Cookieless Targeting Benchmarks 2026: What Losing Cookies Really Costs

Losing third-party cookies cuts publisher ad revenue by roughly 29% to 35%, according to a 2026 field experiment on about 200 million impressions, and Google's Privacy Sandbox recovered only 4.2% of that loss. For cookieless targeting benchmarks in 2026, plan on contextual click-through rates landing 5% to 8% below behavioral, and on ID match rates well under vendor claims.

That is the short answer. The rest of this report shows where each number comes from, how far to trust it, and what we would change in a media plan because of it.

Most cookieless advice is opinion dressed up as strategy. This is the opposite: every figure below traces to a named source, a stated period, and a stated sample. Where sources disagree we give the range. Where a source is a vendor summarizing its own network, we say so.

The headline finding: cookies still carry real money

The strongest evidence of 2026 is academic, not vendor-produced. Researchers at Boston University's Questrom School of Business (Zhengrong Gu, Garrett Johnson and Shunto Kobayashi) ran a randomized experiment with Chrome users assigned to cookies on, cookies off, or the Privacy Sandbox. The work appeared as CESifo Working Paper 11931 and was published in PNAS in May 2026, using ad-management data on about 200 million impressions.

Removing third-party cookies cut publisher ad revenue by 29.1%. For impressions that actually had a cookie available, the drop was 35.3%. Under the Privacy Sandbox, revenue still fell 27.9%, which means the Sandbox recovered about 4.2% of the loss. The Boston University write-up also reports a far steeper hit in the EU, about 66%.

Two cautions. This measures publisher revenue, which is the sell-side mirror of what advertisers pay to reach cookie-addressable audiences. And the experiment treated cookies as fully on or fully off, so it says nothing about partial replacements such as universal IDs. It is the best available estimate of what the old system was worth, not of what any replacement delivers.

NAMEX 2026 Cookieless Targeting Benchmarks

The table below collects every figure we could verify this run, with its source and period. Rows from vendors or secondary write-ups are labelled as such. Rows we could not trace to a named source were dropped.

The DataBeat price row needs context. The publisher network behind it attributes part of the June jump to World Cup and Prime Day demand, and says that attribution is probable, not tested. We include the figure because it shows what buyers are paying now, not because cookie loss caused it.

The NAMEX 2026 Cookie-Loss Planning Range

NAMEX 2026 Cookie-Loss Planning Range: 28% to 35% of addressable ad value at risk without a replacement signal, midpoint 32%. Contextual substitutes cost a further 5% to 8% in CTR.

Three inputs feed it. The lower bound of 28% is the Privacy Sandbox arm of the Boston University experiment (27.9%, rounded), the best outcome available from Google's own replacement. The upper bound of 35% is the loss on cookie-bearing impressions (35.3%) and matches the roughly 35% headline in the university write-up. The 5% to 8% contextual penalty comes from the Integral Ad Science analysis.

The 32% midpoint is the arithmetic centre of 28% to 35%, not a measured value. We publish it because a planner needs one number to model against. Treat it as a stress-test input for your own forecasts, not a prediction for any single campaign.

What the match-rate gap means for your plan

The widest disagreement in the data is also the cheapest to resolve. Survey respondents report cookie matching below 45%. Vendors quote 90% or more. A 45-point gap is not a rounding difference, it is a different product.

The practical rule: never accept a vendor match rate you did not measure. Hand the partner a hashed sample of your own customer file, ask for the matched share, and run the test before budget moves. Our first-party data activation benchmarks show how match rates translate into CPM and ROAS once a list is live.

Identity coverage is also uneven by browser. With Safari and Firefox at roughly 9% of desktop share, Chrome still carries most desktop reach, but mobile skews differently and we have no verified mobile split. Build your test cells by device, not just by audience.

Contextual is the dependable fallback, not the upgrade

The 5% to 8% CTR gap is small enough to be worth paying for when the alternative is an unproven ID graph. It is also one vendor's dataset from 2024. A separate 2021 meta-analysis in the Journal of Advertising, covering 96 studies, found the effect of ad-content congruence on attitudes and purchase intent to be very weak. Read together: contextual rarely wins on engagement, and it rarely loses badly either.

That makes it the right control cell. Run a contextual line alongside every audience line and judge the audience on its lift over contextual, not over nothing. For setup detail, see our practical guide to targeting without cookies.

Want a second opinion on your own cookieless plan? We will compare your current targeting mix against these benchmarks in a free audit, no minimum spend and no contract attached.

Price pressure and where deal structure helps

Programmatic CPMs rose 51% year over year in June 2026 across the DataBeat network, with video up 41.3% and display up 26.9%. The same data shows CTV CPMs falling year over year, at $5.74 in June, which is a reminder that the network figure blends very different inventory.

When open-exchange prices rise and signals thin out, curated and private deals are the usual response. One secondary source reports PMPs averaging 38% higher CPMs with 2.1x the viewability of open exchange, but the original study is not named, so we hold it out of the benchmark table. Our own open exchange vs PMP benchmarks and display CPM by industry give the fuller picture. Our programmatic buying and audience targeting work is priced on a CPM basis, with no retainer.

What we would do with these numbers

First, stress-test every forecast at the 32% midpoint. If a plan only works with cookie-era efficiency, it is a plan with a hidden assumption. Second, measure match rates yourself. Third, keep a contextual control cell in every flight. Fourth, review the whole mix on a services level rather than channel by channel, because signal loss hits search, social and programmatic differently.

Methodology

This report draws on Gu, Johnson and Kobayashi (CESifo Working Paper 11931, PNAS 2026) as reported by Boston University and MediaCat UK; Integral Ad Science's 2024 analysis as summarized by Webtonic in October 2026; the Setupad 2026 identity-signals guide citing StatCounter (May 2026), AdExchanger and Didomi; DataBeat's June 2026 data as reported by PPC Land on 4 August 2026; and the AdAge publisher-adoption report via ATDb. Figures were retrieved on 22 October 2026. Secondary sources are labelled, and several vendor figures could not be independently confirmed.

Cite this data

Cite this data: North American Media Experts, 'Cookieless Targeting Benchmarks 2026: What Losing Cookies Really Costs' (2026). https://www.namediaexperts.com/blog-posts/cookieless-targeting-benchmarks-2026-cookie-loss-match-rates-contextual

Frequently asked questions

How much revenue do advertisers and publishers lose without third-party cookies?

A 2026 field experiment on roughly 200 million impressions found that removing third-party cookies cut publisher ad revenue by 29.1%, and by 35.3% on impressions that would have carried a cookie. Google's Privacy Sandbox recovered about 4.2% of that loss. Treat 29% to 35% as the planning range.

Is contextual targeting as effective as behavioral targeting?

On click-through rate, close. An Integral Ad Science analysis of 1.2 billion impressions found contextual within 5% to 8% of behavioral. That is a measured gap on one vendor dataset, so test it against your own campaigns before committing budget.

What is a realistic ID match rate in 2026?

Industry survey respondents report successful cookie matching below 45%, while vendors often cite 90% or more as a best case. Ask any partner to run a match test on your own customer file and plan on the lower number until they prove otherwise.

Does Safari and Firefox traffic matter for cookieless planning?

Yes, but less than most decks imply on desktop. StatCounter put Safari at 5.3% and Firefox at 3.8% of desktop browser share in May 2026, about 9% combined. Mobile share is higher, and this page does not have a verified mobile figure.

Ready to plan around the real numbers? Book an intro call with Ryan or get a free audit of your current media plan.