Location tracking

On this page 7

Location tracking is the collection, storage and analysis of data that reveals where a person or device is, either at a single moment or as a trajectory over time. In privacy, it is treated as one of the most sensitive forms of tracking because a sequence of positions exposes home and work addresses, routines, relationships, health visits and religious practice without the person ever stating any of them.

Location data reaches companies and authorities through device sensors, network infrastructure, apps, advertising software and connected vehicles. Privacy law, platform permission systems and privacy engineering techniques all aim to limit who can collect it, how precisely, for how long and for which purpose.

How location tracking determines position

Location tracking works by combining positioning signals from several sources, each with its own accuracy and its own party that sees the data. A smartphone typically fuses satellite positioning, Wi-Fi scans, cell tower identifiers and Bluetooth beacons into one estimate, and the operating system then passes that estimate to apps that hold the relevant permission.

The table below compares the main positioning methods by typical accuracy and by who can observe the signal, which determines where privacy controls must apply.

MethodTypical accuracyWho observes itWorks without user action
GNSS (GPS, Galileo)3 to 10 meters outdoorsThe device, then any app granted accessNo, requires an app or OS service
Wi-Fi positioning10 to 50 metersThe OS vendor and location database operatorsPartly, through background scans
Cell tower data100 meters to several kilometersThe mobile network operatorYes, always while connected
Bluetooth beacons1 to 5 meters indoorsVenue operators and SDKs in appsPartly, if scanning is enabled
IP geolocationCity or region levelEvery website and online serviceYes, on every connection
Positioning methods and their privacy exposure

Accuracy alone does not define the risk. Network-side data such as cell tower records is generated whether or not the user opens an app, which is why it is often the basis of lawful access requests and why device-level permission settings cannot fully prevent it.

Why location data is so revealing

Location data is highly identifying because human movement is regular and unique. Research on mobility traces has repeatedly shown that a handful of time-stamped points, often four or fewer, is enough to single out most individuals in a large dataset, even when names and device identifiers have been removed.

From repeated positions an analyst can infer sensitive attributes that are often protected as special categories of data under the GDPR and similar laws. Visits to a clinic suggest health conditions, nightly stays reveal the home, and co-location with another device over weeks reveals relationships. This is why regulators increasingly treat precise location as sensitive data in its own right, as the CPRA in California explicitly does for precise geolocation within a radius of about 1,850 feet.

  • Home and workplace inference from overnight and weekday dwell times
  • Health and religion inference from visits to clinics, pharmacies or places of worship
  • Social graph inference from repeated co-location with other devices
  • Behavioral profiling for advertising from shop visits and travel patterns

Location tracking differs from other techniques in the Tracking family because it follows the person through physical space rather than through websites or apps. It often works together with them, since an advertising ID or a device fingerprint is what links location points to a single profile.

TechniqueWhat it followsTypical identifierRelation to location
Location trackingPhysical movementDevice or advertising IDCore subject
Web trackingBrowsing activityCookies and scriptsAdds IP-based location
Device fingerprintingDevice configurationDerived fingerprintLinks location across apps
Advertising IDApp usageResettable mobile IDMain key in location data markets
Behavioral advertisingInterests and habitsProfile segmentsUses visit history for targeting
How location tracking relates to neighboring techniques

A particular concern is the location data broker market. Software development kits embedded in ordinary apps collect positions, attach the advertising ID and sell the stream onward, so a user who granted location access to a weather app may unknowingly feed dozens of downstream buyers.

Location tracking is governed by consent requirements, purpose limitation and platform permission models that operate in parallel. In the European Union, the ePrivacy Directive requires consent for processing location data from electronic communications beyond what is needed to deliver the service, and the GDPR requires a lawful basis and data minimization for all other location processing.

Mobile operating systems add technical controls that narrow what apps can obtain. The following table summarizes the main levers and what each protects against.

ControlMechanismWhat it limits
Foreground-only permissionAccess only while the app is openContinuous background trails
Approximate locationOS returns a coarse area, about 1 to 3 kmAddress-level inference
One-time permissionAccess ends when the app closesLong-term silent collection
Wi-Fi and Bluetooth scan controlsDisables background radio scanningIndoor and beacon tracking
Consent under ePrivacy and GDPRLegal requirement before collectionUnlawful secondary use
Main controls that limit location tracking

Enforcement has targeted brokers that sold visits to sensitive locations, and several authorities now regard the sale of precise location without clear, specific consent as an unfair practice. A consent screen buried in terms of service is generally not considered valid consent for this purpose.

Privacy engineering for location data

Privacy engineering reduces location risk by collecting less, coarsening what is collected and deleting it early. Simply removing names does not work, because trajectories re-identify people; effective protection changes the data itself.

TechniqueHow it worksMain trade-off
Spatial generalizationRounds positions to grid cells or postal areasLoses street-level utility
Temporal aggregationReports counts per hour or day instead of pointsHides individual journeys
Differential privacyAdds calibrated noise to aggregate statisticsReduces accuracy for small areas
On-device processingKeeps raw positions on the phoneLimits server-side analytics
Short retentionDeletes raw traces after days, not yearsRestricts historical analysis
Engineering techniques and their trade-offs

The most common mistake is to treat hashed device IDs or truncated coordinates as anonymization. Hashing keeps records linkable, and three decimal places of latitude still locate a person within roughly 100 meters. A sound design starts from the purpose, asks what minimum precision and retention serve it, and documents that decision in a data protection impact assessment.

Stripped to the essentials

Location tracking means keeping a record of where you go. Your phone knows its position from satellites, nearby Wi-Fi networks and phone masts, and apps that you allow can see that position too. Some of them pass it on to other companies.

One location point says little, but a trail of points tells a story: where you sleep, where you work, which doctor you visit and who you spend time with. That is why location history is treated as very sensitive. The best protection is to give apps location access only while you use them, choose approximate location when exact position is not needed, and remove access from apps that have no clear reason to know where you are.

Where it is applied

A city transport authority wants to know how commuters move between districts so it can redesign bus routes. A vendor offers raw, per-device location trails bought from app SDKs. The authority’s data protection officer rejects the offer because the trails would allow staff to identify individual residents, and the source consent is doubtful.

Instead, the authority asks the mobile network operator for aggregated origin-destination counts: trips between 1-kilometer cells, grouped by hour, with any cell pair under 20 trips suppressed and differential privacy noise added. No device identifiers leave the operator. The planners still see that morning demand between two suburbs is twice the current capacity, and they add an express line. The result is useful mobility insight without building a database of individual movements, documented in an impact assessment that the authority publishes with the route plan.