UPDATE 2021/01 : keep track of this fast moving situation here
With this major announcement on January 14 about Chrome browsers rendering all third-party cookies obsolete by 2022, Google has shaken the ad/MarTech world to its core. Most digital marketers expected it to happen, perhaps with the hope that it would only materialize in the next five to ten years. Alas, the countdown to the Cookie Apocalypse has now been ushered in, and this leaves many ad/MarTech solutions no more than two years to either adapt or literally go extinct.
The Analytics and Data Science team at Adviso has been doing intensive research on this subject, especially in the last few months. In the coming days and weeks, the team will be publishing a more extensive and technical take on the implications of the third-party cookie collapse to the Adviso blog. Certainly, there are many implications to be listed and the effects will be far–reaching, especially for those who have not been heeding the warning signs for the last few years, including very strong signals from Apple’s Safari and GDPR.
For now, we simply want to look at two things at a very high level:
These two points will only be a preface to the slew of content coming shortly from our Analytics and Data Science team. So stay tuned by signing up for our newsletter.
If you’re a developer or MarTech expert, you most definitely should read the details of the Chromium Privacy Project. For the rest of you, I will outline what the Cookie Apocalypse means in a nutshell.
In essence, Chrome will create a browser sandbox which will serve as somewhat of a black hole for cookie IDs during server calls. Currently, when ads are fired on a publisher website, the pixel captures a host of data about the specific user. These data include when the user clicked, where, what device the user was on and, in some cases, especially with third-party data vendors, it tells the host server about your previous navigation patterns, namely if you’re in-market for various products or services. All of this is going out the browser window (pun intended).
Starting in 2022, Chrome’s sandbox will intercept all this data and dissociate it from any specific ID. After stripping out the data from the server call, the browser will return it as a sort of batch file in the form of metadata, which will not be connected back to any individual user.
Some global information will remain, such as the number of clicks and conversions overall, but outside of knowing from which network or website you obtained visits to your landing page, your third-party cookies won’t be able to remarket users. The information will be lost in the server transaction, a.k.a. the cookie black hole.
With a 66 percent market penetration rate, Chrome effectively puts an end to many careers with this one move. Furthermore, all of this takes place in the context of a broader and more complicated data war involving GAFAM (Google, Amazon, Facebook, Apple, Microsoft) and telecom giants like Verizon and AT&T. You can read my 2018 academic article on the subject published on The Conversion to learn more about this subject (co-signed by my thesis professor, pioneer data engineer Dr. Daniel Lemire, and my data science Adviso colleague Dr. Nicolas Scott).
Having said that, let’s not go into a panic about all of this. For those who will adapt through this important evolution of the market, there is still hope.
According to Ari Paparo in his excellent article on Adexchanger, important areas of the ad/MarTech world are about to take a fall, with little chance of ever getting back up. They will get sucked into the black hole of the Chrome browser sandbox. We already see signs of this. For example, on the day of the announcement, Criteo’s stock fell dramatically, along with that of several other ad tech companies. This indicates clearly how remarketing solutions based entirely on third-party server calls will be the first and main casualties, particularly in display.
According to Ari Paparo, the short list of digital marketing activities and technologies at very high risk of extinction are the following:
For each of the above points (as well as others, such as the implications for mobile tracking with an SDK), our Analytics and Data Science team is preparing a more exhaustive list of detailed articles to go into more technical explanations about what the Cookie Apocalypse means and what the alternatives are. Also, keep in mind that technology and the market are always changing. Therefore, it is always possible that new alternatives might emerge to salvage some of those third-party cookie–based platforms and features, but at the moment it seems unlikely.
In the meantime, I’ll briefly discuss three ways in which marketers can prepare for this now–imminent and official apocalypse date, though even these points will also be defined in more detail in upcoming articles. (You can subscribe right here if you want to stay tuned!)
While user privacy is a highly noble cause, it would be somewhat naive to believe Google is making this move purely for altruistic purposes. After all, the winners in this post-apocalyptic world will be the companies with the most first-party data, namely from login and apps, etc. This of course puts Google, Facebook and Amazon in a very good place within the digital advertising landscape (as if they weren’t already).
In fact, Facebook was one of the early advocates of the Cookie Apocalypse. After all, they need none of it, with 100 percent of their users having to log in. Google and Amazon are in similar places with their own hefty bags of users.
However, for the marketer, it means you have to get used to living in between walled gardens. Data from one ecosystem won’t transfer to the next and you won’t be able to cross these data directly. This is already the case for Campaign Manager users who can’t add impression pixels on Facebook ads. The reason is simple: Facebook won’t let Google siphon its precious data. Likewise, Google Display Network and Amazon DSP won’t allow any DMP impression pixels into their ecosystems. In essence, marketing data will become more fragmented and siloed. Get used to it!
If Google, Facebook, Amazon and even Verizon Media (which is planning to launch its own search engine, by the way) can amass first–party data, why can’t you? If you own a DMP licence without the option of tracking at the first-party domain server level, you might want to have a serious talk with your current vendor. If you do leverage first-party server calls with your DMP, you should still have a serious talk with your vendor.
In truth, most DMP solutions will be deprecated in the next two years. Even with first-party tracking subdomain servers, a long list of cool but expensive features will simply be obsolete by then.
As such, advertisers will need to turn to ad/MarTech solutions prioritizing the collection, management and control of your own first-party customer or user data. Among that list of solutions, customer data platforms/infrastructures are emerging as a potential alternative option. Moreover, there might be a ray of light coming from second-party data collaborations, but we won’t go too far down that road at this time. Stay focused on the first-party!
You might not know it yet, but very soon, a fellow by the name of Reverend Thomas Bayes will be your new best friend. This is because Bayesian statistics and data science will play a big role in stitching back together the mess that will be your marketing performance attribution and measurement data.
As reported by The Drum, according to James Parker, Chief Solutions Officer of Data and Planning at Jellyfish:
“There’s going to be a limited amount of platforms where you can join all the data together. We are kind of going back to the days of media mix modeling.”
If you’re not familiar with media mix modelling, it is a statistical technique going back to the 70s that was used to optimize media spend, namely TV, radio, print, etc. This was long before the internet and direct user-level click-based attribution models.
Today, with the advent of machine learning and more advanced statistical methods, those approaches are going to play key roles in measuring marketing success and performance. Optimization will require advanced statistical methods of inference, which is why you will need to have access to data science expertise. With the end of pixel tracking in combination with large amounts of fragmented data from various walled gardens, fields like Bayesian statistics for marketing are about to take centre stage again, along with marketing AI.
The above is true for both Darwinian evolution of species and business ecosystems in general. While the news might be frightening for some marketers, I see this change as not only good, but inevitable. While programmatic ad buying had its days of glory and was great, it also created many frustrations, both for the end user and the marketer. In fact, solutions like DMP, without proper technical guidance and a solid strategy, often oversell and overpromise on the dream of an ad/MarTech nirvana.
For many advertisers and agencies, this dream became a technological nightmare with a very salty price tag, often gathering dust among other would–be technological silver bullets. Marketers in the post-Cookie Apocalypse will have to return their focus to actually doing marketing analytics, as opposed to chasing vanity metrics. Going back to basics means revisiting your data with a new scope and new lens. Ultimately, let’s hope the new era of user privacy brings advertising back to what really matters, which is establishing a useful connection between a brand solution and a consumer need.