The requisite recap: third-party cookies (the kind used by ad tech to measure performance and attribution across the web) are under attack by the GDPR, CCPA, Safari ITP, and a host of other ghoulish party poopers.
But you already know this. You may also know that not all cookies are created equal and that first-party cookies (the kind that let your favorite sites automatically “recognize” you) are, by contrast, not in the same imminent peril.
Lastly, you may have an inkling that this whole “first-party” thing may be key to figuring out how measurement and attribution are going to survive and thrive in this new era.
You would, of course, be right. Between a dusting-off of classical ad measurement techniques and some serious geeking out over first-party data, advertisers will find that, far from being the airless, suffocating hellscape they have been taught to fear, this new “cookieless” world will prove a hospitable and possibly verdant environment for measurement and attribution.
You Maniacs! You Blew It Up!
But back to third-party cookies—their troubles are throwing a wrench into a lot of measurement tactics. A crumbling cookie is making it harder to identify “User X” in ad server logs and know that User X saw ads around the web on Days 1 and 2, and then converted on Day 3.
This blindness ravages user-based, deterministic attribution models. Impression, view-through, and conversion tracking all depend to some extent on third-party cookies.
Cookies: the Blockbuster Video of Ad Tech
As if cookies’ current troubles weren’t bad enough, cookie-based measurement always had glaring deficiencies. Cookies are easy for bad actors to fake, they’re ill-suited to today’s cross-device world, and finally, disparate cookie pools can be near-impossible to deduplicate, encouraging “cookie bombing” and leading to rampant double-counting.
Pour one out for those plucky little TXT files and let’s move on.
Sometimes the Old Ways Work Best
Did you know that marketers spent the latter half of the previous century perfecting cookieless measurement techniques with zero user-level insight? It’s true, I swear!Cookies are easy for bad actors to fake, they’re ill-suited to today’s cross-device world, and finally, disparate cookie pools can be near-impossible to deduplicate, encouraging “cookie bombing” and leading to rampant double-counting.
(Actually, that’s a mischaracterization—even our slow-witted 20th-century forebears used direct-response advertising to amass troves of “user-level” data.)
Luckily this isn’t the 1950s and my car gets more than 5 miles to the gallon. When you take “classical” measurement techniques and deploy them within a digital ad architecture, you can get pretty powerful results. If user-level measurement no longer yields enough data, just dust off your “Ad Ops 101” textbook and start parsing data by geography, publisher, creative, keyword and so forth. Then throw in some off-the-shelf machine learning software to find new and non-obvious patterns in your measurement data.
Last but certainly not least, panel data and incrementality (A/B) testing can deliver excellent insights without the omniscience of user-based, cookie-dependent tracking.
Options abound.
And the Detail-Oriented Shall Inherit the Earth
However, dusting off “the old ways” means adhering to data taxonomies, trafficking campaigns and buying media with near-fanatical diligence. Being able to glean crisp insights from full-funnel reporting data that no longer lets you see user-level journeys simply isn’t possible without strict attention to detail.
“Data Clean Rooms”: A Safe Space for First-Party Data
With third-party data on the ropes, it follows that advertisers, publishers and walled gardens might come together and see what they can learn by comparing their respective first-party data sets. Did your readers who saw my ads then convert on my site? Let’s put our first-party data together and see!
Two big problems with this. First: first-party data usually contains personally identifiable information (PII). Customer privacy needs to be protected when user data leaves a company’s walls, even in hashed form.
Second: first-party data is super valuable! Parties are understandably reluctant to share it.
A clean room can match my timestamped sales transaction records to your timestamped ad server logs and I get a report showing how many users appear in both data sets (people who saw ads and converted) along with the time lag between ad exposures and conversions.
Enter the concept of “data clean rooms.” They offer a novel solution to the dual problems of user privacy and the self-interest of data owners. Multiple parties upload data sets, but none of the parties can directly access the pooled data. The reporting you get from a clean room is an aggregated look at where your data sets intersect with others.
For example, a clean room can match my timestamped sales transaction records to your timestamped ad server logs and I get a report showing how many users appear in both data sets (people who saw ads and converted) along with the time lag between ad exposures and conversions. Voila! A basic attribution model!
Google’s Ads Data Hub and solutions being developed by Amazon and Facebook are a few examples. However, the idea isn’t limited to walled gardens. Any publisher, media company or retailer (think New York Times, Netflix, or Walmart) with a large owned audience could offer a similar service.
The Future Is Where It’s At
Even if a cookieless future consisted of nothing more than incrementality tests and data clean rooms, advertisers would have years worth of new toys to master and traditional tactics to revisit. But the post-cookie technology and regulatory landscape is just now dawning, and we can look forward to a great deal of innovation.
And frustration.
And confusion.
And surprise.
But our industrious protagonists, Measurement and Attribution, far from playing the victims in a world beyond their control, have been laying the foundations for a bright future.