Deprecating IDs cuts publisher revenue 29.1%

Publisher ad tech is having one of those awkward moments where the academic evidence finally catches up to the industry argument. For years, publishers said killing cross-site IDs would hurt the open web’s business model. Privacy advocates said new privacy-preserving tools could replace the old tracking stack. Now there’s a real field experiment on the table, and the answer is pretty blunt: remove third-party cookies, and publisher revenue drops hard. Privacy Sandbox helps a little. Not much. (ifo.de) ### What exactly is new here? The new thing is a June 2025 working paper from Zhengrong Gu, Garrett A. Johnson, and Shunto J. Kobayashi at Boston University. They worked with Raptive, an ad management firm, and measured more than 200 million ad impressions across 5,000-plus publishers in an open industry experiment. That scale matters — this is not a lab toy or a single-site case study. (ifo. ([ifo.de)What are “IDs” in this story? Basically, this is about third-party cookies and the cross-site identifiers built on top of them. Those IDs let ad buyers recognize a browser across different websites, which helps with targeting, frequency control, attribution, and bidding. If you remove that recognition layer, buyers know less about who they are reaching and what happened after an ad ran — so they bid less aggressively. (ifo.de) ### How big was the revenue hit? Big enough that publishers were not exaggerating. The paper says removing third-party cookies reduced publisher revenue by 29.1%. Then it tested Google’s proposed replacement stack — Privacy Sandbox — and found it preserved just 4.2% of the lost revenue. Put differently, the replacement recovered only a sliver of the money that disappeared when IDs went away. (ifo.de) ### Why doesn’t Privacy Sandbox make up the gap? Because the old cookie system did several jobs at once, and Sandbox tools only replace parts of that stack in a more constrained way. The paper also found higher ad latency and a 2.9% reduction in impression delivery. That sounds technical, but it matters. Digital ads are an auction business. If pages load slower or fewer impressions make it cle(ifo.de) about targeting quality. (ifo.de) ### Why should anyone outside ad tech care? Because this is really about who pays for the open web. A lot of journalism, forums, recipes, niche blogs, and creator-run sites are funded by advertising rather than subscriptions. If ad revenue falls by something like 29%, the missing money does not vanish harmlessly. It usually turns into more paywalls, more affiliate commerce, more sponsored cont(ifo.de)ivacy versus funding — and this paper makes the funding side feel a lot less abstract. (ifo.de) ### Didn’t Google already back off cookie deprecation? Yes — and that is part of why this paper lands with extra force. Google had planned to phase out third-party cookies in Chrome, but by 2025 the company had restated that it would not deprecate them, and the UK CMA moved to unwind the commitments built around that plan. So this is not just a forecast of a future Chrome change anymore. It is(ifo.de)much resistance from publishers and parts of the ad industry. (gov.uk) ### Does this settle the privacy debate? Not really. The paper does not argue that pervasive tracking is good. It argues that the replacement tools, at least in this test, do not come close to replacing the economics of cookies for publishers. That leaves the industry in the uncomfortable middle — everyone wants less creepy tracking, but nobody has shown a privacy-safe alternative that keeps the open web equally funded. (ifo.de) ### Bottom line? The cleanest read is this: the revenue function of cross-site IDs was larger than many privacy-first proposals assumed. That does not mean the old system should last forever. But it does mean “just swap in PETs” was never a serious economic answer for publishers. (ifo.de)