Way way way back in 2021, I wrote an article called The Great Cookie Kerfuffle (https://trustwebtimes.com/the-great-cookie-kerfuffle/) where I unpacked Google’s grand cookie killing announcement. It quickly became clear to me and everyone else that this was meant to be an optic PR ploy because, in fact quietly, Google was trying to hang onto cookie tracking through every technical privacy loophole it could find.
More than that, while everyone understood why Google made the announcement, it was also clear that, despite the talk, nobody’s heart was really in a user privacy-first adtech ecosystem. Not Google’s or any of the adtech firms in the ecosystem because as I observed at the time; “Beneath the mundane chatter about the cookie crumbling, you will detect something quite extraordinary; the existential terror that lies at the heart of the Great Cookie Kerfuffle,” (https://trustwebtimes.com/the-great-cookie-kerfuffle/).
Plainly, the killing of the cookie was a profit killer too because adtech made big bucks in tracking – data folks, media buying people and retargeting platforms were just some of the adtech players who all were reaping great profits from collecting and tracking all this personal data. In fact; “Adtech firms were ill-prepared for such a radical change because they were perfectly happy participating in the greatest data heist of all time. This gravy train was too good to worry about some niggling privacy issues that may never really materialize.” https://trustwebtimes.com/the-great-cookie-kerfuffle/
The inevitable adtech “pivot” to address this cookie apocalypse became concocting every mechanism that allowed adtech to continue tracking folks without appearing as though they are tracking people.
Everyone was in on the game – from Google to even the smallest adtech startups. In 2022 and 2023, we saw ridiculous “privacy solutions” that quickly fizzled out. Google’s ill-conceived FLoC proposal (Federated Learning of Cohorts) came and went with a painful whimper. Then, we heard about privacy sandboxes and data cleans rooms from vendors who gushed about its privacy-first data architecture. Except they weren’t. Clean rooms, for example, was all the rage for a bit even though; “There is nothing clean about data clean rooms… Yet, too many adtech firms are working hard to convince – er everyone – that data clean rooms are pro-digital privacy. ” (https://trustwebtimes.com/there-is-nothing-clean-about-data-clean-rooms/).
All these failed experiments in the tracking/ not tracking tech stacks underscored the PR game adtech and Goole were trying to perpetuate on politicians, advertisers and even consumers. As soon as anyone looked under the hood for any of these non-tracking solutions, the truth became abundantly clear – it required people to jump through a major trust hoop to understand how a profile is targetable anywhere online without crossing the intent of protecting digital privacy.
Too often, these solutions were simply incomprehensible despite all the slick math lingo in the sales pitch.
Worse, far worse, turns out all the non-tracking/ tracking gyrations were expensive for everyone – brands, adtech firms, data firms and even VCs who invested in these ill-fated experiments.
Now, after four long years, the verdict is in and Google waved the white flag of surrender. They acknowledged they have no intention of killing the cookie yet “somehow” the old schema will be user private. No one knows what that meant but people are not even trying very hard to look under the hood anymore because, frankly, most are just breathing a sigh of relief.
This takes us back to where we started – how does adtech continue to track people even though people do not want to be tracked?
The answer is you can’t. In effect, we are about to date our old tracking girlfriend in a new ID dress.
Dubbed this time Universal ID, or sometimes known as Alt ID, it is; “…a unique user ID that allows AdTech companies to identify users across different websites and devices. Universal IDs are created using a piece of deterministic data, such as an email address or phone number. A hashed and encrypted ID is then created, allowing companies to identify the individual without exposing the raw data — i.e. email address or phone number,” (https://clearcode.cc/blog/adtech-id-solutions/).
Got that? More likely you didn’t because no adtech vendor really wants you to “get it.” If you did, you may realize one of two things; a) all this data is bound to misfire – a lot; or b) it still violates users’ privacy. Or, more likely, both are true at the same time.
Big data warehouses and data stores get a lot wrong – A LOT. For instance, check your Experion marketing data and you will see hundreds of attributes about what you are likely or not likely to do. You may be surprised how much is wrong.
My data file was an eye opener. Aside from the fact that they didn’t even get my last name right – “Shadiro,” this model uses algos to determine what I am likely to do or not do. This long list of attributes included all the brands of cars I am likely or not likely to buy. Problem is I haven’t owned a car in 30 years since having a car in NYC is not necessary at all. I have no intention of buying car, yet Experion’s data file clearly expects me to buy a specific brand of car. The fact that I am “supposed” to buy an automobile, means car brands will spend a ton of ad dollars marketing to me which is 100% mis-aimed.
Data behaves in funny ways. Whether “deterministic” or “probabilistic,” (deterministic system is one where a given set of initial conditions always leads to the same predictable outcome, with no randomness involved, while a probabilistic system incorporates uncertainty and randomness, meaning the same initial conditions can produce different outcomes with varying probabilities), data can be accurate at the formulaic level but fail miserably at the individual, real word level – as in my case. I should buy a car using deterministic math but I won’t no matter what the algorithm says.
Tracking data, like all marketing “profile” data before it, has always had efficacy issues because this data is largely unverified and unverifiable. Going all the back to direct mail lists marketers used to buy in the 1990s, it became untrusted because the data quality of these lists was often was abominable. This version of Universal ID is likely to meet a similar fate – untrusted by marketers to get the privacy job done.
The way forward offers brands a few options. First, it may be more efficient to not target anyone and just run media. It will be cheaper (no data ID resolution fees), more efficient and likely better on the privacy front too. Or, you can activate brand specific topic data that matches a user’s topic journey to conversion (Topic Intelligence as an example) that tracks critical topics but never people.
Sometimes less tracking data is really better than more tracking data. Sometimes, it is necessary to swim against the current and be privacy-first despite what the deterministic or probabilistic models think.
