One of the most persistent objections to privacy-first advertising is the measurement problem. Advertisers, understandably, want to know whether their campaigns are working. Publishers want to demonstrate the value of their inventory. And for most of the past two decades, answering both questions depended on the same infrastructure that privacy regulation is now constraining: persistent user identifiers, cross-site tracking, and conversion pixels that followed users from ad click to purchase.
The argument runs: without cookies, you can’t attribute conversions to campaigns, which means you can’t measure ROI, which means advertisers can’t justify spend, which means publisher revenue falls. This argument is treated by some as a fatal objection to privacy-compliant advertising. It isn’t - but addressing it properly requires understanding what attribution actually needs to do and what technical approaches are available.
What attribution is trying to measure
Attribution is the process of understanding which advertising touchpoints contributed to a measurable outcome - typically a purchase, a lead, a sign-up, or some other conversion event. It answers the question: “If a user saw our ad and then bought something, how much of that purchase decision was influenced by the ad?”
This is actually a difficult statistical question even with full tracking data. The traditional approach - last-click attribution, which gives all credit to the final touchpoint before conversion - is widely acknowledged to be a poor model of how advertising actually works. Multi-touch attribution models are better but require even more data and more complex analysis.
The honest answer is that even in a world of unlimited tracking data, the advertising industry’s attribution capabilities were substantially overstated. What looked like precise measurement was often correlation masquerading as causation, with attribution systems taking credit for purchases that would have happened anyway.
What changes without cookies
Without third-party cookies, cross-site attribution - tracking a specific user from an ad exposure on one site through to a conversion on an advertiser’s own site - becomes much harder. This is a genuine loss of information, but it is a loss that needs to be evaluated realistically rather than catastrophised.
The inventory categories where cross-site attribution matters most are direct-response campaigns with short conversion windows: e-commerce ads where the goal is an immediate purchase, or lead generation campaigns where a form completion happens within hours of an ad view. For these campaign types, the deprecation of third-party cookies does create measurement challenges.
For brand advertising, awareness campaigns, and longer-cycle purchase decisions - which account for a substantial proportion of advertising spend - cross-site attribution was always a poor fit. A car manufacturer’s advertising doesn’t need to attribute individual sales to specific ad exposures; it needs to understand whether brand perception is improving in its target audience. A financial services company running awareness campaigns knows that the purchase cycle is months or years, not hours.
Privacy-preserving attribution approaches
Several approaches are available for measuring ad performance without cross-site tracking.
Aggregated measurement: Rather than tracking individual users, aggregated measurement looks at population-level correlations between advertising activity and conversion rates. When campaign spend is increased, do conversion rates in the target audience rise? When specific placements are active, is there measurable uplift? This approach requires larger sample sizes but is legally clean and surprisingly powerful when implemented correctly.
Privacy sandbox measurement APIs: Google’s Attribution Reporting API, part of the Privacy Sandbox initiative, provides attribution data at an aggregated level with differential privacy noise added to prevent individual identification. It allows advertisers to understand which campaigns and creatives are driving conversions without any individual-level tracking.
Server-side conversion tracking: Advertisers can implement server-side conversion tracking - sending conversion events from their own server rather than through a browser-based pixel - which avoids the cross-domain data sharing issues of traditional pixels while still providing accurate conversion data. When combined with privacy-preserving matching through hashed email addresses or similar first-party identifiers, this can provide strong attribution for users who have explicitly shared their contact information.
Modelled conversions: Both Google and Meta now use machine learning to model conversions that cannot be directly observed due to consent choices or browser restrictions. These modelled figures are acknowledged estimates rather than hard numbers, but they provide advertisers with a directionally accurate view of campaign performance.
Media mix modelling: The oldest form of advertising measurement - understanding the contribution of different media channels to business outcomes through statistical analysis of historical data - is having a significant renaissance as real-time click-level attribution becomes less available. MMM doesn’t require any user-level data at all; it works from aggregate sales, revenue, and media spend data.
What this means for publishers
For publishers, the cookieless attribution environment creates both challenges and opportunities. The challenge is that some direct-response advertisers, accustomed to the (often illusory) precision of cookie-based attribution, are reluctant to spend with publishers who cannot offer comparable measurement. This is a commercial headwind.
The opportunity is that publishers who invest in helping advertisers understand the value of their audience through privacy-preserving measurement - providing reach and frequency data, engagement metrics, and whatever consented first-party signals are available - can differentiate themselves from publishers who simply accept that measurement is broken and offer advertisers nothing.
The publishers who will win in the post-attribution world are those who can make a compelling, evidence-based case for the quality and relevance of their audience, even without the crutch of individual-level conversion tracking. This requires better audience understanding, better reporting capabilities, and better advertiser relationships than the old programmatic model demanded. It is more work - but it produces more durable, more valuable advertising partnerships.