What PPID is actually worth on the inventory nobody can identify.

Where this number comes from01One sentence, four qualifiers

In November 2021 Google published a figure that has been requoted ever since:

In our initial experiments, we observed that Beta partners experienced an increase of 15% or more in programmatic auction revenue when passing PPIDs in inventory without other identifiers.

Four qualifiers in that sentence do the work, and all four are routinely dropped when the number is repeated. Initial experiments — not a steady state. Beta partners — a selected cohort. 15% or more — a floor for that cohort, not an average. And the one that decides what the number is worth to you: in inventory without other identifiers.

That last phrase means the 15% was measured on the signal-poor slice of inventory. Not on the whole book.

Why it matters again02The deadline vanished. The slice did not.

Identity work fell down publisher roadmaps for a good reason. In April 2025 Google confirmed it would keep third-party cookies in Chrome — no deprecation, and not even the promised choice prompt. In October 2025 it went further and retired most of the Privacy Sandbox, Topics and Protected Audience included, citing "low levels of adoption".

So the cliff everyone was bracing for never arrived, and the replacement was cancelled. It is entirely rational that identity slipped down the list.

But the signal-poor slice was never created by Chrome. Safari and Firefox have blocked third-party cookies by default for years, and between them they are 18.6% of global browsing. That share does not move when Google changes its mind. Add consent rejection, first-party cookie expiry under ITP, ad blockers and opt-outs, and the real figure on your own inventory is higher — by an amount only your logs can tell you.

This is inventory you already sell, every day, at a discount you may never have quantified.

What others measured03Three comparable results, and the useful one

Google publishes three publisher results alongside its own. Grupo Prisa reported more than 15% RPM uplift against no signals. Mediavine reported 7%. News Corp Australia reported 2× eCPM — worth noting separately, because an eCPM lift and a revenue uplift are not the same measure and do not belong on the same axis.

Mediavine's is the least flattering number and by far the most useful one. It is a clean A/B on Safari and Firefox traffic, PPID against no identifier — and it came in at under half Google's floor. It also settles the objection that stops most publishers before they start: their PPIDs were built from first-party cookies, not logins. You do not need a registered audience to do this.

Treat all of them as directional: they are vendor-published customer results, and two of the three are stated as floors rather than averages.

The arithmetic04Where 15% actually lands

Here is the mistake that makes publishers abandon PPID after a quarter. They hear 15%, model 15% of programmatic revenue, deploy to Safari and Firefox, and see something closer to 2%. The conclusion — "it doesn't work" — is wrong. The model was.

Weight by revenue, not impressions. Signal-poor inventory monetises worse, so its share of revenue is smaller than its share of impressions. With s as the impression share of the slice and r as its eCPM relative to identified inventory:

revenue share = (s · r) ÷ (s · r + (1 − s))

On a 19% impression share at half the eCPM, that is 10.5% of revenue sitting in the slice. A 15% uplift on 10.5% of revenue is about 1.6% of total programmatic revenue. Real money at scale, and nothing like the number in the headline.

The recommendation05Do not scope it to the slice

The fix follows directly. PPID is not only a cookie replacement. On inventory that does carry an identifier it still supports frequency capping, audience segmentation and sequential rotation. Scope it to Safari and Firefox and you buy the 1.6%. Pass it on every ad request and you buy that plus whatever those controls are worth across the whole book.

This is what reconciles Google's 15% with our own figure below. They are not the same measurement and were never in conflict: Google measured the slice, we measured the blend.

What we measured06MSN, on every ad request

On MSN we deployed PPID on every ad request, display included, rather than confining it to browsers without cookies. Measured across all inventory, that was worth approximately 10% overall, with Publisher Provided Signals added where applicable worth a further 10% in video inventory performance.

That is our own reported measurement, not an audited one, on a single publisher operating at very large scale. Your number will differ. The method for finding it will not.

Before you commit07What PPID is not

1. 01
   It is an Ad Manager 360 feature

   "Only available in Google Ad Manager 360." If you are not on 360, this whole document is a business case for that conversation rather than a technical one.

2. 02
   It is publisher-scoped and does not travel

   A PPID identifies a user on your properties only. It does not support frequency capping across publishers, and it is not a universal ID competitor.

3. 03
   It is not measurement

   Google states PPID is not supported in reach reporting or conversion tracking. It is a delivery-side control.

4. 04
   Consent still governs it

   Targeting use is disabled when a user opts out via Ads Settings, EU consent controls or RDP, and direct sharing with third-party demand happens outside the EEA, UK and certain US states. The addressable benefit is geographically uneven.

5. 05
   It has to be stable, and hashed

   22–150 characters, meaningless to Google, never raw PII, and consistent across sessions. Changing your generation algorithm later resets frequency capping and segmentation — so decide the algorithm once.

6. 06
   It decays

   Removed after 180 days of inactivity unless resent.

The cliff never came, so the work stopped. The discount on a fifth of your inventory carried on regardless.

Provenance08Sources

1. Google, New ways for publishers to activate first-party data — 3 November 2021. The 15% figure and its scope.
2. Google Ad Manager, Increase your revenue in browsers with limited signals — Grupo Prisa, Mediavine and News Corp Australia results.
3. Google Ad Manager Help, About publisher provided identifiers — requirements, permitted uses and limits.
4. Anthony Chavez, Next steps for Privacy Sandbox and tracking protections in Chrome — 22 April 2025. Third-party cookies retained.
5. Anthony Chavez, Update on plans for Privacy Sandbox technologies — 17 October 2025. Retirement list.
6. StatCounter Global Stats — browser market share worldwide, June 2026.
7. MSN engagement figures are APH's own, reported and not independently audited.