The web moves from cookies to browser-tracking

Google's FLoC, Apple's Privacy Update, and the Page Experience Update all have one thing in common: real-world user tracking. That shows a broader trend I highlight in this article.

The web moves from cookies to browser-tracking

What do Google’s Page Experience update, FLoC, and Apple’s privacy update have in common? They all measure in the browser. But, to really understand the answer, we need to take a step back.

Disguised as an argument for more privacy, big tech advertising platforms move tracking from cookies to browsers. Third-party cookies are used to track users from site to site and show them relevant ads. However, users often don’t know this and behave differently when asked for consent (let’s ignore how big of a problem cookie-tracking really is, for now).

When Apple rolled iOS 14.5 out and asked users for consent to be tracked by 3rd parties, only 15% opted in worldwide and 6% in the US, according to Flurry Analytics. That is a big deal for advertisers.

But not only Apple positioned itself in a privacy-first world.


Google’s answer to the call is FLoC, Federated Learning of Cohorts. It groups large numbers of users (“thousands”) with similar behavior into profiles by tracking their browser activity. With FLoC, Chrome tracks visited sites, content on these sites, and “other factors” to create anonymized cohorts.

Google claims to achieve at least 95% of the performance of third-party cookies. To be clear, cohorts are not perfectly anonymized because sites can join PII with the cohort ID.

Apple’s privacy update

With iOS 14.5, Apple started asking users for consent to be tracked, also known as App Tracking Transparency. Before, app developers could track users, store that information, and even sell it with IDFA (Identifier for Advertisers).

Apple provides new solutions, SKADnetwork and Private Click Measurement, as workarounds. Make no mistake, though, the billion-dollar fruit will still track users and display ads in Apple News, App Store, Stocks, and now Apple Search - just on Apple’s terms. As outlined in the Advertising & Privacy terms, Apple won't share data with 3rd parties but collect data from your device, location, App Store, News, and Stocks.

What Apple defines as “not tracking you” actually means “3rd parties not tracking you.” If Apple records your behavior and uses that information to show you (better) ads, it’s okay.

In fact, Apple follows the same approach as Google with FLoC! The Advertising & Privacy terms go on to say (highlights mine):

We create segments, which are groups of people who share similar characteristics, and use these groups for delivering targeted ads. Information about you may be used to determine which segments you’re assigned to, and thus, which ads you receive. To protect your privacy, targeted ads are delivered only if more than 5,000 people meet the targeting criteria.

Here’s the excerpt from the FLoC GitHub repo for comparison:

“We plan to explore ways in which a browser can group together people with similar browsing habits, so that ad tech companies can observe the habits of large groups instead of the activity of individuals...A FLoC cohort is a short name that is shared by a large number (thousands) of people, derived by the browser from its user’s browsing history. The browser updates the cohort over time as its user traverses the web.”

Same approach, different name.

Core Web Vitals

The third instance of in-browser tracking is Core Web Vitals (CWV), a trifecta of metrics that quantify user experience. Google has been using certain Page Experience metrics like SSL or mobile-friendliness in ranking for a while, but CWV are measured differently. Google uses CRuX (Chrome User Experience) reports that collect real-world (field) data:

The Chrome User Experience Report is powered by real user measurement of key user experience metrics across the public web, aggregated from users who have opted-in to syncing their browsing history, have not set up a Sync passphrase, and have usage statistic reporting enabled.

For the first time, Google uses field data from real users instead of bot-derived lab data from extrapolations. “Speed” has always been hard to measure because there is so much variety between devices and internet connections. It was easier to abstract how fast a site loads for users, but never better. CRuX is a deviation from that approach because real data comes with all the context “speed” happens in. It was just harder to measure to this point, but it seems Google figured it out.

2nd and 3rd-order consequences

To summarize, Google and Apple - the two biggest (only?) players on the smart device operating system market - replace 3rd-party tracking with inhouse solutions that group users by behavior and provide cohort-based data to advertisers. Instead of websites tracking users with cookies, Google and Apple record and aggregate user behavior with the help of machine learning. This shift fits into the privacy-conscious zeitgeist but is also an indicator for the technology that seems to be ripe for broad tracking and clustering.

As a consequence, big advertising platforms gain more control and power, advertisers become more dependent on G and A. Ironically, we see the same trend in Google Ads, where Google took away targeting capability in February 2021 by incorporating broad match modifiers into phrase match.

In July 2021, both phrase and broad match modifier keywords will have the same updated phrase matching behavior for all languages, and will show ads on searches that include the meaning of your keyword.”

In other words, Google will make keyword targeting choices for you - powered by machine learning. Already in 2018, Google started to include “close variants” in exact match campaigns. At the same time, advertisers see almost 30% fewer keywords in Google Ads. That’s very similar to Apple changing from IDFA to SKADnetwork and controlling a bigger piece of the advertising journey.

So, what do Google’s Page Experience update, FLoC, and Apple’s privacy update have in common? They’re all tracked in the browser and cohort-based.