With The End Of Browser Cookie Support, What Will Happen To Measurement And Attribution?

Angelina Eng headshotData-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Angelina Eng, vice president of measurement and attribution at IAB and IAB Tech Lab.

In less than two years, many publishers, advertisers, data vendors and technology platforms will need to reevaluate and retool their approaches to data collection, audience targeting, measurement and attribution analysis. All of those activities will change, and along with them the ways that we track and measure advertising and user behavior and how campaigns and websites will be optimized.

The past decade’s vast improvements in reporting and attribution analytics were mostly based on third-party systems and approaches relying on device-based identifiers for addressability, including cookie-based identifiers. But now everything is in flux. It’s unclear whether all browsers will converge on a standard approach and what data would be made available.

So, for the moment, let’s focus only on Google’s latest Privacy Sandbox proposal.

Three important things in the latest Google Chrome Privacy Sandbox

1. Post-impression and post-click attribution may be supported after all. A few weeks ago, it seemed support for attributing conversion data to an impression – whether clicked on or not – would disappear. However, the most recent Google Chrome team has updated its proposal to indicate that there may be some support or solution for post-impression attribution, as well as offering some customizations in attribution modeling. However, the data would be limited and only in aggregate. User or device IDs would not be exposed or captured by any system or platform, including site analytics platforms. Log-level data as we know them today will most likely look very different.

Why it matters: Post-click-only attribution would be bad on the buy side and sell side. Advertisers would compete for clicks, even if clicks are not very important to them, and publisher CPM rates would probably rise. And, with no post-impression ads attributed to a conversion, marketers’ cost per conversions would soar. For publishers, many brands will shift their spend toward other partners that may appear more efficient or effective, and as a result, some good publishers could see a decline in revenue.

Bottom line, post-impression attribution is an important metric for many advertisers and publishers since clicks are not always a good indicator of performance.

Read more: Privacy Sandbox article, “Conversion Measurement with Aggregation Explainer” (April 9, 2020).

2. Aggregated data only.Given privacy and data concerns, the Chrome team is reluctant to give any company the ability to stitch audiences across different browsers and/or devices, fingerprint users based on their device or browser settings, analyze user behavior or create audience segments using specific identifiers or signals. Google “wants to make conversion reports anonymous and unlinkable to individual users or clicks.”

Why it matters: Only aggregate data will be made available, without any user event-level visibility. The browser would control what information can be used and when it is sent. There would be significant changes to the data exposed in “log” or “raw” files. Information such as user-assigned ID, device ID, screen resolutions, IP data, operating systems, browser versions, timestamps, etc. will probably not be available or will be masked.

Read more: Privacy Sandbox article, “Aggregated Reporting API” (March 9, 2020).

3. Scheduled reporting will be controlled by the browser. The users’ browsers will collapse all the data from multiple sites into a single aggregated report at scheduled times. The data delivery mechanism currently in place will no longer be directly delivered in real time to servers or reporting platforms. The data that they can collect will be limited by the users’ browsers and will be delivered in aggregate after an event happens in order to not overtax the browsers from making too many calls.

Why it matters: Many will not be happy with the reporting timeframes and the implications this could have on financial reconciliation for both the buy side and sell side – as well as any intermediaries. Since the impression and click event data would not be supplied in real time, reconciliation could take anywhere from one to 30 days. Or it might neverhappen if a user opens the same browser at a later time.

Marketers would no longer be able to:

    1. Get real-time confirmation that campaigns are live or have the ability to assess whether the right ads are properly served to the right places and point to the right landing pages.
    2. Look at reports accurately until all browsers send through the impression and click data to the respective servers. This could take days or weeks to fully capture. As a result, the delivery performance could appear to be unstable and prevent buyers and bidding platforms from making proper campaign optimization decisions, including avoiding fraudulent inventory.
    3. Reconcile delivery. It would be impossible for stakeholders to agree on the delivery stats, as they may continue to roll in afterwards – or maybe never. In cases where impression data can’t be delivered to the servers because the users have not launched the browser in the scheduled time frame, those impressions might never be reported.

Read more: “Scheduled Reports” section of this Privacy Sandbox article, “Click Through Conversion Measurement Event-Level API” (On GitHub, last accessed May 11, 2020).

Everyone is impacted, and there’s a lot at stake

Many companies in our industry rely on cookies to decide where to place ads and whom to serve ads to, as well as to make business decisions that impact the size of their customer base, revenue and sales.

Brands, publishers, tech providers and data solution providers must come together and collaborate and find new solutions. Everyone needs a seat at the table to define and prioritize critical business use cases, encourage public policy that meets the industry’s needs for privacy-centric addressability and drive standards-based solutions.

Follow Angelina Eng (@angelinaeng), IAB (@iab), IAB Tech Lab (@IABTechLab) and AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. there is a point that if CPMs rise, the value to an advertiser goes down as their budget doesn’t go as far and ultimately questions arise around “is it worth spending here the way we used to”

  2. That’s what will happen… It happens today already for those who are not expertts… They dont understand what Attrbution Model is os what is a statistical model in most part… The explanations will rise, arguments may decrease if we dont get the data needed to justify (“dada-driven”).

  3. Arnaud Blanchard

    Thank you for this! It is a welcome contribution on the impacts of some privacy sandbox proposals on reporting, which are more disruptive than the cohort-based targeting we tend to reduce it to.

    As you said, everything is in flux, the proposals are not precise enough to grasp the actual impact on measurement, and new ideas and other proposals have and could continue to emerge. Indeed, many of the concepts used in the proposals are put in the air but without parametrization, among them:

    – the level of “noise” (differential privacy) that they will put in the reporting and billing (yes, random bills). In average, it’s all the same, but depending of the value they choose, the effect on day-to-day campaign management could be very dramatic.

    – the minimum volume of displays/clicks/conversions necessary to get access to reporting (for example, publishers would not be able to see what ads run for a particular ad group unless there are more than certain amount of displays).

    – the delays to receive for budget consumption/security and ad quality control/machine-learning driven strategy building. One can only guess what is going to happen when publishers or advertisers have to wait for hours or days to change their strategy, or kill a malware-infected creative.

    All of these points are crucial, and definitely need advertisers and publishers attention. Also, they are not set in stone! The discussions are happening right now, and needs contribution from all parties. Advertisers and Publishers should join the W3C Web Advertising Business Group, and make their voices heard.