Attribution Reporting API (retired)
Browser measured which ad click led to a purchase. Deprecated Chrome 144; removal under way in 2026.
In one line
Can't measure which ad click led to a purchase in Chrome anymore
Browser measured which ad click led to a purchase. Deprecated Chrome 144; removal under way in 2026.
Matched ad clicks or views to later conversions with noised event reports (3 bits click data for views) or encrypted aggregate reports. Google retired it in Oct 2025 and plans a W3C PATWG attribution standard instead.
What it can and can't do
What the user does: Nothing. It used to happen after an ad click and a later purchase
It can
- Have linked an ad click to a later purchase in Chrome, without a cookie
- Have sent a report with limited detail after a delay
It cannot
- Run in Chrome 144 or later, where it is deprecated and being removed in 2026
- Run in Edge or Samsung Internet
- Name you in the report; it was built to be limited
- Link a click to a purchase on a device without the feature
How it differs between platforms: Chrome is removing it. Safari and Firefox have their own partial versions.
Browser support
| Browser | Status | First version | Released | Note |
|---|---|---|---|---|
| Chrome | Removed | 153 | 2026-09-08 | Deprecated 144, removal per 153 notes |
| Edge | Not supported | Not available | Not listed | None |
| Firefox | Partial | 1 | 2004-11-09 | Firefox shipped a different PPA attribution experiment |
| Safari | Partial | 15.4 | 2022-03-14 | Private Click Measurement is Safari's separate API |
| Safari on iPhone | Partial | 15.4 | 2022-03-14 | Private Click Measurement |
| Chrome on Android | Removed | 153 | 2026-09-08 | None |
| Samsung Internet | Not supported | Not available | Not listed | None |
How a site asks for it
| API | Attribution-Reporting-Eligible / Attribution-Reporting-Register-Source headers; <a attributionsrc> |
| Permissions policy | attribution-reporting |
| Browser compat keys | api.HTMLAnchorElement.attributionSrc |
| How access is granted | none |
| What the user sees | None |
| Needs a secure context (HTTPS) | No |
| Needs a user gesture | No |
| How long it lasts | Source lifetimes up to 30 days; reports delayed 2 days to 30 days |
| How to take it back | Clear site data, browser privacy settings |
| Inside an iframe | Not listed |
Data you get
What comes back: Noised conversion report
Shape: event
How often: delayed by hours to days
| Field | Type | Unit | Range | Precision |
|---|---|---|---|---|
trigger data | integer | bits | 1 to 3 bits | noised |
Technical limits
- Event reports carry 1 bit (views) or 3 bits (clicks) of conversion data
- Reports delayed by hours to days
- Chrome and Android only
- Cannot link to a cross-site identity
The story
What it is
A Chrome tool that measured which ad led to a sale, with some noise added. Being removed.
For you
You never saw it work. When you clicked an ad and bought, Chrome could send a fuzzy report instead of a name. It is now being removed, so most ad measurement goes back to cookies and server events.
For makers
It was Google's answer to counting sales without following people. Fuzzy numbers kept identity out. Advertisers disliked the noise and the work, and Google retired it in 2025 and plans a new standard through a web group.
Where you see it
| Type | Example | Apps |
|---|---|---|
| ads | An ad network counted sales from a banner on a news site, with random noise, so no single buyer was named. | Criteo, Google Ads |
| analytics | A shop got a daily total of clicks that led to purchases for each ad, instead of a person by person list. | None named |
| shopping | A fashion brand learned which ad format drove sales, with privacy noise added to small groups. | None named |
| games | A mobile game maker's web ads could report installs and sign ups tied to a campaign, without a device ID. | None named |
| media | A publisher showed advertisers proof that an ad on its site led to a sale, in aggregate only. | Washington Post |
| security | Fraud teams worried that noisy reports hid fake clicks, one reason many advertisers did not adopt it. | None named |
Risks
- Was never widely adopted
- Being removed, so measurement moves elsewhere
What the platform says
- Google Chrome: Attribution Reporting. Chrome pitched it as ad measurement without cross site IDs, with noisy event reports and encrypted summaries.
Good uses and abuse
Legitimate uses
- Privacy preserving ad measurement (historical)
Known abuse
- Cross-site conversion linkage
Scores
| Score | Value | Meaning |
|---|---|---|
| Sensitivity | 3 of 5 | How revealing the data is |
| Utility | 2 of 5 | How useful it is to honest apps |
| Friction | 3 of 5 | How much the user has to do |
| Fingerprint bits | Not rated | Rough identifying entropy, when known |
History
- 2023-07, Chrome 115: Shipped (Chrome)
- 2025-10, Chrome 141: Retirement announced
- 2026-01, Chrome 144: Deprecation per blink-dev
- 2026-08, Chrome 152: Stub in 152; removal per Chrome 153 release notes; no blink-dev update since 2026-07-02 as of Oct 2026 review
Related capabilities
Related
- Private Aggregation API (removed). Can't send noisy ad totals from Chrome anymore, since it was removed in August 2026
- Ad click IDs (gclid, fbclid, msclkid and more). Can tag your ad click so a later purchase links back to that exact ad
Sources
- Privacy Sandbox: Update on plans (Oct 17, 2025)
- Privacy Sandbox feature status page (updated Aug 14, 2026)
- Privacy Sandbox history and removal timeline (checked Oct 2, 2026)
- Attribution Reporting spec
How well checked: Checked against real behavior.