A Simple Question No One In Ad Tech Will Answer

"Data-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 Sarah Sluis, senior editor at AdExchanger. It’s the first in a series of perspectives from AdExchanger's editorial team. 

There’s a question in ad tech that’s been surprisingly difficult to get a straight answer to lately: What happens to programmatic advertising as cookies crumble away?

Many ad tech companies will tell you everything’s fine – they’ve got a solution for that. They use first-party cookies, local storage or a people-based identifier. Or the solution lies in a customer data platform, or an agency’s first-party data platform.

And there are solutions to the crumbling cookie. It’s just hard for a journalist to know which ones are smoke and mirrors, and which ones actually solve problems.

What’s worse, companies that build solutions are often cagey about them. Their explanations can be vague about privacy and fuzzy on the user experience.

Remember fingerprinting? Almost nobody came bragging to AdExchanger about their just-launched fingerprinting-based identity solution.

Likewise, no companies have been eager to share with AdExchanger’s editors how they tack on information to a URL – link decoration – to enable tracking, or redirect users to their own site before a marketer’s so they can set a first-party cookie. Another sleight of hand, moving third-party cookies to a publisher URL to transform them into first-party cookies, similarly doesn’t come up in any of our editorial briefings.

Their reticence, while regrettable, is not terribly surprising. When companies do share how their tech works, they’ve gotten in trouble. When Criteo said on an earnings call that it had outsmarted Safari ITP in 2017, Apple quickly closed that loophole. Criteo later said it wouldn’t share any details on how it circumvents ITP. And as word got out that ad tech companies had started using local storage, Apple cut off local storage with the release of ITP 2.3 in September. Preserving the cookie is a cat-and-mouse game, and no one wants to tip off the cat.

Does it even matter if cookies go away?

It seems indisputable that taking away third-party cookies will ding programmatic revenue in the short term. Consider the following:

  • Publishers tell me they see lower CPMs on Safari traffic, where most cookies don’t persist long enough to help retargeters and other precisely targeted ads.
  • Criteo’s stock has fallen in lockstep with Apple’s Safari restrictions. The company is widely considered a canary in the coal mine for what happens when cookies go away.
  • Google has published research on the impact of removing behavioral advertising from ad impressions, associating it with a 62% decline in news publisher revenue. (Although Google certainly has a vested interest in seeing behavioral advertising thrive, this study aligns with research conducted by Boston University’s Garrett Johnson that saw a 52% decline in exchange price).

There is a (much smaller) body of evidence supporting the opposite conclusion: The end of third-party cookies will cause only a small blip in revenue. A study trumpeted by Digital Content Next claims that targeted advertising adds virtually nothing to a publisher’s revenue. This conclusion rings false to those of us who have been tailed by thousands of just-for-me product ads on the internet (i.e. all of us).

Behind the anti-cookie phalanx is a bet that the loss of third-party cookies will stave off Google’s takeover of the ad business. If third-party cookies die, maybe first-party data will restore publishers to prominence once again.

So which is it? Is doomsday coming, with Google Chrome on the verge of stamping out third-party cookies completely, following in the footsteps of Firefox and Safari, as cookieless proponents want? Or will tech companies keep building workarounds and playing cat-and-mouse until they stumble into that mythical cookieless solution – thereby preserving online advertising and marketing as we know it?

7 Comments

  1. Long term Google and Facebook will profit from this, while publishers will see revenue declines. I don’t see this decline to be massive and immediate, but rather slow and continuous

    Reply
  2. The other identifiers already exist and are already in use to a degree. Ad ID's are becoming freely available, like TTD's public release of the TDID. It comes down to engineering the ability to key on a different identifier and creating the linkages (or licensing the linkages). From an adtech/martech perspective, I'm not sure that the answer is that difficult.

    Reply
  3. THANK YOU, this has been bothering me for MONTHS on the agency side after trying to get answers from close tech partners (DSP and CRM onboarding) and receiving the cagiest of replies, or outright "we'll get back to you on this in a few months".

    Reply
  4. A significant, but not the first major change impacting how advertising dollars are spent. Marketers look for two things, audience reach and high effectiveness. It is rare to get both, which is why campaigns comprise a mix of activities (e.g., TV ad and paid search). Ad spend mix and CPMs will adjust, likely slowly the past two decades of accelerate shift to digital spend but the spend will still follow the eyeballs.

    Reply
  5. Justin Scarborough

    It's no coincidence that programmatic is shifting rapidly into the $70B TV market.

    Reply
  6. Outside of Retargeting with companies like Criteo, 3rd Party Data has always been so horrible from an advertiser ROI perspective anyway, that I don't think it will make much of a difference except for the companies that sell data as a business. First-party data has always been far superior. Most of the data that Criteo uses for targeting is their advertiser's own first-party data anyway. All other data is highly suspect.

    Reply

Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>