Persistent identifiers let websites and advertisers recognise a browser or device across visits, even after users think they’ve removed tracking. Understanding the technical methods behind persistent tracking is the first step toward effective defenses.
This explainer breaks down common techniques used to create persistent IDs, gives real-world examples, and offers practical mitigations for both end users and site administrators.
Common persistent tracking techniques explained
Storage-based identifiers
Trackers use whatever client-side storage is available to store identifiers tied to a browser. Cookies are the oldest and most visible option, but they are just one of many places an identifier can live.
Cookies, localStorage, IndexedDB, ETags, HTTP cache, and legacy Flash or Silverlight storage are all used by trackers to rebuild identifiers even after deletion.
Redundancy: the evercookie pattern
Rather than relying on a single storage slot, some techniques write the same identifier into multiple locations so removing one copy is not enough.
Evercookies (also called ‘supercookies’) are techniques that redundantly store the same identifier across multiple storage mechanisms so that removing one copy does not remove others.
Fingerprinting
Fingerprinting collects many small, often innocuous signals from a device to create a probabilistic identifier. Each signal by itself is weak, but combined they can be highly distinctive.
Canvas and audio fingerprinting collect tiny rendering differences from a device’s GPU, drivers, font rendering, and audio stack to produce a device fingerprint.
Network and cache-level identifiers
Servers can use long-lived cache entries, ETags, or application-level tokens that are returned by the browser on subsequent requests. Service workers and the cache API increase the persistence options available to web applications.
Service workers and persistent cache entries can be abused to re-establish identifiers independent of standard cookie clearing.
Third-party and first-party combinations
Trackers increasingly use combinations of first-party and third-party channels: a high-privilege first-party cookie can be used to link activity observed by multiple third parties. Blocking only third-party cookies is no longer a complete answer.
Real-world examples and what they look like
Standard cookies and SameSite behaviour
Cookies remain widely used for session management and tracking. Server-side attributes control their scope and behaviour: Domain, Path, Secure, HttpOnly and SameSite determine when and how cookies are attached to requests.
SameSite cookie attribute and Secure flag are server-side controls that influence when cookies are sent and whether they require HTTPS.
localStorage and IndexedDB
These client-side databases are persistent until explicitly cleared. A script can store identifiers and later read them to re-identify a browser without touching cookies.
Canvas and audio fingerprinting examples
A page can render hidden text to an HTML canvas and read pixel outputs, or perform an audio rendering and analyse the output. Small differences introduced by the device produce a fingerprint that is hard to change without altering the device or browser.
ETags and cache-based rebuilds
ETags and cache-control responses can be used as a covert storage mechanism because the browser will continue to send validator headers that reveal prior state.
Browser and extension defenses
User-facing browser settings
- Enable built-in tracking prevention or strict privacy mode. These modes block known trackers and isolate storage for third parties.
- Use private browsing or profiles for sensitive sessions; private mode isolates some storage and reduces cross-session state.
- Periodically clear site data, cookies, and browser caches, and review which sites have persistent storage rights.
Fingerprinting-specific protections
Anti-fingerprinting features aim to reduce the uniqueness of a browser by standardising responses or denying access to high-entropy APIs. They are not perfect, and some legitimate sites may break.
Browser privacy modes and tracking prevention settings reduce many forms of cross-site tracking but do not eliminate all fingerprinting vectors.
Careful use of extensions
Extensions can enhance privacy, but they also add risk. Only install well-reviewed extensions and limit permissions to what is strictly necessary. Be cautious with extensions that request broad host access.
Extension host permissions let extensions read and modify page content when granted, which can expose identifiers to malicious or poorly reviewed extensions.
Server-side and architectural techniques to reduce tracking
Minimise and partition identifiers
Design systems so that long-lived, cross-site identifiers are not issued unnecessarily. Use short-lived tokens for sessions and avoid using the same identifier across different services.
Use cookie attributes and storage policies
Set cookies with appropriate Scope, SameSite and Secure attributes. Ensure cookies that must not be shared with third parties are not sent in cross-site contexts.
Adopt privacy-preserving analytics
Consider server-side analytics that aggregate behaviour without creating device-level identifiers, or adopt differential privacy and sampling to reduce identifiability.
Cache control and ETag hygiene
Avoid using caching headers that could be repurposed as identifiers. Treat validator headers and client cache behaviour as part of your threat model.
How to audit your site or browser for persistent IDs
Audit steps for site owners
- Inventory: catalog all third-party scripts, pixels and SDKs included on pages.
- Storage scan: use browser developer tools to inspect cookies, localStorage, IndexedDB and service worker registrations after a clean visit.
- Network analysis: review request headers and responses for ETags, Set-Cookie, and unusual cache behaviour.
- Threat modelling: identify where identifiers might be used to link cross-site activity and design mitigations (partitioning, short lifetimes, consent gating).
Audit steps for users and admins
- Use the browser’s privacy dashboard or developer tools to list third-party domains contacted by a page.
- Inspect storage entries after visiting a site; remove unexpected data and retest to see if identifiers reappear.
- Install reputable privacy extensions that block known trackers, and review their logs to understand blocked connections.
- For more comprehensive checks, replicate visits in a fresh profile and compare stored state.
Auditing tools like browser developer tools, privacy extensions, and server logs can reveal persistent identifiers and third-party requests.
Checklist: actions to reduce persistent tracking
- For users: enable tracking prevention, use privacy-conscious profiles, clear site data regularly, and limit extension permissions.
- For site admins: limit cross-site identifiers, set cookie attributes correctly, avoid using cache headers as storage, and prefer aggregated analytics.
- For both: review included third-party scripts and document why each needs persistent storage.
Closing: start small, iterate
Persistent tracking is a layered problem: storage mechanisms, fingerprinting, and network behaviours can all contribute. Begin by auditing the obvious storage and third parties, then apply incremental controls—strong cookie attributes, storage partitioning, and privacy-focused analytics. Users should enable browser tracking prevention and limit extensions, while site operators should prioritise reducing shared, long-lived identifiers.
For a quick privacy refresh on mobile, consider the practical tips in 5 Privacy Settings You Should Change on Your Phone Today, and if you’re reviewing account hygiene alongside tracking, see A Beginner’s Guide to Password Managers and Passkeys for related advice.