An identifier, an auction, and a set of segments attached to it. Here is the chain that turns scattered page views into a profile, and what actually breaks it.
On this page
You open a page about running shoes. Somewhere else on the web, an ad for running shoes appears.
The word “profile” suggests a dossier assembled deliberately and stored in one place. The reality is less tidy and more effective: labels attached to an identifier, updated as pages load, and used the instant an ad slot needs filling. Following the chain from a single page view to a bid explains what advertisers actually know, and where the leverage sits.
The identifier on the page
Everything starts with a way to recognise the same browser twice. Historically that was a cookie. When it worked, the cookie was the thread: every participating site loaded a request to a common domain, the cookie came along, and a note was made that this identifier had been seen.
When third-party cookies are blocked, the identifier has to come from elsewhere. A site can use its own first-party cookie and pass the identifier onward itself, or it can compute a browser fingerprint — a value derived from your device rather than stored in it. Either way the goal is the same: a stable handle to hang data on.
The auction behind every ad slot
The identifier makes everything after it possible. When a page loads an ad slot, it does not display a stored advert; it runs an auction, often within the time it takes the page to render. The page sends what it knows — the site, the page context, and the identifier — to an ad exchange, which gathers bids from advertisers who want to reach someone matching it. The highest bid wins, so the advert is the result of a competition rather than a fixed placement.
Segments are the profile
What advertisers bid on is not you as a person but the segments attached to your identifier. A segment is a label: a category, an interest, a behaviour, an inferred attribute. The profile, in practical terms, is the current set of segments associated with the handle.
| Piece | What it refers to |
|---|---|
| Identifier | The handle a site recognises on a return visit |
| Segment | One label: a category, interest, behaviour or inferred trait |
| Profile | The current set of segments attached to that handle |
Segments come from converging evidence. A visit to a page about a product, a search, a purchase, a location, a time of day, a device type — each is a signal, and signals accumulate into patterns more revealing than any single event. None of it requires your name, which is precisely the point: an anonymous but persistent identifier is enough to target advertising.
On-site versus cross-site
The distinction that matters most is the boundary of that profile.
| On-site profile | Cross-site profile | Held together by | |
|---|---|---|---|
| Built by | The site you are on | A network across many sites | A shared identifier |
| Recognises you | Only there | Wherever the network operates | The same handle |
| Survives clearing cookies | Usually not | Often | A fingerprint fallback |
| Laws apply | Where the company sits | Each site’s jurisdiction | Where you are sitting |
An on-site profile is a company remembering what you did on its own pages; you agreed to that relationship, and the data stays inside it. A cross-site profile is assembled from many unrelated sites by a party you never visited directly — what most people mean when they worry about being followed. It depends on a shared cookie or a shared fingerprint to hold the pieces together.
What happens when a cookie is cleared
Clearing cookies removes the handle, and on its own that is effective: without the identifier, new events cannot be attached to the old segment set. This is why the tracking pixels and third-party cookies story matters — much of the architecture was designed around an identifier you could delete.
But if the site can recognise your browser by fingerprint, the handle is regenerated on the next visit and the profile reattaches. Delete the cookie and the data is not lost if the identity can be reconstructed; the fingerprint is the fallback that survives the removal.
What actually breaks the chain
Real leverage comes from attacking the link between you and the identifier, not from scrubbing the profile once it exists.
- Your fingerprint’s stability — since the fallback is what reclaims you, the durable fix is blending in rather than standing out. A mainstream browser in its default state joins a crowd far larger than a heavily customised one, as laid out in how to reduce your fingerprint.
- The machinery you can see — content blockers stop the requests that transmit identifiers. An ad block detector shows what your setup is filtering.
- Your stated preference — opt-out signals do not break the technical chain, but they carry weight in some jurisdictions. The Do Not Track test shows whether your browser is transmitting one.
- The continuity between sessions — clearing data between sessions, or isolating sites from one another, limits how much can accumulate under a single handle.
Where the chain gets away from you
Run the browser fingerprint test to find out how much of your device is distinctive, then check the ad block detector to see what your browser is already blocking. The first tells you how easily you could be re-identified; the second tells you how much of the visible machinery is still reaching you.
None of it is complete. The chain is resilient because each link has a replacement, and the replacements move further from your control the deeper you go. Every fix narrows the opening rather than closing it: this is a decision you keep making, not a setting you finish.