Who and what will survive the coming Cookiepocalypse?

The Cookiepocalypse is a term we’ve been using around KMTX Towers for some time now to our cookieless future. We can’t claim it, and attribute to our friend (and Adtech influencer, Beeswax CEO etc.) Ari Paparo. It’s an appropriate term in our view because, not only is it painfully accurate, it’s also the right level of tongue in cheek.

Do any of us really think that our ability to target ads is going to disappear? Or somehow become less accurate? Will we be flung back to the dark ages of newspaper advertising with little-to-no attribution? Do we really believe the successive moves by Google and Apple are actually in the interests of our personal privacy? 

All that being said, there is still a looming horizon and a lot of uncertainty. Uncertainty that defies all logic considering the size and value of the market which will be affected. Therefore we’ve taken the opportunity to look at the reality of our collective, cookieless futures and try to understand the likely outcomes. 

Here then, is our story of the pending Cookiepocalypse.

The good, bad and ugly of the Cookiepocalypse.

As in any good story, it makes sense to set the scene by outlining the roles of the various protagonists. In this case, the good, the bad… and the ugly.

The Good – 1st party cookies

Our good guys are held by the sites you visit. In most cases in ad-land these are publishers: media companies which vary in size, sophistication and revenue. They keep your log-in details, track your behavior for analytics purposes (on that site only) and generally optimize your experience.

The Bad – 3rd party cookies

These are considered bad guys because – just like a bad guy – they’re always lurking around the next corner, watching your every move. They’re the reason those shoes you like – which you already bought and were delivered two weeks ago – still appear in home page takeovers while you’re reading about pot plants. In Adtech parlance, they also look after frequency caps or the number of times you see an ad, help control the price of an ad (bid) and how an advertiser such as tat  shoe brand knows that you’ve purchased the item because of that ad (attribution).

The Ugly – Chrome

Here’s where our story starts to get messy. Chrome is the only one of the three dominant browsers to still allow 3rd party cookies. Safari and Firefox, numbers two and three in browser market share, stopped their usage a while back. The problem here lies in three parts. First, Google owns Chrome. Second, they own large chunks of the ad network and Third and finally, they own Google Analytics: upon which most 1st party cookies do their best work. They’re now going to remove 3rd party cookies from Chrome – the action which brings the Cookiepocalypse – but have yet to work out how to replace some of the work these cookies do.

Google’s proposed solution – the mysterious stranger in our story – is the Privacy Sandbox.

The Privacy Sandbox

Google wants to keep the ad market alive. Google NEEDS to keep the ad market alive. It is, after all, how they make more than 80% of their revenue. For this reason, it is in their own interest to maintain the market. Therefore they’ve started exploring a bunch of options including targeting alternatives and privacy mitigation tools. 

Targeting options include:

  • Contextual Targeting – which is essentially just placing an ad next to relevant content
  • FLoC (Federated Learning of Cohorts) – where users are placed into interest clusters
  • Turtledove – a ‘family’ of options of which Fledge is the first ‘locally-executed decision over groups’ experiment

While the privacy tools include:

  • Privacy Budget – a way to combat user ‘fingerprinting’ by changing browser behavior
  • Gnatcatcher – or Global Network Address Translation Combined with Audited and Trusted CDN
  • Trust Token – an API for bot detection.

We encourage you to follow these links and read more about each of these options. 

We’re going to delve a little deeper into each of the targeting options below, how they affect us as adtech business (and likely many others in adtech and programmatic) and whether they’re options that we as an industry are likely to adopt.


As described above… which is about the extent of it and therein lies the limitation. While many may claim that contextual is the ultimate, and best solution, it has a multitude of limitations which we’ve discussed before. 

These include volume, keyword ambiguity, cannibalization and fraud. Most especially, there is something very 90’s about contextual targeting. In the 21st century, we’d hope to find a more elegant solution.


Sounding like the bad guys in a dystopian sci-fi novel or a teen-movie joke*, Federated Learning of Cohorts is one of the more ambitious solutions being proposed to stop the Cookiepocalypse. Essentially, users with similar interests are clustered into groups of unspecified size and then targeted with ads based on the similarity of their consumption. 

These clusters will be based on the sites they visit and the analysis of who fits in each cohort will be decentralized, meaning the browsers (and their owners) don’t know anything about the user. This, coupled with non-descriptive IDs, is meant to protect user privacy. But there are issues. For example FLoC is currently in beta testing… testing which it isn’t running in Europe. Yet. 

There are also some concerns that users will be clumped together to reveal ‘sensitive’ categories. But for the moment, FLoC seems to be Google’s best bet. 

For adtech businesses, FLoC allows us to build semantic profiles, and augment the data available for even better targeting. 

*Yes we know they’re all bird-related. Which is weird in itself.

Turtledove and Fledge

As mentioned earlier, Turtledove is a ‘family’ of options and Fledge is the first experiment in this family. These are based around interest groups held by advertisers, publishers and adtech companies. 

These interest groups are stored at the browser end, and bidding occurs on-device. To aid in privacy, ads are rendered in a ‘fenced frame’ so that no interactions can be shared with the browser, publisher or advertiser. 

Such solutions require a lot of work due to the fact that the bidding occurs on-device. To have access to standard metrics such as auctions won, how many clicks were generated and how much budget is left, the device must send some information. To maintain the anonymity, this information must be shared to a trusted server, which can then aggregate before sending the data.

For adtech companies, all of this means that cohorts will be targeted instead of specific users, and those cohorts will be based around a variety of interest groups: at its most basic, an interest group could be everyone who placed an item in their cart on an ecommerce site.

How does the story end?

While there is potential that each of these solutions are widely used in future, this will also depend on other integrative solutions to cover some of the ‘legwork’ which cookies already do. This includes:

  • Frequency Capping
  • Bidding
  • Attribution 

There are a plethora of mooted solutions which we’ll talk about in future posts.

But like in any story, you still need a conclusion, a climax… a denouement. The fact is, we’re nowhere near the end of this story and the characters are sure to change a lot before we get there.

In any case, for us like many others in the ad ecosystem, we’re currently in a holding pattern. If we do nothing from here on out, our own testing on available solutions shows there are some wins and some losses along the way. CPC’s may even drop, but using video as an example CPVs look like increasing by multiples. Frequency capping will likely be handled by the DMPs, while attribution is probably going to be in the hands of ad servers (both of which can be provided by Google. Surprise!) 

However it ends, you can be sure we’ll be doing our best to deliver optimum performance. In the meantime, feel free to grab some popcorn and remain on the edge of your seat.

Special thanks to Pierre Bros for his words and info.