Goodbye to SEO as we knew it, not to the people who understand search

The question has stopped sounding absurd. A few years ago, “Is it time to say goodbye to SEO teams?” would have been treated as a provocation from someone who did not understand acquisition. In May 2026, it is a serious management question. Google has put AI deeper into Search, OpenAI has turned ChatGPT into a search destination, Microsoft has kept pushing generative search inside Bing, Perplexity has taught users to expect sourced answers rather than result pages, and publishers are fighting over whether AI systems are taking more value from the web than they return. Search has not disappeared. The old bargain behind SEO has changed.

The question is real, but the answer is not a firing plan

That bargain was simple enough to build departments around. Users searched. Google showed ranked links. Brands invested in technical access, useful pages, authority signals, and content. In exchange, they received qualified traffic. Traffic could be measured, attributed, forecast, and defended in budget meetings. SEO teams became the stewards of that machine. They fixed crawling issues, improved templates, built internal links, mapped keywords to landing pages, prepared briefs, advised developers, audited competitors, and reported on rankings, clicks, conversions, and revenue.

The machine is still running, but it no longer behaves like a straight line. Google’s AI Overviews and AI Mode may summarize answers, link to supporting sources, and use query fan-out to search across related subtopics before returning a response. Google’s official guidance says its generative AI features are rooted in core Search ranking and quality systems, and that work described as AEO or GEO is still SEO from Google’s perspective. That is not a small detail. It means Google is not telling companies to abandon SEO. It is telling them that search visibility now includes generative surfaces that draw from the same broad web index and quality logic.

The hard part is that the traffic side of the bargain looks weaker. Pew Research Center found that users who encountered a Google AI summary in March 2025 clicked a traditional search result in 8% of visits, compared with 15% when no AI summary appeared. Ahrefs reported that AI Overviews reduced clicks to top-ranking pages in its sample, first estimating a 34.5% drop and later updating its work with a larger decline in its own data. SparkToro and Datos had already shown that most Google searches in the United States and European Union ended without a click to the open web in 2024.

So the answer is not “fire the SEO team.” The answer is sharper. It is time to say goodbye to SEO teams that exist mainly to manufacture pages, chase rank reports, and treat Google as a predictable traffic pipe. It is not time to say goodbye to people who understand search behavior, crawlability, information architecture, authority, content quality, conversion paths, analytics, and the messy link between how people ask questions and how businesses earn demand.

The next SEO team will look less like a production desk and more like a search intelligence unit. It will need technical people who understand crawl control for both classic bots and AI crawlers. It will need editors who can create evidence-rich material that deserves to be cited. It will need analysts who can read Search Console, server logs, referral patterns, brand demand, AI citation traces, and conversion data without pretending every impression is worth the same. It will need product, PR, legal, and data partners because AI search makes brand identity, source permission, and content rights part of search strategy.

This is uncomfortable because many companies still use “SEO” as a synonym for “blog output” or “keyword work.” That version is exposed. AI can generate rough briefs, rewrite metadata, cluster queries, summarize competitor pages, draft schema, flag internal linking ideas, and produce copy at a speed no human team can match. If a team’s value was tied to repetitive text production or routine checklist work, the business case is collapsing.

But visibility in search is not routine checklist work. It is a system problem. AI search has made the system more complex, not less. Google is still crawling, rendering, indexing, scoring, classifying, ranking, summarizing, citing, filtering, and personalizing. ChatGPT Search needs web sources. Bing generative search reviews sources and displays labeled references. Perplexity has built its identity around answers with citations and publisher partnerships. These systems all need something to retrieve, trust, interpret, and present.

The best SEO people have always worked near that intersection. They understood that a page is not valuable because it ranks; it ranks because it solves a retrieval and trust problem better than alternatives. What has changed is the output layer. Search engines are no longer only sending users to documents. They are using documents to assemble answers, comparisons, recommendations, shopping paths, and task flows. That shifts the question from “How do we rank?” to “When machines and people assemble an answer, why would our brand, data, product, expert, or page deserve to be part of it?”

That is not the end of SEO. It is the end of SEO as a department that can be isolated from strategy.

Search did not die; it changed the shape of demand

The mistake in many executive conversations is the assumption that a decline in blue-link behavior means a decline in search behavior. The evidence points in the other direction. People are still asking questions. They are asking longer ones, more conversational ones, more comparative ones, and more task-oriented ones. The old search box trained people to break intent into keywords. AI search trains them to bring the whole problem.

That matters because demand has not disappeared. It has become harder to observe. A user who once searched “best CRM for small consulting firm,” opened five tabs, compared pricing pages, and clicked a review result may now ask an AI search system for a shortlist, request a comparison, ask for trade-offs, and click only after the answer has narrowed the field. The brand that appears in the answer may gain influence before a visit. The brand that receives the click may get a more qualified visitor. The brand that appears nowhere may never know it was excluded.

Google says AI Mode uses query fan-out, breaking a question into subtopics and issuing multiple related searches at once. The official Search Central documentation says AI Overviews and AI Mode may use that technique across subtopics and data sources to develop a response. That means the visible user query can be only the starting point. The system may run hidden retrieval steps around entities, comparisons, attributes, definitions, reviews, requirements, locations, and follow-up intent.

This changes the shape of SEO work. A keyword list is no longer enough because the system may expand the query before a human sees the answer. A page built for one exact phrase may be too thin if the model needs corroboration across related concepts. A product page with weak specifications may be invisible in comparison flows. A service page with vague claims may be excluded when the AI system looks for proof, constraints, pricing logic, case evidence, or industry-specific language.

Search demand is also moving across interfaces. Google remains central, but it is not alone in the user’s mind. OpenAI introduced ChatGPT search in October 2024 with source links and a references sidebar. Microsoft’s Copilot Search in Bing promises summaries and layouts that reduce scrolling. Perplexity markets itself as an answer engine with real-time trusted answers. These products are not identical, and none should be treated as a Google replacement at web scale. Yet they change user expectations. A user who gets a sourced synthesis in one place expects less friction everywhere else.

For SEO teams, the practical shift is from page acquisition to answer participation. That phrase can sound abstract, so it needs grounding. Answer participation means the brand’s public information is clear enough to retrieve, credible enough to cite, and distinct enough to be selected. It means the company has pages that answer factual questions, product pages that expose real attributes, documentation that resolves edge cases, expert content that carries named accountability, and external references that confirm the brand is not only praising itself.

Classic SEO did some of this already. Strong teams have always cared about technical access, entity clarity, internal linking, structured data, digital PR, user satisfaction, and topical depth. The difference is that AI search punishes shallow separation between those disciplines. A brand cannot fix answer visibility with metadata while its documentation is thin, reviews are weak, product data is messy, and executives appear nowhere in the topic conversation. AI search compresses brand, content, technical, and reputation signals into a single retrieval problem.

This is why the “goodbye” framing can be dangerous. If a company interprets AI search as proof that search no longer matters, it may cut the only people who understand how public information becomes discoverable. The better move is to ask whether the current team is organized around the right work. A team built to publish 50 generic articles a month deserves scrutiny. A team that can diagnose crawl waste, defend source authority, guide product content, shape editorial evidence, advise on AI crawler access, and explain changing demand patterns deserves more influence, not less.

The shape of demand has also become less linear. A user can search, ask an AI follow-up, watch a YouTube result, read Reddit, compare a product in an AI answer, visit a brand directly, and convert days later. Google’s AI features may include links, but the click is no longer the only evidence of influence. That breaks many reports. It also breaks many teams’ self-image. SEO can no longer defend itself only by pointing to last-click organic sessions. It has to explain its role in discovery, evaluation, trust formation, and brand selection.

The strongest teams will stop treating AI search as a separate channel with a new acronym. They will treat it as a new retrieval layer over the same public web, shaped by different interfaces and different incentives. That is a more demanding view. It asks SEO leaders to understand both old mechanics and new surfaces. It also keeps the discipline honest. Search did not die. The measurable click became a smaller part of a larger influence system.

The traffic bargain behind SEO is under pressure

SEO became a budget category because it promised compounding returns. Paid media rented attention. SEO built assets. A strong page could rank for years, pull demand at low marginal cost, and reduce dependence on advertising. That story was never perfect, but it was persuasive because it matched what many companies saw in analytics. Build content, improve technical quality, earn authority, gain traffic, convert some of it, repeat.

AI search has not destroyed that story, but it has made the assumptions weaker. The first assumption was that ranking high would usually earn a meaningful share of clicks. The second was that informational queries were useful because they filled the top of the funnel. The third was that search engines needed publishers, brands, and experts enough to keep sending them traffic. Each assumption now needs fresh proof.

The click evidence is uneven by sector, query type, geography, and device, but the direction is hard to ignore. Pew’s study showed lower click behavior when AI summaries appeared. SparkToro and Datos showed that zero-click behavior was already high before AI Overviews became a mainstream feature. Ahrefs and other industry analyses have reported click declines on queries affected by AI summaries. Semrush found that AI Overviews were heavily informational earlier in 2025 but expanded into commercial, transactional, and navigational query types later in the year, a shift that matters because those intents sit closer to revenue.

The pressure is not only from Google. ChatGPT Search, Bing generative search, Perplexity, and other answer systems are teaching users that a synthesized answer can be the product. When the answer is the product, the source page becomes infrastructure. Some pages will still receive clicks, especially when the user needs depth, proof, purchase, service, community, tools, or personal judgment. Many pages will not. The issue for SEO teams is not whether clicks vanish entirely. They will not. The issue is that the value of a ranking position becomes less stable when the interface can answer before the visit.

That instability matters for forecasting. A page can hold a high organic position while traffic declines because the search result now contains a stronger answer unit. A query can show more impressions while producing fewer visits because AI surfaces and expanded features change what “visibility” means. A brand can appear inside an AI answer without getting enough referral traffic to justify old attribution models. A competitor can be cited in an AI answer while ranking below you in classic results. Those are not edge cases. They are the new reporting headaches.

The traffic bargain is also under pressure because source economics are being renegotiated. Cloudflare introduced Pay Per Crawl in 2025, giving domain owners options to allow, charge, or block AI crawlers. Its later analysis described a widening crawl-to-click gap, with AI crawling rising while referrals to publishers fell. Reuters reported that Cloudflare’s tool was backed by major publishers and platforms and framed it as a response to AI systems using web content without sending enough value back.

This is not only a publisher problem. Any company that invests heavily in original research, documentation, reviews, tools, market data, or expert analysis now has to decide how it wants machines to access that material. Blocking all AI crawlers may protect some content but reduce answer visibility. Allowing all AI crawlers may increase inclusion but weaken the value exchange. Charging may make sense for some media businesses but not for brands that want discovery. These are business decisions, not only technical settings.

The old SEO team rarely owned that conversation. Robots.txt and crawl controls were often treated as implementation details. In the AI search era, crawler policy can affect licensing, visibility, attribution, legal posture, and commercial strategy. That is a sign of SEO becoming more strategic, not less. But it also means a team that cannot speak with legal, product, communications, and finance will struggle.

The pressure on the traffic bargain should make companies more selective. Not every page deserves creation. Not every informational query deserves investment. Not every glossary term has business value. Not every AI citation will matter. SEO teams must get better at distinguishing content that creates market authority from content that merely fills a calendar.

A useful test is whether a page has a reason to exist beyond matching a query. Does it contain original evidence? Does it clarify a decision? Does it present product or service information better than competitors? Does it earn trust through named expertise or real-world detail? Does it help a user do something the AI answer cannot fully complete? Does it make the brand a better source for future retrieval? If the answer is no, AI has made the page less defensible.

Traffic is still valuable. Organic visits still convert. Search still reveals demand. Google remains a powerful discovery system. But the old bargain has become conditional. SEO investment now needs to prove not only that it can win rankings, but that the visibility it creates survives answer interfaces, supports brand choice, and leads to measurable business outcomes beyond raw sessions.

That is a harder argument. It is also a better one.

Google’s own answer is still SEO

Google’s May 2026 guidance for generative AI features in Search is blunt on one point: from Google’s perspective, work described as answer engine optimization or generative engine optimization is still SEO. The company says the best practices for SEO remain relevant because AI features in Search are rooted in core ranking and quality systems. That does not settle every argument about traffic, fairness, or publisher economics. It does settle one practical question for website owners: Google is not presenting AI search as a reason to abandon the fundamentals of being crawlable, indexable, understandable, useful, and trustworthy.

This matters because the SEO industry is prone to renaming itself whenever the interface changes. Mobile SEO, voice SEO, app SEO, local SEO, YouTube SEO, Amazon SEO, AEO, GEO, LLMO, AI SEO. Some names are useful because they clarify the surface. Others become sales packaging. Google’s position is that AI Overviews and AI Mode are not a separate universe detached from Search. They are part of Search. For brands, that means foundational work still matters: technical access, page quality, structured information, user experience, helpful content, and policy compliance.

Google’s AI features documentation says AI Overviews and AI Mode can display links and may use query fan-out to find a wider and more diverse set of helpful pages than classic search. Its AI features guidance also explains that normal Search controls such as preview controls can affect how content appears in AI experiences. Google’s robots meta tag documentation covers page-level indexing and serving controls, and its crawler documentation explains Google-Extended as a product token for managing whether content may be used for Gemini model training and grounding in certain Google AI contexts.

For a search team, these details are not trivia. They prove that AI search is still connected to web infrastructure. A page blocked from crawling has limited visibility. A page without clear structure may be harder to interpret. A site with poor internal links may hide important material. A content program that floods the index with low-value pages may create quality problems. A brand that cannot distinguish Googlebot, Google-Extended, OAI-SearchBot, GPTBot, and other crawlers may make crude access decisions.

Google’s Search Essentials still define eligibility and spam boundaries. The spam policies warn against behaviors that can lead to lower ranking or omission from Search, and Google has expanded its language around spam to include manipulation of generative AI responses as well as classic rankings. Google’s generative AI content guidance says AI can be used in content workflows, but scaled production without added value can violate spam policy.

This is where many “SEO is dead” arguments fail. They confuse a change in output with a disappearance of input quality. AI systems need retrieval. Retrieval needs sources. Sources need technical access and trust signals. Trust signals need real substance. That does not mean every old SEO tactic works. It means the deeper reasons behind SEO still apply.

Old SEO team model versus search visibility model

This table does not argue for replacing SEO with a fashionable acronym. It shows that the same discipline has moved from page-level tactics toward system-level visibility. Teams that cannot move with it will look expensive; teams that can will become harder to replace.

Google’s answer also creates an uncomfortable standard for SEO vendors. If AI visibility is “still SEO,” then selling it as a magical separate technique is suspect. There are new measurement problems and new tactics, but the core work remains grounded in the same public facts: crawlability, relevance, quality, authority, usefulness, and user satisfaction. Any agency promising instant AI answer inclusion without improving the underlying source deserves skepticism.

The same applies to in-house leaders. A chief marketing officer should not ask only whether the SEO team knows how to “rank in AI.” The better question is whether the team can explain which parts of Google’s AI features are controllable, which are not, what data is visible, what data is blended, which content types are most at risk, which pages are still gaining qualified demand, and which parts of the site are poor sources for machine interpretation.

Google’s position does not remove the threat to clicks. The company argues that AI features create new opportunities for exploration and links. Publishers and independent studies have raised concerns that summaries reduce visits. Both can be true in different contexts. Some queries may produce more searches, more follow-ups, and higher-quality clicks. Others may satisfy the user without a visit. SEO teams need to stop treating platform claims as either gospel or conspiracy. They need to test outcomes by query class, page type, market, and business model.

The strategic point is clear. If the board asks whether it is time to say goodbye to SEO teams, Google’s own documentation supports a better answer: say goodbye to narrow SEO, not to search expertise. The work still exists. The standards are higher. The interfaces are less generous. The reporting is messier. The role is bigger than it used to be.

AI Mode turns one query into many hidden queries

Classic SEO trained teams to map one query to one page. That was never the whole truth, but it was a useful simplification. A user searched a phrase. Google matched pages. The SEO team studied the result page, inferred intent, improved a page, and tried to rank. AI Mode weakens that model because the visible phrase may trigger a chain of invisible retrieval work.

Google’s query fan-out explanation is central here. AI Mode can break a question into subtopics and issue multiple searches simultaneously. Google’s AI features documentation says both AI Overviews and AI Mode may use this technique across subtopics and data sources while models identify supporting pages.

A user might ask, “Which accounting software is best for a small architecture firm with project billing?” A classic SEO program might target “best accounting software for architects” and create a comparison article. An AI search system may fan that out into project billing, time tracking, job costing, client retainers, invoicing, integrations with project management tools, regional tax needs, firm size, software reviews, pricing tiers, implementation difficulty, and alternative tools. The answer may cite sources that do not rank for the original phrase but do provide strong information on one subtopic.

This changes the meaning of topical authority. It is not enough to own a keyword cluster with similar articles. The brand must cover the decision structure around the topic. That includes definitions, comparisons, constraints, use cases, edge cases, data, product details, expert commentary, and proof from outside the brand’s own site. AI search rewards sources that can support a multi-part answer. Thin pages built for exact-match capture will struggle when the system is assembling a response from a wider intent map.

For SEO teams, query fan-out demands better research. Keyword tools still help because they show how people phrase demand. But teams need to model questions the way a user would develop them in a conversation. After the first answer, what would the user ask next? What evidence would they need to trust a recommendation? Which attributes decide the choice? Which objections stop the purchase? Which regulatory, geographic, financial, technical, or operational details change the answer?

This is where experienced SEO practitioners become more useful, not less. AI can generate lists of related questions. It can cluster topics and draft outlines. But it cannot decide which questions matter to the business without judgment. It cannot know which objections sales teams hear every week unless the organization feeds that knowledge into the content process. It cannot create original benchmarks or customer evidence out of nothing. It cannot replace a subject expert who understands why one answer is legally unsafe, commercially weak, or technically incomplete.

Query fan-out also changes internal linking. Old internal linking often focused on pushing equity toward target pages. That still matters. But AI-era internal linking should also help machines understand relationships between entities, subtopics, product attributes, and decision paths. A documentation hub should not be a pile of disconnected articles. A service site should not bury industry-specific evidence. A marketplace should not hide filters, specifications, and comparison logic behind JavaScript that crawlers struggle to render or pages that cannot be reached through stable links.

Structured data plays a supporting role here. Google’s documentation says structured data helps it understand page content and gather information about entities such as people, books, companies, recipes, and other items. Structured data does not guarantee inclusion in AI answers. It does make information more explicit. For large sites with products, reviews, events, organizations, articles, recipes, jobs, videos, and FAQs where supported, clean markup is part of source hygiene.

The hidden-query model also creates a new competitor set. Your competitor in AI search may not be the company selling the same thing. It may be a government page defining a requirement, a Reddit thread discussing real usage, a YouTube video demonstrating a workflow, an analyst report explaining the category, a documentation page from an integration partner, or a review site with stronger comparative data. Pew’s analysis of AI summaries found that Wikipedia, YouTube, and Reddit were among the most cited domains in Google AI summaries, according to Search Engine Land’s summary of the Pew findings.

That finding should make brands uncomfortable. If AI systems use third-party communities and reference sites to answer questions about your category, then brand-owned SEO is not enough. You need reputation work, community understanding, product truth, documentation quality, and external validation. A page can be perfectly written and still lose influence if the wider web contradicts it or ignores it.

The hidden-query model also weakens simplistic content gap analysis. Tools can show pages competitors have that you do not. They cannot always show the subtopics an AI system uses to assemble an answer. Teams need to combine search data with customer research, sales calls, support tickets, product analytics, forums, review mining, and expert interviews. This is slower than generating keyword briefs. It is also harder to commoditize.

AI Mode turns SEO from matching known queries into preparing for unknown retrieval paths. That is the clearest reason SEO teams should not vanish. Someone has to map those paths, improve the source material, and decide where the brand deserves to be present. The team just cannot be organized around the assumption that one keyword equals one page equals one ranking equals one click.

Rankings are becoming inputs, not final outputs

Ranking still matters. That sentence needs to be plain because some AI-search commentary has become reckless. Google’s AI features draw from Search systems. Pages that perform well in classic organic results often have a better chance of being discovered, interpreted, and cited. A page that cannot be crawled, indexed, or trusted is unlikely to become a reliable AI source. Rank tracking is not useless.

But rankings are no longer the final output. They are one input into a more complex answer surface. A classic result page asked the user to choose from ranked documents. AI search asks the system to choose, summarize, synthesize, and cite before the user clicks. The user may still open a link, but the decision environment has changed.

This distinction matters for incentives. If a team is judged only on rank movement, it may keep doing work that does not protect business value. A page can rank first and lose clicks when an AI Overview answers the query. A page can rank lower but be cited because it contains a precise fact the system needs. A brand can gain demand because it appears in synthesized recommendations even if Search Console does not cleanly separate that exposure. A brand can lose demand because it is absent from AI answers that shape consideration before anyone reaches its site.

Google’s Search Console documentation now addresses how AI features are counted within performance data, but visibility into AI surfaces remains less clean than old rank reports. Google says AI Overviews and AI Mode are counted in Search Console data, and industry reporting has described AI Mode data being included in Search Console totals. But the measurement picture is still limited because AI answer exposure, citation, and downstream influence are not as transparent as a traditional click.

This creates a management problem. Executives like simple dashboards. SEO teams have often complied by presenting average rank, traffic, and conversions as if the channel were stable. AI search punishes that simplicity. The same query can behave differently depending on whether AI Overviews appear, whether the user expands links, whether the answer contains citations, whether a source is visible above the fold, whether the user switches to AI Mode, and whether the query becomes a conversation.

The next SEO dashboard needs to separate at least five layers: classic ranking, search feature presence, AI summary presence, brand/source inclusion, and business outcome. Not every company will have perfect data for each layer. That is not an excuse to keep pretending old metrics explain the whole system. Teams should build directional evidence from Search Console, analytics, server logs, third-party monitoring, referral data from AI platforms, brand search trends, conversion quality, sales feedback, and manual review of high-value queries.

This is also where rank tracking vendors and AI visibility tools will face scrutiny. Some will become useful. Others will sell false certainty. AI answer monitoring is difficult because outputs can vary by location, time, device, user state, query wording, follow-up context, and platform. A report that says “you rank third in ChatGPT” should be treated carefully. There may be useful sample-based visibility data, but it is not the same as a stable Google position.

The best SEO teams will explain uncertainty clearly. They will say which data is observed, which is sampled, which is inferred, and which is unavailable. That honesty is part of the new job. AI search creates enough confusion already; teams should not add fake precision.

Rankings becoming inputs also changes content evaluation. Under the old model, a page that ranked and attracted traffic was considered successful. Under the new model, the page may need to do more. It may need to provide extractable facts, cite-worthy definitions, comparison tables, original data, named expertise, useful images or diagrams, product specifications, and pathways to deeper action. The goal is not to write for machines at the expense of people. It is to make human-useful information explicit enough for machines to understand and credible enough for them to use.

There is a subtle but important shift in conversion thinking too. If AI search pre-qualifies users, organic traffic may shrink while conversion rate rises for some pages. A smaller number of visitors can still be valuable if they arrive with clearer intent. SEO teams must avoid panic when raw sessions decline and instead ask whether revenue, leads, assisted conversions, branded demand, and customer quality changed. Some declines will be real losses. Some will be the removal of low-value visits. The team’s job is to know the difference.

This is why firing SEO teams because rank reports feel less meaningful would be a weak move. The reports are less complete, not the work. The discipline must move from rank ownership to visibility interpretation. Rankings still matter, but they no longer tell the whole story. A company that loses that interpretive capacity will be blind at the moment search becomes harder to read.

The new visibility gap is between being read and being chosen

AI systems can read more than they choose. That gap is the core anxiety for brands. Being crawlable is not the same as being indexed. Being indexed is not the same as ranking. Ranking is not the same as being cited. Being cited is not the same as being trusted by a user. Being trusted is not the same as earning a click, lead, sale, or subscription. Each step filters the brand.

Old SEO focused heavily on the crawl-index-rank path. AI search adds another layer: selection inside generated answers. A page can be available to the system yet not chosen as a supporting source. A brand can be known to the model yet not recommended. A fact can be extracted without a prominent link. A source can be cited but receive minimal traffic. For teams measured by clicks, this is frustrating. For users, it may feel convenient. For search platforms, it keeps more activity inside the interface.

The visibility gap puts pressure on the quality of source material. Vague marketing pages are poor machine sources because they do not resolve uncertainty. “We help businesses grow” is not useful. “Our platform supports project-level invoicing, retainer billing, and revenue recognition for architecture firms with 10 to 200 employees” is more useful. The second sentence contains entities, attributes, use cases, and constraints. It helps both people and systems understand fit.

This is not keyword stuffing. It is specificity. AI search makes specificity more valuable because generated answers need facts, comparisons, and distinctions. The old content habit of writing broad, polished paragraphs around a keyword becomes less useful when the system can synthesize broad information from many sources. The content that survives is often the content that adds something a generic answer cannot: original data, operational detail, expert judgment, product truth, case evidence, legal nuance, local knowledge, or a clear point of view.

The visibility gap also rewards consistency. If a company describes its product one way on the homepage, another way in documentation, another way in partner listings, and another way in review profiles, systems receive mixed signals. Entity clarity becomes part of search work. The brand name, product names, categories, founders, experts, locations, pricing, integrations, certifications, supported industries, and policies should be consistent across the web where accuracy matters. This is not only a knowledge graph issue. It is a trust issue.

Being chosen also depends on external corroboration. A brand’s own claim is weaker than a claim supported by reviews, references, industry coverage, documentation, standards, customer evidence, and independent mentions. That does not mean companies should chase low-quality mentions or fake digital PR. It means the SEO team’s remit overlaps with communications and reputation. If no credible external source confirms the brand’s expertise, an AI answer has less reason to choose it.

This is particularly important for YMYL topics such as health, finance, legal, safety, and major life decisions. Google’s systems have long emphasized quality and trust for sensitive topics. AI summaries raise the stakes because users may act on synthesized advice. A brand in a sensitive category needs named experts, evidence, review processes, citations, clear dates, disclaimers where appropriate, and content that distinguishes educational information from professional advice. Generic AI-written content is especially dangerous here because it can sound fluent while adding no verifiable authority.

The same principle applies outside YMYL. A B2B software company should not publish generic “best practices” content with no product evidence, customer examples, or implementation detail. A travel brand should not summarize destinations without current, first-hand, locally accurate information. An ecommerce site should not rely on manufacturer copy if it wants to be chosen in comparison answers. A local service business should not hide credentials, service areas, before-and-after proof, pricing logic, reviews, and availability behind thin pages.

The visibility gap also changes how teams think about snippets and summaries. Some publishers want more control over how much content appears in AI features. Google’s robots meta and preview controls allow certain limits on snippets and previews. OpenAI documents separate crawlers for search, training, and user-triggered browsing. These controls create choices. More restriction may protect content but reduce the ability of systems to understand and cite it. More openness may improve discoverability but reduce bargaining power.

SEO teams should not make those choices alone. They should bring evidence to the table: which content drives business value, which content is expensive to produce, which content is already copied or summarized, which platforms send referral traffic, which crawlers consume resources, and which pages need visibility more than protection. That is a governance role.

The future visibility gap is not between companies that “do SEO” and companies that do not. It is between companies whose public information is reliable enough to be chosen and companies whose content is merely available. That is a higher bar. It is also a clearer one.

Clicks are becoming scarcer and less evenly distributed

The click decline story is easy to exaggerate and easy to dismiss. Both instincts are wrong. Organic search is not collapsing uniformly. Many sites still get meaningful traffic from Google. Transactional searches, local searches, brand searches, image and video searches, product searches, and navigational searches can still drive visits. Yet AI summaries and zero-click behavior are real enough that no serious SEO strategy can ignore them.

Pew’s 2025 study is one of the clearest public signals because it looked at actual user behavior. The finding that users clicked traditional results nearly half as often when an AI summary appeared is not a minor interface detail. It suggests that summaries can satisfy enough intent to change behavior. Pew also found that users rarely clicked links inside the AI summaries themselves.

Ahrefs’ work adds another angle. Its initial study estimated a 34.5% reduction in clicks to top-ranking content when AI Overviews were present. Its later update, based on its own Search Console data, reported a larger decline. Every study has limits, and no single dataset should be treated as universal law. But the direction aligns with what many publishers and SEOs are seeing: the top organic position is less reliable as a traffic guarantee when an AI unit sits above or within the user journey.

SparkToro and Datos showed that the zero-click pattern was already mature before Google’s AI Overviews became the center of the debate. In 2024, they estimated that 58.5% of U.S. Google searches and 59.7% of EU Google searches ended without a click. That baseline matters because AI summaries are not the first search feature to answer queries on the results page. Featured snippets, knowledge panels, maps, calculators, weather boxes, flights, shopping modules, lyrics, sports results, and direct answers have been shifting value to the interface for years.

The new risk is distribution. Clicks may not only become scarcer; they may concentrate around certain query types, brands, and sources. When an AI answer cites three or four sources, the rest of the result set may become less visible. When a platform shows product recommendations, trusted marketplaces, large review sites, Reddit threads, YouTube videos, or official documentation, smaller publishers and brands may struggle to appear. When answers satisfy broad informational needs, only sources with deeper value may get the follow-up click.

This forces a shift in content economics. A company cannot assume that every informational page will earn enough traffic to justify itself. Top-of-funnel SEO used to tolerate low conversion rates because traffic volume was large and cheap. If AI answers reduce that volume, top-of-funnel content must work harder. It has to create brand memory, support sales, earn citations, attract links, feed email capture, support retargeting, or become a source for higher-intent pages. Traffic alone may not carry the business case.

The same pressure applies to affiliate and review models. If AI search summarizes “best” lists, compares products, and sends fewer users to review pages, sites built on ranking for commercial comparison queries face a direct threat. They will need stronger original testing, first-party data, community trust, tools, video, newsletters, subscriptions, or direct brand relationships. Thin affiliate pages are vulnerable because AI can reproduce their basic function without needing to send many clicks.

For B2B and professional services, click scarcity may produce a different pattern. Informational traffic may decline, but high-intent visits may become more valuable. A user who clicks after reading an AI comparison may be closer to action. That can raise conversion rates while lowering sessions. SEO teams need to segment this carefully. A 30% traffic decline is bad if leads and revenue fall with it. It is less alarming if low-value informational visits disappear while pipeline remains stable. It is promising if fewer visits produce better-qualified opportunities.

Click scarcity also changes the role of brand demand. Users who trust a brand may search for it directly, click it inside AI answers, or navigate to it after seeing it mentioned. Brand search, direct traffic, newsletter subscriptions, community membership, and repeat visits become more important. SEO teams cannot own all of that, but they should influence it. Search visibility and brand building are no longer separate in practice. A brand with stronger recognition may win clicks even when result pages are crowded with summaries.

The worst response to click scarcity is to publish more low-cost content. That floods the site with pages least likely to survive the shift. The better response is to classify content by function. Some pages should attract demand. Some should support AI citation and answer inclusion. Some should help sales teams. Some should convert existing demand. Some should build authority through original research. Some should be removed because they weaken the site.

Scarcer clicks do not make SEO irrelevant. They make weak SEO more visible as waste. The next SEO team must be comfortable killing pages, not only creating them. It must protect the parts of organic search that still produce revenue while building new ways to measure influence where clicks are no longer the whole story.

News publishers are the warning system for everyone else

News publishers feel AI search pressure early because their economics depend heavily on attention, referrals, and control over original reporting. When search engines summarize news or background information, publishers can lose the visit that supports advertising, subscriptions, registration, and brand loyalty. That is why the publisher debate around AI Overviews, AI crawlers, and source compensation matters far beyond media.

Reuters reported in July 2025 that Google faced an EU antitrust complaint over AI Overviews from independent publishers, who alleged that the feature caused traffic, readership, and revenue harm. In April 2026, Reuters reported that Italy’s communications watchdog AGCOM asked the European Commission to investigate Google’s AI-powered search features over concerns about publisher harm and media pluralism. These are not abstract industry complaints. They show that AI search has become a regulatory and economic dispute about who captures the value of web content.

Cloudflare’s Pay Per Crawl initiative is part of the same warning. It gives publishers and site owners a way to charge, allow, or block AI crawlers. Reuters framed the launch as a response to traditional web traffic declining while AI systems extract content without sending enough users back. Cloudflare’s own analysis described a crawl-to-click gap in which AI crawling activity grew while referrals were weaker.

Perplexity’s publisher program shows another path: licensing, revenue share, analytics, and partnerships. Reuters reported that Perplexity expanded the program with publishers including the Los Angeles Times and The Independent, while also facing legal pressure from other media groups. Le Monde later announced a partnership with Perplexity that allowed its editorial content to be used in answers with links back, while not permitting model training.

For non-media companies, the lesson is not “become a publisher lobbyist.” The lesson is that public information now has two kinds of value. It can attract humans. It can also feed machines. SEO teams have historically focused on the first value. AI search forces companies to manage both.

A software company’s documentation may train users, support customers, and help AI systems answer category questions. A medical device manufacturer’s educational material may support clinicians and be cited by answer engines. A travel company’s local guides may inform AI trip planning. A retailer’s product data may feed agentic shopping recommendations. A university’s research pages may become sources in AI summaries. In each case, the company has to decide what it wants from machine access.

News publishers also warn against depending too much on one platform. Many publishers built audience strategies around search and social referrals, then watched platforms change distribution. Brands can make the same mistake with Google organic traffic. AI search does not mean companies should abandon Google. It means they should stop treating any single interface as a guaranteed path to customers.

The publisher experience also exposes a measurement problem. If an AI system summarizes a publisher’s work and the user does not click, the publisher may still have influenced the answer but received no measurable benefit. A brand can face a milder version. An AI answer may mention the brand, shape perception, and drive a later direct visit. Traditional analytics may miss the causal path. This does not mean teams should invent credit. It means they should treat attribution as incomplete.

There is also a quality lesson. News publishers with original reporting have something AI systems need. Commodity rewrites have less bargaining power. The same applies to brands. Original research, proprietary data, expert interviews, field experience, customer evidence, and product documentation are stronger assets than generic explainers. If AI can synthesize the same article from five public sources, the page is weak. If the page contains facts or judgment that cannot be found elsewhere, it has more strategic value.

The publisher fight may also shape future law, platform policy, and crawler behavior. If regulators force more granular opt-outs, licensing, data access, or transparency, SEO teams will need to adapt quickly. The EU’s Digital Markets Act scrutiny of Google and the U.S. antitrust remedies around search show that search is no longer only a marketing channel; it is a policy arena.

Companies outside media often ignore these fights until they affect traffic. That is a mistake. Publishers are the warning system because their business model makes the damage visible sooner. The same mechanics can reach ecommerce, travel, education, software, local services, finance, healthcare, and B2B research. Once AI systems answer more commercial and transactional questions, the effects move closer to revenue.

The publisher lesson is blunt: if your business invests in knowledge, you need a strategy for how machines use that knowledge. SEO teams are natural participants in that strategy, but only if they understand content rights, crawler control, brand value, and platform incentives. Teams that only know keyword volume will not be enough.

SEO teams built around production volume are exposed

The easiest SEO team to cut is the one that looks like a content factory. If the team’s main output is a steady flow of generic articles built from keyword lists, AI has already weakened its case. Generative tools can produce drafts, outlines, metadata, FAQs, and variations quickly. They can summarize competitor pages, generate schema templates, and fill a calendar. If human SEO work is reduced to feeding that machine, the team becomes a cost center waiting to be automated.

This does not mean content no longer matters. It means volume without evidence matters less. Google’s guidance on generative AI content does not ban AI use, but it warns against using automation to generate many pages without adding value. Google’s Search Essentials and spam policies focus on helpfulness and avoiding manipulative behavior, including scaled content abuse.

The production-volume model had flaws before AI. Many sites published hundreds of pages with overlapping intent, weak expertise, thin differentiation, and no maintenance plan. Some pages ranked because competition was weak or search engines had not yet tightened quality systems. AI search raises the bar because generic informational needs can be satisfied inside the interface. If a page only restates public knowledge, why should a user click it? Why should an AI system cite it? Why should Google rank it over a stronger source?

The new content test is evidence density. A useful page should contain enough original or well-organized evidence to justify attention. That may include data, methodology, expert review, screenshots, product specifications, pricing details, use-case boundaries, implementation steps, examples from practice, limitations, and current dates. It may include external citations where appropriate. It should also make clear who is responsible for the information and why they are qualified.

SEO teams built around volume often lack access to the people who can provide that evidence. They sit in marketing, far from product managers, engineers, doctors, lawyers, consultants, analysts, sales teams, support teams, and customers. They receive a keyword and produce an article. In the AI era, that workflow is too weak. The team needs source access. It needs interviews. It needs data. It needs permission to say less, but say it better.

This will reduce page counts. That is healthy. Many companies have too many pages competing with themselves. Search teams should audit for decay, duplication, cannibalization, thinness, and outdated information. They should consolidate where one stronger resource can replace ten weak pages. They should delete or noindex pages that drag quality signals down. They should update pages with real changes, not cosmetic rewrites.

The production-volume model is also exposed by maintenance costs. AI-generated content is cheap to create but not free to govern. Someone must verify claims, check legal risk, update dates, maintain internal links, monitor performance, handle broken references, align with product changes, and avoid hallucinated or copied material. A company that publishes too much creates a liability. Content debt becomes search debt.

A smarter team will use AI to reduce low-value labor, not to increase low-value output. AI is useful for research assistance, query clustering, transcript analysis, outline drafts, internal-link suggestions, schema drafts, log pattern summaries, and first-pass content audits. Humans should decide what deserves publication, where evidence comes from, what the brand can credibly claim, and how the page supports business strategy.

This requires a different staffing mix. Fewer pure content operators. More subject editors. More technical SEOs. More analysts. More people who can work with product and legal. More strategists who can connect search demand with commercial priorities. The total headcount may shrink in some organizations, but the average skill level should rise.

There is a painful talent issue here. Entry-level SEO roles often involved routine tasks: metadata, keyword mapping, simple audits, reporting, content briefs. AI now handles many of those tasks well enough to reduce junior hiring. That can create a pipeline problem because people still need to learn the craft. Strong teams will redesign junior work around review, research, QA, experimentation, and cross-functional exposure rather than repetitive production. Weak teams will simply cut juniors and later wonder why they have no senior talent.

The SEO team that survives is not the team that publishes the most. It is the team that improves the business’s public knowledge base in ways search systems and customers can trust. That is a smaller, harder, more valuable job than filling a blog.

Technical SEO has become more strategic, not less

Technical SEO used to be seen as plumbing. Important, but often invisible. Fix crawl errors, submit sitemaps, improve canonical tags, manage redirects, speed up templates, check robots.txt, handle pagination, diagnose JavaScript rendering, and keep Search Console quiet. AI search has made that work more strategic because technical decisions now govern not only classic search visibility but also machine access, answer eligibility, source control, and content economics.

Google’s Search systems still need to crawl, render, index, and understand pages. Google’s Core Web Vitals documentation frames loading performance, interactivity, and visual stability as real-world user experience metrics aligned with what core ranking systems seek to reward. Structured data helps Google understand content and can make pages eligible for rich results. These are not new concepts, but they remain necessary because AI features are still connected to the web index.

AI crawlers add another layer. OpenAI’s crawler documentation distinguishes robots and user agents used for products, including OAI-SearchBot and GPTBot. Google’s crawler documentation includes Google-Extended for managing certain uses of content in Gemini-related products. Cloudflare’s AI Crawl Control and Pay Per Crawl show that access policy is becoming a business setting.

This means technical SEO teams need a crawler access matrix. Which bots are allowed? Which are blocked? Which are allowed for search but blocked for training? Which high-value content is open? Which content is behind paywalls or registration? Which APIs or feeds are available? Which server logs show heavy AI crawler activity? Which crawlers ignore rules? Which bots create cost without value? Which platforms send referral traffic worth tracking?

Those questions cannot be answered by copying a robots.txt template from a blog post. The right policy depends on the business. A media company may want to limit training use while allowing search citation. A B2B software company may want broad visibility in AI answers. A paid research provider may restrict most content while exposing summaries. An ecommerce site may allow product discovery but protect proprietary data. A local business may not care about training but needs visibility. Technical SEO becomes the translator between business intent and machine access.

JavaScript and rendering also matter more when systems need to extract precise information. If product specs, prices, availability, documentation, reviews, or critical content are hidden in difficult scripts, behind interaction, or rendered inconsistently, crawlers may miss or misinterpret them. Modern search engines can render many pages, but that does not mean every implementation is equally reliable. Large sites should still test rendered HTML, crawl paths, canonical signals, structured data, hreflang, pagination, faceted navigation, and content parity.

Site architecture is another strategic layer. AI search makes comprehensive source quality important, but comprehensive does not mean chaotic. A site needs clear hubs, stable URLs, logical taxonomy, internal links, breadcrumbs, and canonical topic ownership. If a company has 20 near-duplicate articles on a topic, systems and users receive a weaker signal. If the company has one strong guide connected to supporting pages, documentation, product pages, and expert profiles, the source is easier to understand.

Technical SEO also intersects with data feeds. Ecommerce, travel, jobs, real estate, events, and marketplaces depend on structured product or listing data. Search and AI systems cannot recommend what they cannot parse. Clean feeds, schema, inventory updates, canonical URLs, image quality, and policy compliance are not “SEO details” in the agentic era. They are commercial infrastructure.

Security and reliability matter too. If AI crawlers increase server load, a site may face cost and performance issues. Cloudflare’s crawl-to-click analysis shows why site owners are paying attention to AI bot behavior. Technical teams need rate limiting, bot verification, cache strategy, CDN rules, and monitoring. Blocking too aggressively can hurt visibility; allowing everything can waste resources.

The technical SEO role also needs to support experimentation. If a company changes preview controls, blocks certain crawlers, restructures content, adds schema, or exposes new data feeds, someone must monitor effects. The results may be noisy because AI surfaces are volatile, but doing nothing is worse. The team needs test design, documentation, and rollback plans.

This is where saying goodbye to SEO teams becomes risky. Many executives underestimate technical SEO until traffic drops after a migration, a JavaScript change, a canonical bug, an accidental noindex, or a robots.txt mistake. AI search adds more failure modes. A company can block the wrong crawler, hide critical content, break structured data, weaken internal links, or starve answer systems of the details that make the brand eligible.

Technical SEO is no longer only about making pages accessible to Google. It is about governing how machines access, interpret, use, and value the company’s public information. That is not a role to eliminate. It is a role to move closer to engineering, product, legal, and leadership.

Content quality now means source quality, not polished text

The phrase “quality content” has been abused into near uselessness. For years it often meant well-written, long enough, formatted with headings, and matched to intent. AI has made that standard inadequate because fluent text is now cheap. A page can be polished and still be worthless. It can sound authoritative and still contain no original value. It can satisfy an old SEO checklist and still fail as a source.

Source quality is a stricter idea. A high-quality source is useful because it contains accurate, current, specific, attributable, and verifiable information that helps a person or system answer a real question. It shows its work where necessary. It names experts where expertise matters. It distinguishes fact from opinion. It includes dates when timing matters. It states limits. It provides enough detail for a reader to make a decision or continue research.

Google’s long-standing guidance around helpful content and AI-generated content points in this direction. The company says it rewards helpful, reliable, people-first content rather than content made primarily to manipulate rankings. It does not prohibit AI assistance, but it warns against automation used to produce low-value pages at scale.

AI search makes source quality visible in new ways. A generated answer needs support. It may prefer pages that directly answer subquestions, provide concise definitions, include clear evidence, or have strong external trust. A bloated article that hides the answer under generic paragraphs is a poor source. So is a product page that uses adjectives instead of specifications. So is a medical article with no author credentials or update history. So is a local service page with no location proof, reviews, or service boundaries.

This does not mean every page should become academic. A restaurant page does not need peer review. It needs accurate hours, menu, location, booking details, photos, reviews, and local relevance. A SaaS documentation page needs exact instructions, version awareness, screenshots, limitations, and troubleshooting. A legal explainer needs jurisdiction, expert review, dates, and warnings about advice. A product comparison needs test criteria, data, and honest trade-offs.

The key is fit between claim and evidence. If a page claims “best,” it should explain the criteria. If it claims “fastest,” it should provide data. If it claims “trusted,” it should show proof. If it recommends a tool, it should disclose basis and limits. AI systems and human readers both benefit from that discipline.

Source quality also changes the role of writers. The best SEO writers are becoming editors, researchers, interviewers, and explainers. They need to extract knowledge from experts, structure it for retrieval, and preserve accuracy. They must be comfortable saying that a topic does not deserve publication until evidence exists. They must resist the pressure to produce generic text because a keyword tool shows volume.

This creates friction with old content operations. Many marketing teams value speed and calendar consistency. AI makes speed easy. Search quality now demands restraint. A company may publish fewer pages, but each page needs stronger inputs. That requires planning cycles around expert availability, data collection, review, design, and updates. It also requires content governance: who owns each page, when it was last reviewed, what claims need verification, and when the page should be retired.

Source quality also includes format. Some answers are better served by tables, diagrams, calculators, tools, videos, or downloadable templates than by prose alone. AI search may summarize text, but users still click when they need a tool, a template, a calculator, a configuration workflow, a visual explanation, or a trusted transaction. Content teams should ask where the page can do something an AI answer cannot fully replace.

The web is likely to fill with AI-written sameness. That creates an opening for brands with real expertise. The opening is not automatic. Originality must be visible. A page based on original survey data should expose the methodology. A field guide should include examples from practice. A product page should show real screenshots or specs. An expert article should identify the expert and their experience. A case study should include constraints, not only praise.

The new content standard is not “write better.” It is “be a better source.” That is a much harder brief. It also gives SEO teams a defensible role if they can enforce it.

Brands need entity management as much as keyword coverage

Keywords show demand. Entities show meaning. AI search depends heavily on meaning because generated answers need to understand people, organizations, products, places, categories, attributes, and relationships. A brand that is poorly understood across the web is at a disadvantage even if it has pages targeting many phrases.

Entity management begins with basic consistency. The brand name should be used consistently. Product names should not drift across pages, press releases, documentation, review listings, app stores, marketplaces, and partner profiles. Founders, executives, authors, doctors, lawyers, engineers, and other experts should have clear biographies and external corroboration where relevant. Locations, service areas, credentials, awards, certifications, and policies should be accurate across major profiles.

This sounds like reputation management, and it is. It is also search strategy. Google uses structured data and other signals to understand entities and the web. AI answer systems need to resolve which organization or product a user means, whether it belongs in a category, and whether claims about it are supported. If the web contains conflicting or thin signals, the brand is harder to place confidently.

Entity management also changes category strategy. Many companies describe themselves using internal language that no buyer uses. Others chase broad category labels without proving fit. AI search can expose that mismatch. If users ask for “best payroll software for restaurants with tip pooling,” a payroll vendor that never connects itself publicly to restaurants, tip pooling, compliance, integrations, and customer examples may be absent from the answer even if it ranks for “payroll software.”

Keyword coverage might say the company needs a page for “restaurant payroll software.” Entity management asks a deeper question: is the company recognized as a relevant solution for restaurant payroll across its own site and the wider web? Does it have documentation, customer stories, reviews, partner listings, integration pages, and support content that reinforce the relationship? Does it explain limitations? Does it have experts or legal reviewers where compliance matters?

This is where SEO, PR, partnerships, product marketing, and customer marketing overlap. External mentions from credible sources can shape how the brand is understood. So can customer reviews, analyst coverage, app marketplace profiles, awards, industry associations, podcasts, webinars, and conference pages. The SEO team does not need to own all of that, but it should map the entity gaps that affect discoverability.

Entity management is especially important for personal expertise. Google’s quality concepts around experience, expertise, authoritativeness, and trust have pushed many sites to improve author pages, review processes, and expert attribution. AI search raises the same issue because users and systems need to know whose knowledge is behind a claim. A faceless article in a sensitive field is weaker than one reviewed by a named specialist with verifiable credentials.

Brands also need to manage negative and contradictory entity signals. If review sites consistently mention a product weakness, the brand should not pretend it does not exist. AI systems may surface those concerns. A better strategy is to address them honestly: explain who the product is for, who it is not for, what has changed, and where documentation supports the claim. Search visibility is not only about amplification. It is about making reality legible.

Schema can support entity clarity, but it cannot substitute for truth. Organization, Person, Product, Article, Review, LocalBusiness, FAQ where appropriate, and other structured data types can help machines parse information. But markup that contradicts visible content or overstates facts is not a strategy. The underlying page and external signals must support it.

The rise of AI answers also means brand distinctiveness matters more. If ten vendors describe themselves with the same phrases, an AI system may treat them as interchangeable. Distinctiveness comes from clear positioning, named use cases, proof, product specifics, recognizable expertise, and external validation. SEO teams should care about messaging because generic messaging creates generic retrieval.

The keyword era rewarded being findable for phrases. The AI search era rewards being understood as a credible entity in a specific problem space. Companies that treat brand architecture as separate from SEO will miss that shift.

Measurement is the part that breaks first

When search changes, reporting usually breaks before strategy catches up. AI search is no different. The old organic dashboard was built around impressions, rankings, clicks, sessions, conversions, and revenue. It was imperfect but familiar. AI Overviews, AI Mode, ChatGPT Search, Bing generative search, Perplexity, and other answer systems introduce visibility that may not produce a direct click, may be blended into existing Search Console data, may vary by prompt, and may influence conversion later.

Google’s AI features documentation explains how AI Overviews and AI Mode are counted toward Search Console data and points site owners toward combining Search Console and analytics to understand changes. Search Engine Land reported that AI Mode clicks, impressions, and positions were included in Search Console totals as Google updated documentation around counting. But inclusion in totals is not the same as a clean, fully segmented understanding of influence.

This is a problem for budget defense. A CFO may ask whether SEO is down because AI took traffic, because content quality declined, because competitors improved, because demand fell, because tracking changed, because the site lost rankings, because ads crowded the page, because a core update hit, or because users shifted to other platforms. The honest answer may involve several of those factors. A weak SEO team will point to one metric and guess. A strong team will build a measurement framework that separates signals.

The first signal is demand. Search impressions, keyword trends, brand search volume, category demand, and market data help determine whether the market is shrinking or the site is losing share. The second signal is visibility. Classic ranks, search features, AI Overview presence, AI Mode samples, and citation monitoring help show whether the brand is present where decisions happen. The third signal is behavior. Click-through rate, sessions, engagement, scroll depth, assisted conversions, lead quality, and sales feedback reveal what traffic does after arrival. The fourth signal is business outcome. Pipeline, revenue, subscriptions, bookings, quote requests, store visits, and retention matter more than visits.

AI search adds a fifth signal: machine referral and citation. OpenAI says ChatGPT search can include links to web sources, and its publisher FAQ says publishers can track referral traffic from ChatGPT using analytics, including a UTM source parameter. Similarweb reported that AI platform visits grew while AI referrals to external sites were flat between January 2025 and January 2026 in its data, which suggests referral volume alone may understate usage growth.

Teams should tag and track AI referrals where possible. They should monitor referral sources such as chatgpt.com, perplexity.ai, copilot-related sources where visible, and other AI platforms. They should also use server logs to understand bot activity. But they should not overclaim. AI platform referrals can be small, inconsistent, and hard to attribute. Citation monitoring tools can sample visibility but may not represent all users. Manual checks are useful but limited. The correct posture is disciplined uncertainty.

Measurement also needs query segmentation. AI Overviews affect informational queries differently from commercial or transactional queries. Semrush found that AI Overview triggers shifted during 2025, with commercial, transactional, and navigational intents gaining share compared with the start of the year. A site that relies on informational top-of-funnel content may see different effects from a site dominated by brand and product queries.

Page segmentation matters too. Educational content, product pages, comparison pages, category pages, local pages, documentation, tools, and thought leadership have different risk profiles. A glossary definition is more likely to be answered directly. A complex implementation guide may still earn clicks. A pricing page may gain from better prequalification. A comparison page may lose traffic if AI answers summarize the comparison, unless the page contains unique data.

Measurement also has to account for brand lift. If AI answers mention a brand repeatedly, some users may search the brand later or go direct. That influence may appear as branded organic, direct, paid search, or sales activity, not as AI referral. Teams should watch branded search trends, direct conversion quality, CRM source notes, sales-call mentions, and surveys asking how prospects first heard about the brand. None of this is perfect. All of it is better than pretending last-click organic tells the full story.

The worst measurement response is to invent a vanity metric called “AI visibility score” and manage the business around it. A useful score may support analysis, but it should not replace business outcomes. The second-worst response is to ignore AI visibility because it is hard to measure. The correct response is a layered dashboard with confidence levels.

AI search does not remove measurement discipline. It makes measurement more demanding. SEO teams that can explain messy data will become more valuable. Teams that only export rank reports will lose credibility.

AI crawlers turn access into a business decision

Robots.txt used to be a technical file most executives never discussed. AI crawlers changed that. Access to content now affects discovery, model training, answer citation, server cost, licensing leverage, and legal posture. A single allow-or-block rule can express a business strategy, even when the business has not consciously chosen one.

OpenAI’s crawler documentation says it uses web crawlers and user agents for products, with OAI-SearchBot and GPTBot robots.txt tags enabling webmasters to manage how sites and content work with AI. GPTBot is associated with training, while OAI-SearchBot is related to search discovery. Google-Extended lets publishers manage certain uses of their content for future Gemini models and grounding in Gemini-related products, separate from classic Googlebot behavior.

This separation matters. A site owner may want to appear in ChatGPT Search but not train future models. A publisher may want to allow Google Search indexing but limit AI training. A brand may decide that broad AI access is worth it because visibility matters more than content protection. Another may decide that original research behind a paywall should not be freely crawled. There is no universal answer.

Cloudflare’s Pay Per Crawl makes the decision more explicit. A site can allow, charge, or block certain AI crawlers. The feature uses payment intent and HTTP 402 responses as part of the access flow. Whether this model becomes common or remains one part of a larger negotiation, it marks a shift: web crawling is no longer treated as a one-way courtesy.

SEO teams need to help companies classify content by access strategy. Public marketing pages usually benefit from broad discoverability. Product documentation may benefit from search and AI citation but needs accuracy monitoring. Original research may deserve summaries publicly and full access behind registration. Paid media content may require licensing. User-generated content may carry moderation and privacy issues. Sensitive data should not be crawlable at all.

This classification should be documented. Which crawlers are allowed and why? Who approves changes? How often is the policy reviewed? Which systems monitor bot behavior? What happens when a new AI crawler appears? Which legal terms apply? Which business units own the content? Without governance, crawler decisions become accidental. A developer blocks too much to reduce server load. A marketer opens everything to gain visibility. Legal demands a blanket block. None of those one-sided decisions is good enough.

Server logs become more important in this environment. Analytics tools show human sessions, but logs show crawlers. A company should know how often major bots visit, which pages they request, how much bandwidth they consume, whether they respect rules, and whether crawl patterns align with business value. This is technical work with financial implications.

Crawler access also intersects with paywalls and registration. If a publisher blocks content behind login, AI systems may not see it. If the publisher exposes too much, subscription value may weaken. Flexible strategies can expose summaries, metadata, structured data, and limited previews while protecting full content. SEO teams familiar with news SEO and paywall markup have experience here, but AI crawlers add more stakeholders.

For ecommerce and marketplaces, AI crawler access may influence product discovery. If AI agents and answer systems compare products, clean access to product data can matter. But exposing price, availability, reviews, and attributes also carries competitive considerations. Retailers need coordination between SEO, merchandising, feed management, legal, and platform partnerships.

The business decision also includes attribution. If an AI platform sends referral traffic, allows source links, respects publisher terms, or offers partnership tools, access may be easier to justify. If a crawler consumes content without visible referral or compensation, blocking or charging may become attractive. This is not ideology; it is cost-benefit analysis.

The access question will keep changing. New bots will emerge. Platforms will adjust user agents. Regulators may require more choice. Publishers may negotiate deals. Search engines may blend more AI features into core products. A static robots.txt policy will not be enough.

Crawler governance is now part of search strategy. A company that says goodbye to SEO without assigning this work elsewhere is not reducing complexity. It is leaving a business decision to chance.

Paid search will not rescue weak organic strategy

When organic clicks get harder, some executives reach for paid media. That reaction is understandable. Paid search offers control, budget levers, audience targeting, and clearer reporting. If AI Overviews reduce organic visits, why not buy more visibility? The problem is that paid search is being pulled into the same AI-driven interface changes and cannot replace the strategic value of being a trusted source.

Google has been adding AI across ads, shopping, and commerce. Its January 2026 announcement around agentic commerce described tools and an open standard for agentic shopping, while earlier shopping updates brought agentic checkout into Search and AI Mode for eligible U.S. merchants. These changes suggest that paid, organic, product data, merchant feeds, and AI-assisted buying will become more connected, not less.

Paid search can capture demand, but it does not create all the information that AI systems need to answer users. An ad can put a brand in front of a buyer. It cannot by itself make the brand credible in an AI-generated comparison. It cannot replace product documentation, reviews, expert content, or external validation. If the organic information layer is weak, paid traffic may become more expensive and less persuasive because users can ask AI systems to compare claims before clicking.

AI summaries may also affect ad click-through rates. Some industry studies have reported lower organic and paid click-through rates when AI Overviews appear, though results vary by query type and market. The more important point is structural: as search results become answer pages, ad placement and user behavior will keep shifting. A company that relies only on buying visibility is exposed to platform pricing and interface changes.

Organic strategy supports paid performance in several ways. Strong landing pages improve relevance and conversion. Clear product information supports shopping feeds. Helpful content builds remarketing audiences. Brand authority improves click trust. Search demand analysis informs paid keyword selection. Technical site health supports conversion and quality. Content that answers objections reduces wasted spend. SEO and paid search are not interchangeable, but they are connected.

The AI era strengthens that connection. If a user sees a brand in an AI answer, later searches the brand, and clicks a paid ad, the paid channel may get credit for demand influenced by organic source visibility. If a user clicks an organic documentation page, then later converts through paid search, SEO supports paid. If a brand is absent from AI answers, paid may have to work harder to create trust from scratch.

This is why cutting SEO to fund paid media can be short-sighted. It may improve short-term lead flow while weakening the information base that supports future demand. The better move is integrated search governance. Paid teams should share query data, conversion insights, and ad copy tests. SEO teams should share content gaps, AI answer patterns, organic decline risks, and landing page opportunities. Product feed teams should align attributes, taxonomy, and structured data. Brand teams should align messaging and proof.

Paid media also has its own AI automation. Google Ads increasingly automates bidding, creative, targeting, and campaign formats. If companies are comfortable with AI managing more paid decisions, they need even stronger human strategy around positioning, offer quality, measurement, and landing-page truth. Otherwise automation can efficiently spend money promoting weak assets.

The boardroom lesson is that paid search is a channel, not a substitute for being understood. A brand can buy a sponsored position, but it still needs to be named, described, reviewed, cited, and trusted in the wider information system. AI search makes that wider system more influential.

The strongest search teams will not divide the world into SEO versus paid. They will manage demand capture, answer visibility, product data, and conversion paths as one commercial system. That does not mean one person owns everything. It means the old walls are too expensive.

Ecommerce SEO enters the agentic buying era

Ecommerce has always been a search discipline. Shoppers search for products, categories, reviews, prices, comparisons, sizes, local availability, shipping policies, and alternatives. Traditional ecommerce SEO focused on category pages, product pages, faceted navigation, internal linking, schema, reviews, content hubs, merchant feeds, and site speed. AI search adds a new layer: agents that help users decide, compare, track prices, and sometimes buy.

Google has been explicit about agentic commerce. Its January 2026 announcement described new technology and tools for retailers in an agentic shopping era. In November 2025, Google said agentic checkout was starting to roll out on Search, including AI Mode, with eligible U.S. merchants such as Wayfair, Chewy, Quince, and select Shopify merchants.

This changes ecommerce SEO from “rank category pages” to “make products legible and eligible across shopping decisions.” Product data quality becomes central. AI shopping systems need attributes: price, availability, size, color, material, compatibility, shipping, returns, reviews, images, variations, sustainability claims, warranty, and merchant trust. If that data is inconsistent across the site, feed, schema, marketplace listings, and Google Merchant Center, the retailer creates friction for machines and humans.

Category pages still matter, but they need to do more than display products. A strong category page explains selection criteria, filters, use cases, size guidance, comparison logic, and buying considerations. It helps users understand which product fits which need. In AI search, that material can also support answer inclusion. A category page for running shoes that only repeats “shop the best running shoes” is weak. One that explains gait, cushioning, terrain, injury history, training distance, and fit trade-offs is a better source.

Product pages need richer truth. AI can summarize product claims, but purchase decisions often require detail. Retailers should expose real specifications, compatibility notes, care instructions, return conditions, shipping timelines, stock status, reviews, Q&A, images, and videos. Manufacturer copy is usually not enough because every retailer has it. Original testing, expert picks, customer usage patterns, and comparison tools create differentiation.

Reviews and user-generated content matter because AI systems and shoppers both look for real-world evidence. But review quality is under scrutiny. Fake reviews, thin testimonials, and manipulated ratings can damage trust. Ecommerce SEO teams need to work with customer experience and moderation teams to present reviews in useful ways: verified purchase status, pros and cons, use-case tags, fit notes, images, and response handling.

Faceted navigation becomes more strategic in agentic shopping. Filters help users and machines understand product attributes, but they can create crawl traps and duplicate pages. Technical SEO teams need to decide which facets deserve indexable pages, which should be crawlable but not indexed, and which should be blocked or controlled. AI shopping may increase the value of long-tail attribute combinations, but not every filter combination deserves a landing page.

Merchant trust is also part of SEO now. Return policies, delivery reliability, customer service, payment security, store reputation, and marketplace ratings affect buyer choice. AI agents may factor these signals into recommendations. A retailer with strong product pages but poor service signals may lose. Search teams should monitor not only rankings but also trust signals across the shopping ecosystem.

Agentic checkout raises an even bigger question: where does the conversion happen? If users can complete parts of the purchase flow through Google or another AI interface, retailers may receive less site traffic but still gain sales. That changes attribution and customer relationship. Retailers will need to evaluate whether agentic transactions support margin, data ownership, loyalty, and upsell opportunities.

The old ecommerce SEO team might have focused on category copy and technical audits. The new team must understand feeds, structured data, product information management, marketplace signals, reviews, AI shopping surfaces, analytics, and merchandising. It must work closely with paid shopping, CRM, ecommerce operations, and customer support.

Ecommerce SEO is becoming product-data strategy plus trust strategy plus classic technical SEO. Teams that only write category descriptions will lose relevance. Teams that can make a retailer’s catalog understandable and trustworthy across human and machine shopping flows will matter more.

Local and service SEO becomes trust infrastructure

Local SEO may prove more resilient than broad informational SEO because local intent often requires action: call, book, navigate, visit, request a quote, check hours, read reviews, compare nearby providers. AI can summarize options, but the user still needs a business. That does not mean local SEO is safe from change. It means the work becomes more focused on trust, availability, and proof.

A local search answer needs accurate business data. Name, address, phone number, opening hours, service areas, categories, booking links, menus, pricing where relevant, accessibility, credentials, photos, and reviews all matter. If AI systems answer “best emergency plumber near me” or “dentist open Saturday for children,” they need structured and corroborated facts. A business with outdated hours, inconsistent listings, weak reviews, and vague service pages is a poor candidate.

Local SEO has long involved Google Business Profile management, citations, reviews, local landing pages, schema, and location-specific content. AI search raises the quality bar. A city page that swaps location names into generic copy is weak. A strong page shows real local proof: team members, service coverage, case examples, licenses, neighborhood references, pricing logic, parking, emergency availability, before-and-after work, and customer evidence.

Reviews become even more central because they provide external language about real experiences. AI systems may use review sentiment, common complaints, and repeated praise to shape answers, depending on platform and data access. Businesses should not chase fake reviews. They should build honest review generation, respond to issues, learn from patterns, and surface useful review details on their own site where allowed.

Service businesses also need to clarify fit. Many local pages try to rank for every possible service. AI answers may favor businesses that state boundaries clearly. A law firm should specify jurisdictions and practice areas. A clinic should specify services, insurance, providers, and appointment rules. A contractor should specify project types, materials, certifications, service radius, and timelines. Clear limits build trust.

Local content should not become generic advice. A roofing company does not need another article on “what is a roof.” It may need local storm-damage guidance, permit explanations, material choices for regional weather, insurance claim steps, and photos of completed projects. A restaurant does not need AI-written food history; it needs menu accuracy, dietary details, reservations, events, photos, and local relevance.

Voice and conversational search also blend with local intent. Users may ask, “Where can I get my tire repaired near Petržalka today?” or “Which pediatric dentist near me has good reviews and speaks English?” AI search may assemble answers from business profiles, reviews, websites, maps, and directories. Local SEO teams need to make that information consistent and current.

For multi-location brands, the challenge is scale without sameness. Each location page needs accurate data and local proof, not duplicate templates with thin differences. Reviews, staff, inventory, services, photos, offers, and local FAQs should reflect reality. Technical architecture must support location discovery without creating duplicate confusion.

Local service SEO also has regulatory and safety implications. Medical, legal, financial, childcare, home repair, and emergency services involve trust. Pages should include licenses, qualifications, insurance, safety procedures, and clear contact paths. AI-generated fluff can be harmful in these categories because it may obscure critical information.

Local SEO is not going away because local decisions still require real-world providers. But the team’s job is less about “near me” keyword insertion and more about building a trustworthy, machine-readable local presence. If a company cuts local SEO as old-fashioned, it risks becoming invisible at the exact moment AI systems are filtering options for ready-to-act users.

Enterprise SEO teams need fewer silos and better authority

Enterprise SEO has always been a coordination problem. Large organizations have many sites, templates, CMS rules, markets, languages, products, compliance requirements, analytics systems, and stakeholders. The SEO team rarely controls the assets it must improve. It persuades product, engineering, legal, content, brand, PR, analytics, and regional teams to make changes. AI search makes the coordination problem harder because visibility depends on even more signals and decisions.

A large company may have technical debt across multiple platforms. It may have old content libraries, inconsistent product naming, duplicate pages, poor structured data, slow templates, incomplete localization, and unclear ownership. Classic SEO could sometimes work around those issues by targeting keywords and building authority. AI search is less forgiving because source quality and entity clarity matter across the ecosystem.

Enterprise teams need authority to govern the public knowledge base. That does not mean SEO dictates everything. It means SEO has a recognized role in how the company is described, structured, exposed, and measured in search and answer systems. Without that authority, teams become advisors whose recommendations die in backlogs.

The first enterprise priority is a search visibility council or equivalent operating group. It should include SEO, engineering, product marketing, content, PR, legal, analytics, paid search, and customer support. The group should own decisions around crawler access, major site architecture, content quality standards, structured data, AI referral tracking, high-risk pages, and search-impacting migrations. This is not bureaucracy for its own sake. It is a way to prevent accidental damage.

The second priority is content ownership. Large sites often have pages nobody owns. They rank, decay, conflict with product truth, or mislead users. AI search makes stale content more dangerous because it can become source material for generated answers. Every important page should have an owner, review cadence, last-updated policy, and evidence source. High-risk categories need expert review.

The third priority is entity architecture. Enterprises often have multiple product names, acquisitions, sub-brands, regional variants, and legacy terms. Search systems need to understand the relationships. SEO teams should work with brand and product teams to define naming, redirects, canonical pages, organization schema, product schema where relevant, and external profile consistency.

The fourth priority is experimentation. Large sites can test changes across templates or content types, but they need careful design. AI search volatility makes tests harder, but enterprise scale can reveal patterns smaller sites cannot see. Teams should test content depth, schema improvements, internal linking, preview controls, crawler access, and page consolidation where risk is manageable.

The fifth priority is executive education. Many leaders still think of SEO as “free traffic.” That framing is outdated. Enterprise SEO now protects discoverability, source integrity, demand intelligence, technical access, and brand understanding across search and AI systems. If leadership does not understand that, SEO will remain underpowered until a traffic crisis occurs.

Enterprise teams also need to resist acronym fragmentation. It is tempting to create separate teams for SEO, GEO, AEO, LLMO, and AI visibility. That can create confusion and vendor sprawl. A better structure is one search visibility function with specialists for technical SEO, content, analytics, local/international, product data, and AI surfaces. The labels can be useful internally, but the work should be governed together.

International SEO adds another layer. AI features roll out at different times across countries and languages. Google AI Overviews expanded to many markets, while regional regulation and publisher pressure differ. A strategy that works in the United States may not apply in the European Union, where DMA scrutiny and publisher complaints shape the environment. Teams need market-specific monitoring rather than global assumptions.

Enterprise SEO also needs better tooling, but tools cannot replace authority. Crawlers, log analyzers, rank trackers, content inventories, schema validators, AI citation monitors, and dashboards are useful. They do not solve political problems. If the SEO team cannot get engineering time, product data fixed, or content reviewed, tools only reveal the gap.

The enterprise SEO team of the future is not larger by default. It is more connected, more senior, and more accountable for how the organization’s knowledge enters search systems. That is a different mandate from producing recommendations nobody implements.

Agencies face the same test as in-house teams

The agency market will feel the SEO reset intensely. Many agencies built retainers around deliverables that AI can compress: keyword research, content briefs, metadata, monthly reports, competitor summaries, blog calendars, and basic audits. Clients will ask why they should keep paying for work that software now performs quickly. Agencies that cannot answer will lose.

This does not mean SEO agencies are doomed. It means the value has to move. Agencies can still be extremely useful when they bring strategic judgment, technical depth, cross-industry pattern recognition, high-quality editorial systems, analytics, digital PR, migration experience, and the ability to challenge internal assumptions. They are weak when they sell activity instead of outcomes.

A good agency in 2026 should be able to explain how Google’s AI features work from a site-owner perspective, including the continued relevance of SEO fundamentals. It should be able to discuss OpenAI’s search crawlers, AI referral tracking, Google-Extended, crawler governance, and preview controls without pretending these settings guarantee visibility. It should be able to audit whether a site is a strong source, not only whether it has title tags.

Agencies also need to change reporting. A monthly PDF full of rank changes and generic commentary will not survive. Clients need segmentation by intent, page type, AI exposure risk, click-through shifts, conversion quality, technical constraints, content quality, and next actions. Where AI visibility tools are used, agencies should explain sampling limits. Honest uncertainty is better than decorative dashboards.

The agency content model needs the biggest change. Low-cost article production is a poor business in a world of cheap AI text. Agencies should either move up the value chain or stop selling content as a commodity. Higher-value content work includes expert interviews, original research, product-led content, technical documentation support, category narratives, case-study development, data storytelling, conversion content, and content governance. That is harder to scale, but it is harder for clients to replace with a prompt.

Technical agencies may gain demand. Migrations, JavaScript rendering, structured data, international architecture, page speed, log analysis, crawl budget, and bot governance are not solved by generative text. AI can assist audits, but implementation still requires judgment and coordination. Agencies with strong technical teams will be well positioned if they can connect their work to business risk.

Digital PR and authority work may also become more important. If AI systems rely on external corroboration, credible mentions and source authority matter. But this area is vulnerable to spam. Agencies that sell low-quality link schemes or fake expert commentary will create risk. Agencies that earn legitimate coverage, build useful data assets, and improve brand authority will be more defensible.

Clients should also change how they buy SEO. Instead of asking for a fixed number of articles, links, and reports, they should ask agencies to solve specific problems: recover from AI-driven click loss, improve product-page eligibility, consolidate content debt, build a crawler governance policy, redesign search measurement, improve entity clarity, or prepare a migration. Project-based and strategic retainers may replace generic monthly packages.

Agencies should not overpromise AI answer visibility. No one controls AI outputs fully. A credible agency can improve the probability of inclusion by improving source quality, technical access, authority, and relevance. It cannot guarantee that ChatGPT, Google AI Mode, Perplexity, or Bing will cite a client for a specific prompt every time. Guarantees in this area are a warning sign.

The best agency-client relationships will become more collaborative. Agencies need access to product experts, sales calls, customer research, data, engineers, and leadership. Without that access, they can only produce surface-level work. Clients that keep agencies at arm’s length and demand magic will be disappointed.

Agencies are not being replaced by AI. Low-judgment agency deliverables are being replaced by AI. The agencies that survive will look more like search strategy, technical consulting, editorial intelligence, and authority-building partners. The agencies that keep selling old SEO packages with new AI labels will fade.

The tasks most likely to disappear from SEO

Some SEO work should disappear. That is not a threat; it is progress. A discipline matures when it stops defending low-value labor. AI and automation are good at repetitive, pattern-based, low-context tasks. SEO has many of them. The question is whether teams use automation to free people for better work or use it to flood the web with more noise.

Metadata drafting is one example. Humans should still set rules, review important pages, and test snippets, but AI can produce first drafts for title tags and meta descriptions across large sets of pages. The value is not in typing every variation. The value is in deciding which pages matter, what intent they serve, and how snippets align with truth and conversion.

Keyword clustering is another. Tools can group queries faster than humans. But clusters are not strategy. A human must decide which clusters align with the business, which deserve pages, which should be consolidated, and which indicate user needs better served by product changes or sales material.

Basic content briefs can be automated. AI can summarize ranking pages, extract headings, list common questions, and propose structure. That is useful as a starting point. It is dangerous as a final brief because it reproduces what already exists. Human editors need to add original angle, evidence requirements, expert inputs, business context, and differentiation.

Routine reporting should shrink. No senior marketer needs a long deck saying traffic was up or down with generic commentary. Automated dashboards can surface changes. Human analysts should explain causes, uncertainty, and decisions. Reporting should become shorter and more strategic.

First-pass technical audits can be partly automated. Crawlers can flag broken links, missing tags, redirects, duplicate titles, canonical issues, and schema errors. AI can summarize patterns. Technical SEOs still need to prioritize, validate, diagnose root causes, and work with engineers. Automation finds symptoms; experts decide what matters.

Content refresh identification can be automated. Tools can flag pages with declining clicks, outdated years, lost rankings, or thin engagement. Humans should decide whether to update, merge, redirect, noindex, or delete. Many “refreshes” are cosmetic. AI makes cosmetic refreshes easier, which means teams need stronger judgment to avoid wasting time.

Competitor summaries can be automated. AI can compare pages, extract claims, and identify gaps. But competitor imitation is weaker than ever. The team’s job is to decide where competitors are right, where they are copying each other, and where the brand can create a better source.

Tasks to cut, keep and redesign

The point is not to preserve old tasks for sentimental reasons. The point is to remove weak work so search teams can spend time on problems AI does not solve alone: judgment, governance, evidence, strategy, implementation, and trust.

Some roles will be squeezed. Junior content SEO roles built around brief production and metadata may decline. Generalist SEO coordinators who only pull reports may struggle. Link builders using templated outreach will face even less tolerance. Content managers who cannot work with experts or data will be exposed. Agencies built on deliverable volume will lose pricing power.

But task disappearance should not mean knowledge disappearance. A person who has drafted metadata manually understands how snippets shape clicks. A person who has built keyword maps understands intent. A person who has done technical QA understands site risk. Teams should redesign training so people learn principles through higher-value work, not endless repetitive tasks.

AI also creates new QA needs. Someone must review AI-generated recommendations, check hallucinations, validate code snippets, confirm schema, verify facts, and ensure compliance. Automation can produce errors at scale. Human review becomes more important when the machine can make thousands of changes quickly.

The teams that win will be ruthless about low-value tasks. They will ask: does this require human judgment? Does it improve source quality? Does it reduce risk? Does it support revenue? Does it create information competitors cannot copy? Does it help machines and people understand the business better? If not, automate it, reduce it, or stop doing it.

Saying goodbye to old SEO tasks is healthy. Saying goodbye to the discipline behind them is not.

The roles that become more valuable

As routine SEO work shrinks, certain roles become more valuable. The titles may vary, but the capabilities are clear. Companies need people who can understand search systems, improve source quality, manage technical access, interpret messy data, and connect visibility to business strategy.

The technical search architect becomes more important. This person understands crawling, rendering, indexing, JavaScript, structured data, sitemaps, canonicals, hreflang, site speed, internal linking, log analysis, and crawler policy. They can work with engineers and explain business risk to nontechnical leaders. In AI search, they also understand how bots, preview controls, and source access affect visibility and content use.

The search intelligence analyst becomes more important. This person does not only report traffic. They segment demand, study query changes, monitor AI surfaces, compare click behavior by intent, connect organic visibility to conversions, and explain uncertainty. They can combine Search Console, analytics, CRM, third-party tools, server logs, and manual SERP review into decisions. They know when a decline is a traffic-quality shift and when it is a serious loss.

The source-quality editor becomes more important. This person turns expertise into publishable, retrievable, trustworthy content. They interview subject experts, demand evidence, structure pages for clarity, maintain update cycles, and protect the brand from generic AI text. They understand search intent but do not worship keyword volume. Their job is to make the company a better source.

The entity and authority strategist becomes more important. This person maps how the brand, people, products, and categories are represented across the web. They work with PR, partnerships, reviews, analysts, communities, and brand teams. They understand that AI answer inclusion may depend on signals outside the company’s own site. They help build external corroboration.

The product-data SEO becomes more important in ecommerce, SaaS, marketplaces, travel, jobs, and local platforms. This person understands feeds, schema, taxonomies, filters, product attributes, reviews, inventory, and search merchandising. They ensure that systems can understand what the company offers and why it fits a user’s need.

The search governance lead becomes more important in larger organizations. This person coordinates SEO, legal, engineering, content, PR, paid media, and analytics around crawler access, AI policies, content standards, migrations, and measurement. They keep search visibility from becoming a collection of disconnected tactics.

These roles may sit inside one team or across several teams. A small company may have one senior search strategist covering many capabilities. A large enterprise may need specialists. The key is not the job title. The key is that the work moves from production to judgment.

This shift also changes leadership. SEO leaders need broader commercial literacy. They must speak in terms of demand, revenue, risk, market positioning, customer trust, and information assets. They must stop hiding behind jargon. The board does not need a lecture on canonical tags unless canonical tags threaten revenue. It needs to know what search changes mean for acquisition, brand, and competitive position.

AI literacy is now required, but not in the shallow sense of prompt tricks. SEO teams need to understand retrieval, citations, crawl control, model limitations, hallucinations, prompt injection risk, source grounding, and answer variability. The Guardian’s reporting on ChatGPT Search vulnerability to hidden content and prompt injection is a reminder that AI search introduces new manipulation and security risks.

Legal and ethical judgment also become more relevant. AI-generated content, copyright, content licensing, medical or financial advice, review integrity, and crawler access all carry risk. SEO teams do not need to become lawyers, but they must know when legal review is necessary.

The human skills matter more too. Search leaders need to persuade engineers, interview experts, challenge weak claims, push back on executives who want shortcuts, and explain uncertainty without losing confidence. AI can produce analysis. It cannot build organizational trust by itself.

The SEO career path is not disappearing. It is becoming less forgiving. People who only know tactics will feel squeezed. People who understand systems, evidence, and business value will have more important work.

A better operating model for the next SEO team

The next SEO team should not be designed around channel maintenance. It should be designed around search visibility as an operating system. That means the team’s work should be organized by the functions that make a company discoverable, understandable, trustworthy, and measurable across classic search and AI answer surfaces.

A practical model has five pillars.

The first pillar is technical access and interpretation. This covers crawlability, rendering, indexation, site architecture, structured data, internal linking, Core Web Vitals, bot governance, migrations, and technical QA. It ensures that search and AI systems can access and understand the right material without wasting crawl capacity or exposing sensitive content.

The second pillar is source quality. This covers editorial standards, expert involvement, original evidence, content consolidation, documentation quality, update cycles, and content risk. It ensures that pages deserve to be used as sources. It also prevents the organization from filling its site with AI-generated sameness.

The third pillar is entity and authority. This covers brand consistency, product naming, author credibility, external mentions, digital PR, reviews, partnerships, local profiles, and knowledge graph clarity. It ensures that the wider web supports the company’s claims.

The fourth pillar is demand and answer intelligence. This covers query research, AI surface monitoring, intent segmentation, competitor analysis, customer questions, sales insights, support data, and emerging search behavior. It ensures the team understands how people and machines ask, refine, and answer questions.

The fifth pillar is measurement and business impact. This covers Search Console, analytics, server logs, AI referrals, citation monitoring, CRM connection, conversion quality, forecasting, and executive reporting. It ensures search work is tied to business decisions rather than activity.

This model can scale. A small business may assign these pillars to one consultant and a few internal stakeholders. A mid-size company may have three or four people and agency support. An enterprise may have dedicated specialists. The important point is that the operating model is not “publish content and report rankings.”

Work intake should change too. SEO teams should not accept every keyword request. They should evaluate requests by business value, source potential, competitive reality, and maintenance cost. A page that cannot be supported with evidence should not enter production. A technical recommendation that cannot be implemented should be escalated or deprioritized honestly. A reporting request that does not inform decisions should be removed.

The team should have quarterly search priorities tied to business objectives. For example: improve visibility for high-margin product categories, protect local search demand, reduce content debt, prepare for a migration, improve AI citation readiness for a strategic topic, or build crawler governance. Each priority should have owners, actions, success measures, and known uncertainties.

Content workflows should include expert input by design. A brief should not only list keywords and headings. It should identify the user decision, required evidence, internal experts, external sources, product truth, legal risk, conversion path, update cadence, and differentiation. AI can draft parts of it, but humans must own the substance.

Technical workflows should include search review in development. Template changes, navigation changes, JavaScript frameworks, CMS migrations, URL changes, and performance work should involve SEO before launch, not after traffic drops. AI-era crawler and rendering issues can be subtle. Preventive review is cheaper than repair.

Measurement workflows should include anomaly detection and interpretation. If impressions rise but clicks fall, the team should investigate AI features, SERP changes, query mix, rank shifts, and demand changes. If AI referrals rise, the team should study which pages and topics receive them. If bot activity spikes, technical teams should assess cost and access policy.

The operating model also needs a kill list. Which content types will the company stop producing? Which reports will be retired? Which outdated pages will be merged? Which vanity metrics will be removed? Strategy is as much about refusal as action.

The next SEO team should be smaller in busywork and larger in authority. It should have permission to improve the company’s public information system, not only permission to chase rankings.

The boardroom question should be cost of ignorance

When executives ask whether SEO teams are still needed, they are often asking a cost question. AI promises productivity. Organic traffic may be declining. Paid media and brand campaigns feel more controllable. The SEO team may look like a legacy function with unclear attribution. The boardroom wants to know whether the company can save money.

That is a fair question. Some SEO budgets are wasteful. Some teams produce little strategic value. Some agencies sell outdated deliverables. Some content programs should be cut. But the better boardroom question is not “Can we eliminate SEO?” It is “What is the cost of not understanding search at the moment search is being rebuilt?”

That cost can appear in many places. A site migration can erase traffic because no one handled redirects, canonicals, internal links, rendering, and indexation. A crawler policy can block search visibility or expose paid content. A content program can produce hundreds of pages that weaken quality. AI summaries can reduce top-funnel traffic without anyone knowing which topics are affected. Competitors can become preferred AI sources while the company watches only old rank reports. Product data can be too messy for agentic shopping. Local listings can feed wrong hours into answer systems. Legal can demand blanket blocking that harms discoverability. Paid media can become more expensive because organic trust is weak.

These are business risks. They are not “SEO tasks” in the narrow sense. They affect revenue, acquisition cost, brand perception, customer experience, and competitive position.

The board should also ask what search intelligence is worth. Search queries reveal market demand in near real time. They show how customers describe problems, which competitors they compare, which objections matter, which categories grow, and which concerns emerge. AI search may obscure some clicks, but the demand signal remains valuable. A strong search team turns that signal into product, content, sales, and brand insight.

There is also a cost to outsourcing understanding. A company can hire agencies and tools, but if no internal leader understands search, it becomes dependent on vendor claims. In the AI era, vendor claims will be noisy. Every platform will sell AI visibility. Every agency will invent a framework. Every tool will claim to track answer engines. The company needs internal judgment to separate useful support from hype.

The boardroom should also distinguish cost reduction from capability destruction. Automating metadata, reports, and first-pass briefs can reduce labor. Consolidating content can reduce maintenance. Cutting low-value agency retainers can save money. But eliminating technical search expertise, analytics interpretation, and source-quality governance can create losses that are harder to recover.

A useful executive framework is to classify SEO spend into four buckets: automate, eliminate, upgrade, and protect. Automate routine tasks. Eliminate low-value content and reports. Upgrade roles tied to source quality, technical architecture, measurement, and authority. Protect capabilities that prevent major search losses and support revenue.

The board should expect SEO leaders to present clearer business cases. “We need SEO because SEO is not dead” is not enough. A strong case might say: this product category generates €X in organic-assisted pipeline, AI Overviews affect Y% of tracked informational queries, our click-through rate has declined by Z points, competitor A is cited in AI answers for high-value questions, our documentation lacks comparison evidence, and these five changes should protect or grow qualified demand. That is a board-level conversation.

Leaders should also be honest about uncertainty. No one can forecast AI search with perfect precision. Google’s interface is changing. OpenAI is changing. Bing is changing. Perplexity is changing. Regulators are intervening. User behavior is still forming. The right response to uncertainty is not paralysis. It is building adaptive capability.

Cutting SEO because the old model is weaker is like firing the navigation team because the map changed. The company still needs people who can read the terrain. It may need different people, fewer people, or better people. It does not need ignorance.

The risk of overreacting

The SEO industry has a habit of declaring death. SEO died when social media grew. SEO died when mobile apps rose. SEO died when featured snippets appeared. SEO died when voice assistants arrived. SEO died when zero-click searches increased. SEO died when generative AI launched. Each declaration contained a piece of truth and a lot of theater. Search kept changing. The work kept changing.

The risk now is not that AI search is overhyped. The changes are real. The risk is that companies overreact in the wrong direction. They may cut SEO expertise, stop investing in useful content, block crawlers without strategy, rely only on paid media, or chase AI visibility tricks that violate platform policies. Each reaction creates damage.

Overreaction often starts with aggregate data. A study shows lower clicks when AI summaries appear. A publisher reports traffic losses. A competitor says blog traffic is down. An executive concludes that all informational SEO is dead. But query classes behave differently. A glossary page, a product comparison, a local service page, and a technical implementation guide do not face the same risk. A blanket cut can remove valuable assets along with weak ones.

Overreaction can also come from platform distrust. Some publishers and brands feel that AI systems use their content unfairly. That concern is legitimate enough to reach regulators and courts. But a blanket block of AI crawlers may reduce visibility in systems where users are shifting attention. The right policy depends on content type, business model, and platform behavior. Anger is not a crawler strategy.

Another overreaction is chasing manipulation. As AI answers become valuable, some marketers will try to poison prompts, flood the web with brand mentions, create fake comparison pages, manipulate reviews, or produce content aimed only at LLMs. Google’s spam policies are already expanding around attempts to manipulate generative AI search outputs. This path may produce short-term experiments, but it creates long-term risk.

Overreaction can also produce measurement panic. If clicks fall, teams may label SEO a failure without checking whether conversions, lead quality, brand search, or assisted revenue changed. A decline in low-intent visits may not be catastrophic. A decline in high-intent visits is serious. Teams need segmentation before budget decisions.

The reverse overreaction is denial. Some SEO teams insist nothing has changed because Google says SEO is still relevant. That is also wrong. Google’s guidance confirms fundamentals, not old traffic guarantees. AI summaries, AI Mode, answer engines, and agentic shopping change interfaces, user behavior, and click economics. Teams that use “SEO still matters” as an excuse to keep old workflows will lose credibility.

A balanced response is not a diplomatic middle. It is a specific diagnosis. Which parts of the site are most exposed to answer satisfaction? Which pages still earn qualified visits? Which topics need original evidence? Which technical barriers prevent source quality? Which AI platforms send traffic? Which crawlers access expensive content? Which competitors are cited? Which reports are obsolete? Which roles are underused? Which vendor work can be automated or cut?

The companies that handle this well will not swing wildly. They will run audits, reduce weak content, improve source quality, revise measurement, govern crawler access, train teams, and test new surfaces. They will accept that some old traffic will not return. They will also see that search remains one of the clearest windows into customer intent.

The danger is not saying goodbye to bad SEO. The danger is saying goodbye to search understanding because bad SEO disappointed you. Those are different decisions.

The verdict: retire the old SEO team, not search expertise

It is time to say goodbye to a certain kind of SEO team. The team that exists to produce generic content at scale should be retired. The team that reports rankings without business interpretation should be retired. The team that treats technical SEO as a checklist rather than a governance function should be retired. The team that ignores AI search, crawler policy, entity clarity, and source quality should be retired. The team that cannot explain revenue impact should be challenged.

But it is not time to say goodbye to SEO expertise. Search remains a central way people express intent. Google remains a dominant discovery system. AI answer engines still need sources. Ecommerce, local services, B2B research, professional advice, travel planning, product discovery, and news consumption still involve search-like behavior, even when the interface no longer looks like ten blue links.

The work is changing from page ranking to search visibility. That includes classic organic rankings, AI Overviews, AI Mode, ChatGPT Search, Bing generative search, Perplexity, local packs, shopping surfaces, video, images, community results, and direct brand demand. It includes both human clicks and machine use of public information. It includes content, technical access, authority, measurement, and governance.

The new SEO team should be judged by different questions.

Can it make the company’s expertise easier to find, verify, and use? Can it identify where AI search is reducing clicks and where it is creating new opportunities? Can it improve technical access without giving away content blindly? Can it build pages that deserve citation because they contain real evidence? Can it coordinate with product, PR, legal, engineering, paid media, and analytics? Can it explain uncertainty without hiding behind jargon? Can it cut weak work? Can it defend search investment in business language?

If the answer is yes, the team is not obsolete. It is one of the few teams equipped to manage a difficult transition.

The phrase “SEO team” may eventually change. Some companies will call it organic growth. Some will call it search experience. Some will call it discovery. Some will fold it into content strategy, product marketing, or demand intelligence. The label matters less than the capability. Every company that depends on being discovered needs people responsible for how its information is found, interpreted, trusted, and acted upon.

The search world of 2026 is less generous than the search world many SEO teams grew up in. It gives fewer easy clicks. It hides more influence inside answers. It demands better sources. It rewards stronger brands. It punishes generic production. It creates new technical and legal choices. It makes measurement harder. None of that is a reason to abandon SEO. It is a reason to stop treating SEO as a cheap traffic machine.

The honest answer to the user’s question is this: say goodbye to SEO teams as they were commonly built. Do not say goodbye to search expertise. The companies that understand the difference will cut waste without cutting their ability to be found.

Questions executives are asking about SEO teams now

Author:
Jan Bielik
CEO & Founder of Webiano Digital & Marketing Agency

This article is an original analysis supported by the sources cited below

Google’s guide to optimizing for generative AI features on Search
Google’s official guidance explaining that SEO remains relevant for AI Overviews, AI Mode, and related generative search features.

AI features and your website
Google Search Central documentation describing how AI Overviews and AI Mode work from a site-owner perspective.

AI in Search: Going beyond information to intelligence
Google’s official May 2025 announcement explaining AI Mode and query fan-out.

A new era for AI Search
Google’s May 2026 announcement on agentic AI features and the upgraded AI-powered Search experience.

Expanding AI Overviews and introducing AI Mode
Google’s March 2025 announcement on AI Overviews expansion and AI Mode testing.

Google Search’s guidance about AI-generated content
Google’s official explanation of how AI-generated content fits into its people-first content standards.

Google Search guidance on using generative AI content
Google documentation on using generative AI in website content without violating spam policies.

Google Search Essentials
Google’s baseline technical, spam, and best-practice requirements for appearing in Search.

Spam policies for Google web search
Google’s official policies on spam tactics and behaviors that can lead to lower rankings or removal.

Introduction to structured data markup in Google Search
Google documentation explaining how structured data helps Search understand page content and entities.

Understanding Core Web Vitals and Google search results
Google guidance on Core Web Vitals, user experience metrics, and their relationship to Search.

Google common crawlers
Google crawler documentation including Google-Extended and its role in managing some Gemini-related content uses.

Robots meta tag specifications
Google documentation on page-level indexing and preview controls.

Introducing ChatGPT search
OpenAI’s announcement of ChatGPT search with source links and a reference sidebar.

SearchGPT prototype
OpenAI’s July 2024 announcement of its search prototype with timely answers and source links.

ChatGPT Search
OpenAI help documentation explaining ChatGPT Search and web source links.

Overview of OpenAI crawlers
OpenAI documentation on GPTBot, OAI-SearchBot, and crawler controls for site owners.

Introducing Copilot Search in Bing
Microsoft’s announcement of Copilot Search in Bing and AI-generated search layouts.

Introducing Bing generative search
Microsoft’s official explanation of generative search in Bing.

Do people click on links in Google AI summaries?
Pew Research Center analysis of click behavior when Google AI summaries appear.

2024 zero-click search study
SparkToro and Datos analysis of zero-click behavior in U.S. and EU Google searches.

AI Overviews reduce clicks by 34.5%
Ahrefs study estimating lower click-through rates for top-ranking pages when AI Overviews appear.

Update: AI Overviews reduce clicks by 58%
Ahrefs follow-up analysis using its own Search Console data to reassess AI Overview click impact.

Semrush report: AI Overviews’ impact on Search in 2025
Semrush analysis of AI Overview growth, volatility, and intent distribution across keywords.

SERPs with ads and AI Overviews grew by over 394% in 2025
Semrush analysis showing AI Overviews moving beyond purely informational intent.

Generative AI statistics for 2026
Similarweb analysis of AI platform visits and AI referral traffic trends.

Enabling content owners to charge AI crawlers for access
Cloudflare’s announcement of Pay Per Crawl and AI crawler monetization controls.

The crawl-to-click gap
Cloudflare analysis of AI bot crawling, publisher referrals, and the widening crawl-to-click imbalance.

Cloudflare launches tool to help website owners monetize AI bot crawler access
Reuters report on Cloudflare’s AI crawler monetization tool and publisher support.

Google’s AI Overviews hit by EU antitrust complaint from independent publishers
Reuters report on the EU antitrust complaint over Google AI Overviews and publisher traffic concerns.

Italy’s media regulator asks EU to investigate Google AI search tools over publisher concerns
Reuters report on AGCOM asking the European Commission to examine Google’s AI-powered search features.

AI startup Perplexity adds The Independent, LA Times to its publishers’ program
Reuters report on Perplexity’s publisher program expansion and revenue-sharing model.

Artificial intelligence: Le Monde signs partnership agreement with Perplexity
Le Monde announcement of its Perplexity partnership and content-use terms.