Quick answer
Cross-site tracking is recognising the same browser or person on websites run by different companies and joining their activity into one record, using third-party cookies, fingerprinting or ids added to links. Browsers block these cross-site mechanisms, not a site's own first-party analytics.
What is cross-site tracking?
Cross-site tracking is the practice of recognising the same browser or person on websites run by different companies, and joining what they did on each into one record. The record is then used to target ads, measure ad conversions or build profiles that are sold or shared. When you see a pair of shoes on a shop's site and then see ads for the same shoes on a news site, cross-site tracking connected the two visits.
The key word is cross. A shop counting its own visitors, pages and orders is measuring activity on its own site. That becomes cross-site tracking only when an identifier links those visits to activity on other sites. This distinction is the core of this page: cross-site tracking relies on third-party cookies, browser fingerprinting and link parameters to connect behaviour across different sites. Browsers block those cross-site mechanisms, not a site's own first-party analytics.
The three mechanisms
| Mechanism | How it links sites | Example | What browsers do about it |
|---|---|---|---|
| Third-party cookies | A tracker's script or pixel loads on many sites and reads the same cookie set on its own domain | An ad network's pixel on a shop and on a blog | Safari blocks them by default; Firefox partitions them per site; Chrome offers users a choice and blocks them by default in Incognito |
| Browser fingerprinting | A script combines device and browser traits (fonts, screen, settings) into an identifier that is the same on every site | A fingerprinting script embedded by an ad tech vendor | Safari limits exposed traits; Firefox blocks known fingerprinters; Apple's app rules forbid it |
| Link decoration and redirects | A click id or user id is added to the URL (?gclid=…, ?fbclid=…) or the click bounces through a tracker domain, so the destination site can store it and send it back | An ad click that lands on /product?click_id=abc123 | Safari caps how long script-set storage from such landings survives; bounce tracking domains lose their data |
Pixels, social widgets and embedded videos aren't separate mechanisms. They are delivery vehicles: each loads a third-party resource that can read a cookie, run a fingerprinting script or receive a decorated link.
What each browser does
The behaviour below is taken from each vendor's own documentation, checked on 2026-11-15. Browser defaults change; check the linked pages before you rely on a detail.
Safari (Intelligent Tracking Prevention). WebKit's tracking prevention policy says Safari blocks third-party cookies by default with no exceptions; sites can ask for access through the Storage Access API. Full third-party cookie blocking shipped in Safari 13.1 in March 2020. Beyond cookies, Safari partitions third-party storage per first-party site, deletes script-writable storage (cookies set by JavaScript, LocalStorage, IndexedDB and similar) after 7 days without user interaction, caps cookies set by JavaScript on a landing page that arrived with click-id link decoration to 24 hours, and deletes all data for domains it classifies as trackers unless the user interacted with them as a first party within the last 30 days of browser use. Apple's Safari guide describes the "Prevent cross-site tracking" setting the same way: a third-party provider's cookies and data are deleted unless you visit and interact with it directly, and websites see a simplified system configuration so fingerprinting is harder.
Firefox (Enhanced Tracking Protection and state partitioning). Mozilla's developer documentation says Firefox has partitioned client-side state by top-level site by default since Firefox 103 (2022). A third-party cookie set while you're on site A lives in a separate jar from the same third party's cookie on site B, so the tracker sees two strangers. Third parties can get unpartitioned access only through the Storage Access API or built-in heuristics. Mozilla's Enhanced Tracking Protection page lists what the default Standard mode blocks on desktop: social media trackers, cross-site tracking cookies, cryptominers and known fingerprinters, with tracking content also blocked in private windows.
Chrome. Chrome's help centre lists options to allow or block third-party cookies and says they are blocked by default in Incognito mode. In October 2025 Google said it would keep its current approach of offering users a choice about third-party cookies, and retired most Privacy Sandbox technologies, including Topics, Protected Audience and the Attribution Reporting API. In practice, many regular Chrome sessions still allow third-party cookies, but you can't count on all of them.
None of these defaults block a site from counting its own visits. They target the shared identifier that makes the same browser recognisable on site A and site B.
Effect on analytics and ad attribution
Two kinds of measurement live on the same page, and browsers treat them very differently.
- Your own analytics count visits, pages, sources and goals on your site. Browser tracking protection doesn't block that. What changes is how long a visitor can be recognised if the tool relies on cookies set by JavaScript: in Safari those expire after 7 days without interaction, so a weekly returning visitor may look new. A content blocker can still block an analytics script entirely; that's an extension's choice, not the browser's default tracking protection.
- Ad conversion measurement sends conversions back to an ad platform that saw the click on another site. That is cross-site by definition, so it loses data when third-party cookies or long-lived click ids are blocked.
Worked example: the conversion postback gap
Sample data: a campaign brings visitors to a shop that records 400 purchases in its own order system. Purchases split by browser: Safari 180, Firefox 20, Chrome and others 200. Assume, for the example, that every Chrome session allows third-party cookies and that 60% of Safari purchases happen within 24 hours of the ad click.
- Third-party cookie pixel only. The ad platform's pixel reads its own cookie. Safari blocks it and Firefox partitions it, so only Chrome purchases are matched: 200 of 400 = 50% reported.
- First-party click id. The landing page stores the click id in a first-party cookie set by JavaScript and sends it with the purchase. Chrome: 200. Firefox: 20 (first-party storage is fine). Safari: the cookie is capped at 24 hours, so only the 60% within a day survive: 180 × 0.6 = 108. Total 200 + 20 + 108 = 328, or 82%.
- Gap left. 400 − 328 = 72 purchases (18%) the ad platform never hears about, all from Safari buyers who came back after a day.
The shop's own count stays at 400 in every case. The ad platform's number moves from 200 to 328 depending on how the conversion is passed back. If you compare the two and conclude "our analytics is over-counting", you've drawn the wrong conclusion: the browser removed the cross-site link, not the purchase.
How to prevent cross-site tracking (as a user)
- Safari: keep "Prevent cross-site tracking" selected in Settings › Privacy and use Private Browsing for sessions you don't want joined.
- Firefox: keep Enhanced Tracking Protection on Standard, or switch to Strict for more blocking (some sites may break).
- Chrome: under Settings, choose to block third-party cookies, or use Incognito, where they are blocked by default.
- Any browser: clear cookies regularly, sign out of large platforms when you're not using them, and be careful with links that carry long ids in the query string.
Common misconceptions
- "Blocking cross-site tracking breaks all analytics." It removes the shared cross-site identifier. Counting pages and visits on one site still works; cookie-based tools may see more "new" visitors in Safari.
- "First-party cookies can't be used for tracking." A first-party cookie can carry a click id that is sent back to an ad platform, which recreates a cross-site link. That's why Safari caps those cookies after link-decorated landings.
- "No cookies means no consent is needed." Consent rules depend on what you collect, the purpose and local law, not only on whether a cookie is set. See cookieless analytics, the GDPR entry and do I need a consent banner. This page is not legal advice.
- "Fingerprinting is a cookieless alternative." It's still cross-site tracking, browsers actively limit it, and Apple's developer rules forbid it in apps.
For how brands buy and combine other companies' audience data, see third-party data. How identifiers are joined into one person inside your own data is covered in identity graph.
Cross-site tracking and TapCub
The usual question is: "Safari blocks cross-site tracking, will my TapCub numbers drop?" Here is how the setup above plays out:
- No cross-site identifier to block. By default the website script writes no cookie and no local storage. The collector separates visitors with a hash of address and browser data and a site salt, so the id only means something inside one project. TapCub doesn't track visitors across other companies' sites and doesn't run ad retargeting. Safari's 7-day cap on script-set storage has nothing to shorten, although a content blocker can still block the script.
- Privacy signals are honoured. If the browser sends Global Privacy Control or Do Not Track, the website script doesn't run by default, so those visitors are absent from TapCub and may still appear in other tools.
- Ad click ids are kept as types, not raw values. Web analytics › Sources classifies visits that carry ids such as
gclidormsclkidas paid search, and the ad click id table shows only the type and a hash. TapCub measures your own site; it doesn't return conversions to ad networks by matching them to clicks. - Consent is a configuration choice. Configuration › Privacy center sets the collection tier and region rules. The default rule for the EEA and the UK requires consent; without analytics consent the hit drops to anonymous tier 1, and pageviews and custom events are still recorded. Whether you need a banner depends on your configuration and local law, so TapCub doesn't claim you can skip consent.
In the worked example, TapCub's goal count would sit with the shop's own 400 orders (minus visitors whose browser sent a privacy signal), not with the ad platform's 200 or 328.
Sources
- WebKit, Tracking Prevention in WebKit.
- WebKit Blog, Full Third-Party Cookie Blocking and More (March 2020).
- MDN Web Docs (Mozilla), State Partitioning.
- Google Chrome Help, Delete, allow and manage cookies in Chrome.
- Google Privacy Sandbox, Update on plans for Privacy Sandbox technologies (October 2025).
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Solutions Engineer
Covers SDKs, API integrations, and data implementation, with an emphasis on validation, system interoperability, and troubleshooting.