For twenty years, ad measurement rested on one trick: a third-party cookie set on one site and read on another. Safari blocked it in 2017, Firefox in 2019, and Chrome spent years dismantling it. That trick is gone.
What actually broke
Cross-site retargeting. Following someone from one site to another to show them the same product. Gone, and good riddance.
Third-party frequency capping. Knowing a person saw an ad on a different site. Gone.
View-through attribution. Crediting a sale to an ad someone saw but did not click. Largely gone.
What still works perfectly
Click-based attribution. A click is a navigation. You can attach an identifier to the destination URL, and the destination site can store it in its own first-party storage. Nothing about that depended on third-party cookies.
First-party frequency capping. Counting how many times someone saw an ad on your site only needs storage on your site.
Impressions, clicks, viewability. Never needed cookies at all.
Click-ID passthrough, concretely
- Ad renders. Click URL points at the ad server.
- Visitor clicks. The server records the click, generates an id, and redirects to the advertiser with
?vt_cid=abc123appended. - The advertiser's page reads that parameter and stores it in their own localStorage — first-party to their domain.
- On purchase, their tag sends the id back with the order value.
- The click and the sale are joined.
No shared cookie. No cross-site identifier. The id means nothing outside the advertiser's own site.
Why small publishers come out ahead
The cookie apocalypse hurt the ad-tech middle far more than it hurt direct relationships.
A programmatic network needed cookies to know who someone was across thousands of sites. You do not. Your pitch was never "we can follow this person" — it was "the people reading this page are interested in this subject". Contextual relevance never depended on tracking.
If anything, the collapse of behavioural targeting makes context more valuable. A sponsor who can no longer retarget efficiently has to go back to putting the right message on the right page.
The honest limitations
Click-ID attribution is not perfect, and you should say so rather than let a sponsor discover it:
- Cross-device is lost. Click on a phone, buy on a laptop — unattributed.
- Cleared storage loses the link. Rare within a 30-day window, but real.
- View-through is not covered. Someone who sees the ad, remembers it, and searches for the brand later shows up as unattributed.
All three under-count in the sponsor's favour, so the number you report is a floor. Say that. "This is what we can prove; the real figure is somewhat higher" is both true and reassuring.
What to avoid
Fingerprinting. Reconstructing an identity from device characteristics is the workaround the industry reached for, and regulators have been clear it is treated as tracking. It also puts you back into consent-banner territory you had escaped.
Believing the platform's number. If a sponsor reports different figures from their own analytics, the discrepancy is usually last-click attribution taking credit for a visit your ad started. Worth explaining calmly, once.