#20: Who Stole the Cookie?
Welcome to Quo Vadis, your periodic source for fresh programmatic news and off-the-beaten-path perspective. Click here to join the conversation.
As you enjoy this article, you can sing along with this classic childhood song.
Who stole the cookie from the cookie jar?
Are you surprised? Don’t be. This is adtech.
Last week, Ebiquity and Usercentrics published the results from a unique EU cookie consent test. If you haven’t been kicking it on the beaches of advertising with your head stuck in the sand, then you will probably not be very surprised by the results they found.
When Ebiquity (media investment analysis company) and Usercentrics (consent management platform) scanned the 1,000 biggest publishers in Europe (from a European IP address), they found a large number of non-compliant cookies. No surprise.
Quo Vadis loves these data-driven Moneyball tests that separate facts from adtech fiction. After all, fictional information and false heuristics have a funny way of influencing how marketers make decisions about where to place their media budget bets.
The results of this data collection exercise prove that Nobel Prize winner Amos Tversky was right when he said:
“A part of good science is to see what everyone else can see but think what no one else has ever said.”
Good Holiday Read: If you want to learn more about Tversky and the interplay between human judgment and probabilities, then check out Michael Lewis’s book The Undoing Project. If you’ve read Lewis’s Moneyball, his book on the life of Tversky (and Daniel Kahneman too) peels back layers into how we humans get probabilities wrong most of the time. The probability-based nature of programmatic is filled with case studies.
Ebiquity + Usercentrics Findings
The analysis found over 50% of third-party cookies come from 2074 adtech companies. Are you surprised yet?
You shouldn't be. Finding adtech cookies across the ad-supported internet is like looking for plastic trash in the Pacific Ocean. It’s everywhere.
The data revealed that one-third of these marketing cookies are fired on publisher sites before users give consent.
What do you, me, and 99.99999999% of other people do? You can’t click “accept” fast enough. Whatever you are there for, you want it now. And as for all those personalized ads, you can hardly wait for that too… at least according to the people that sell them.
From the ad-supported publisher’s perspective, they are often desperate to maximize revenue (instead of maximizing profits), so getting your consent appears to be an afterthought. And for adtech players, asking for forgiveness (on the slim chance anyone ever gets caught) trumps asking for permission.
Why? Mainly because adtech companies have a higher priority economic incentive. They need to constantly expand impression volume in any way possible to reduce marginal cost and get more price-competitive with Google’s programmaticplex.
The study also found that 70% of these cookie-based adtech companies transferred data outside of the EU. That's a big GDPR no-no. But then again, if you’re surprised by that then imagine where else you might be making bad programmatic bets.
Stretch Your Mind, Dig Deeper
Programmatic audience targeting is like fish and chips. Too much batter on your fish ruins the dish into a soggy mess.
It appears that thousands of adtech companies and publishers across the programmatic landscape think of GDPR as a suggestion rather than law. Why is that?
For the obvious answer, look no further than money! If you’re looking for the not-so-observable reason, that brings us back to all that plastic trash floating around the Pacific Ocean.
When marketers get overly influenced by “easy button” adtech sales pitches, their insatiable appetite for audience targeting grows like a tide that only flows. It never ebbs.
Display ad spending growth is one of the big driving forces behind GroupM’s updated forecast estimating that global media spending will top $1 trillion by 2025. For 2021, GroupM is forecasting $763 billion in global media spending, up 22% from 2020.
At its core, audience targeting is a probability game. Making it work requires throwing a massive net into the ad-supported internet ocean to catch user IDs swimming in the bid stream between SSP bid requests and DSP bid responses.
Think about it this way
Imagine you want to catch 100 specific fish swimming around in the Pacific Ocean. If you have a small fishing boat with a small trawling net, you’re unlikely to catch any fish at all. You’ll end up like the Old Man and the Sea — completely salao.
Salao is Spanish slang. It means salty, which is the worst form of bad luck.
To make it work, media buyers need a huge net like the size of Australia. With a giant net, they can swoop up ad impressions from hundreds of thousands of websites sent by every SSP possible.
Rule of Thumb: Marketers buy ad impressions. Adtech buys users that make ad impressions.
Your chances of catching enough cookie fish are now greatly improved. But not without an expensive trade-off. Using such an inconceivably large net means marketers also catch a lot of trash in the process. Whether or not any advertising gets done is a different question.
That’s audience targeting. Like everything else, it comes with trade-offs. If the trade-off costs are not incorporated into ROI, then what is reported as a positive advertising result could very well be a negative one.
The entire concept of audience targeting to specific user IDs in the bid stream is predicated on the assumption that buyers are willing to cast a very wide net to "find" user IDs in the bid stream. This is known as the "Find-Back" Ratio.
Example: You create an audience segment of 100 user IDs based on cookies from those 2074 adtech companies. They all sync data with each other. Even if match rates are terrible, everyone in adtechland will say everything magically matches up 1-to-1. Don’t worry. It’s good enough. Trust us, we got your back.
Over the course of a 30-day campaign, your DSP of choice "sees" 20 out of 100 cookie IDs appear across millions of bid requests sent by SSPs. This puts your Find-Back ratio at 20%.
The higher your Find-Back Ratio, the more budget will flow. This is what media buyers refer to as "scaling" a campaign budget. If there are no cookie find-backs, there is no spend ☹️.
The adtech crew also loves telling marketers they can frequency cap. They did with 3rd party cookies and they are doing it again with cookie replacements. Frequency capping is a sales pitch soft spot for marketers.
The adtech apparatus doesn’t have to actually be able to pull it off. They just have to say they can and it becomes true. If adtech was a commercial fishing outfit, they’d tell you: “Don’t worry, we only catch a certain kind and a certain amount of fish for you. We never overfish. That would be bad for the environment.
Since no one can ever really check if frequency capping is really happening, it’s easier to assume it is and the money keeps flowing like water to gravity.
Information Bias: Also called observation bias or measurement bias, happens when key information is either measured, collected, or interpreted inaccurately.
If media buyers have to constantly cast and recast a massive net through hundreds of SSPs and tens of thousands to millions of websites to get enough scale, they also end up buying tons of low-quality inventory in the process.
The deeper buyers go to find fish in the long tail of publisher inventory, the more trash they end up catching. But as long as adtech can deal with the trash so marketers never see it or smell it, the money (and the pay-to-play fees) will keep flowing.
And it’s not just the small little-known websites. Big brand name sites can sell just as much bad inventory as anyone else. But like any other market, there is a market for trash too. As long as marketers demand it, the ad-supported internet will find a way to supply it.
So, if we ask why 30% of publishers skirt GDPR consent rules and 70% of 3PMC adtech players ignore GDPR, it’s because the benefits far outweigh the risk.
Sure, third-party cookie fish will eventually go extinct as Apple and Google put increasingly restrictive policies in place. And with every dying cookie, audience targeting as we know it will become increasingly difficult to pull off (until a replacement comes along).
In the meantime, why marketers spend $1.00 CPMs on audience data (roughly 5–20% or more of ad budgets) and also have to pay verification vendor fees to count all the trash caught by casting such a wide net might be a perplexing question to ask.
But then again, maybe the answer is really simple. Just like water, the money has to go somewhere. That somewhere in the media biz is usually what leads marketers to the least-worst alternative, aka the choice of least regret.
How Much Carbon Does an Ad Impression Emit?
Mikko Kotila, CTO at Cavai, wrote a fantastic piece asking the question:
“How much carbon dioxide does an ad impression emit into the atmosphere?”
As Mikko points out, for every thousand ad impressions delivered, between 140 Wh and 1930 Wh of energy is consumed, resulting in 0.08 kg to 1.09 kg of carbon emissions before offsetting.
Quo Vadis Math Time:
Last May, Quo Vadis estimated the size of open-web programmatic advertising to be around $53 billion in 2020. Given GroupM’s recent ad market update, it’s probably closer to $70 billion in 2021.
Assume a $5 average fully-loaded CPM (aka “Advertising Cost” in DSP reporting platforms). DR marketers tend to pay below-average CPMs, while brand marketers pay above-average CPMs, give or take.
Let’s take the mid-point in Mikko’s 0.08 kg to 1.09 kg range, which is 0.6 kg per 1,000 impressions.
Given the plug values, marketers bought 14 trillion impressions in 2020, generating 8.4 billion kilograms or 8.4 million metric tons of carbon emissions.
According to the EPA, the average car produces 4.6 metric tons of carbon per year.
So, if marketers somehow realized that audience targeting is not only a regretful strategy to get advertising results when it is overused, but also generates needless carbon waste, then they’d save money and the planet at the same time.
If marketers stopped doing programmatic audience targeting today, it would be like taking all the cars off the roads in New York City every Monday, Tuesday, and Wednesday forever.
Ask Us Anything (About Programmatic)
If you are confused about something, a bunch of other folks are probably confused about the same exact thing. So here’s a no-judgment way to learn more about the programmatic ad world. Ask us anything about the wide world of programmatic, and we’ll select a few questions to answer in our next newsletter.
Join Our Growing Quo Vadis Community
Was this email forwarded to you? Sign up for our monthly newsletter here.