Rabbbits Weeekly: Cookie Monster
Or, what happens when the internet takes all our cookies away? And the usual round-up of analytics, ads, audience, algorithm, and attack vector news.
Last year I wrote a piece for the Blue Ion newsletter about the coming of the Cookiepocalypse. (I've updated a few pieces to make the info more current.)
Cookie Monster
Is the modern era of internet advertising coming to an end? There are plenty of think pieces that will try to convince you of that. iOS 14.5 dismantled tracking users across services, and when Google phases out third-party cookies in Chrome in the not-too-distant-future, it’ll join most other major browsers to have done so.
Let’s recap, shall we?
In iOS, users must now opt-in to allow apps to track them and Facebook advertising is taking a hit
Android will go halfway later this month with the safety section of the Google Play store (more transparency, less opt-in/out)
Soon, most major browsers will have similar measures in place
iOS 15 introduced email tracking opt-out, which means no IP data for location and incorrect open rates (think 100%)
What comes next? [Insert shrug emoji here.] If Google has its way, maybe we’ll FLEDGE and Topics. Perhaps Apple launches an ad platform and we start paying a different tech giant. Or contextual advertising comes back in a big way (the seller side looks to be pushing this angle). A common feature of many prominent alternatives is to collect and maintain user data on a user’s device or in their browser and share an anonymized signal with ad platforms that acts like a yes/no button for whether or not the user is in the targeted audience.
In the meantime, the safe approach looks to be leveraging data you own (CRM* or email list) or the platforms own (interests and engagement actions happening within their walled gardens).
At this point, there is more uncertainty than clarity. But the safest bet is the same as it’s always been: own the customer relationship and build your own database (it does not have to be a literal database).
Buckle up, it’s going to be a bumpy ride.
*We can help
And now you don’t even have to take my word for it, you can take McKinsey’s! The 3 consultant-certified recommendations?
Collect first-party data (the stuff users voluntarily share with you)
Create partnerships to leverage second-party data (your partner’s first-party data)
Run contextual ads (the ads match the content of the page the user is on; or, the way internet advertising used to run.
More Cookie Crumbs
A helpful infographic recapping where we are in the privacy process and some post-cookie options.
MiQ is throwing their hat into the post-cookie solution ring. It's a 3 pronged plan that includes a new identity graph to allow for targeting, an on-ramp partnership for first-party data, and another partnership for contextual intelligence for more targeting. I don't know what solution(s) will win out but it seems pretty clear at this point that it will be cohort based. So instead of targeting individual users (not exactly how it works currently) you'll target groups of users based on similar interests / signals.
And another option! This one is called Seller-Defined Audience and seems like it's essentially first-party data augmented contextual advertising. No cross-domain data sharing or tracking. The crux of this approach looks to be centering the publisher in the value chain, therefore bringing more money to them than an intermediary.
This post also starts with a counter-point to Google's Topics if that's something you're interested in.
First they take away our cookies, then they hand us a plate of statistics. Get ready for probabilistic attribution (aka what The Zuck Bucks Metaverse Printer™ and (probably) Goog are already doing), machine learning-powered probability mapping for conversion and attribution measurement.
We've covered the topic of Topics. So let's talk about FLEDGE now. It seems that FLEDGE operates on the same principle as Topics (the browser is the key) but the interest groups are custom created to operate as remarketing and custom audience groups for advertisers. I think.
Clean rooms could be another solution for post-cookie retargeting (though it only works for those people who have shared their information with you in some way (usually forms) and not just anyone who stopped by your site for a cookie). Roku has launched theirs, allowing (potential) advertisers to upload their first-party data and–securely and privately–match it to Roku’s data for targeting.
It appears that advertisers aren’t terribly concerned about the cookiepocalypse, or potentially even preparing for it, but the IAB is less optimistic. This likely reflects the difference between the IAB being a group that thinks about standards and “the future” and advertisers focusing on what works now and being very used to adapting in the frequently shifting digital world (I expect the Facebook interface to be different every time I open it).
Analytics
This is a weird report in my mind because it basically says what we knew was going to happen, the rollout of Mail Privacy Protection on iOS would make open rates unreliable. (I don't think this was the first time this approach has been taken either, I think Gmail does something similar.) But a good reminder to ignore your open rate.
Predicting the future, now with AI! Researchers at MIT have developed a tool that takes in time-series data and spits out predictions, all without you needing to get your hands dirty. Give it a test drive.
Audiences
A very salient piece of advice from Ben Evans (it was in his newsletter so I don’t have a link for it):
Reminder to anyone going direct: your users are the most important thing in your life, but that doesn’t mean they think about you at all.
But it looks like your audience thinks about spending money. Visa is projecting revenue to "accelerate past pre-pandemic levels" as the less-than-rosy economic conditions don't seem to be slowing consumer spending.
Consumers spent more (in retail categories) on Amazon than at Walmart for the first time ever in the last quarter of 2021.
Metaverse? More like Meta-worst. Currently, the metaverse is like your favorite indie band before they "sell out", not many people know about it. A recent poll found 38% (of Americans) were "familiar" with "the metaverse" but only 16% could actually define it. Seems like a stable idea to build the future of your company on…
Speaking of, Zuck+co lost a whole bunch of money on this plan last quarter. Like, $3 billion bunch.
Ads
Apple’s ATT privacy measures did not direct all that money Facebook complained about to their own bank accounts. Or so this (Apple funded) paper says. Which makes sense, someone probably would have noticed if billions more dollars were being spent on App Store ads.
TikTok has launched a new (and distinctly TikTak-y) ad format: Spark. This is a fancy name for UGC or sponsored content ads, but seems like the setup flow maybe easier than what other platforms have. The content can be yours or someone else's, as long as you have permission. If TikTok advertising is going to explode, this is probably the ad type that'll make it happen.
Ads are coming to Google Goes TikTok, sorry, YouTube shorts. Short-form video eats the internet?
The Goog is also adding new toys to the Performance Max product: optimize for new customer acquisition, get more insights into asset and ad performance, and a one-click upgrade to performance max feature for smart shopping and local campaigns.
Something to watch out for in your PerfMax campaigns:
Location targeting is (currently?) like other campaigns' "interested in" setting vs. a strict geofence. You can exclude locations, which is not the most efficient use of time. Google may be making some changes here, so we'll see what the future holds. If you need it, here's a list that might be helpful.
But wait, there’s more PerfMax stuff! Audience signals are the starting point for learning but the algo will then do its thing after that. Rumors this may change, but who knows.
Facebook does what was probably a whole bunch of research to come to the highly novel and not at all obvious conclusion that the page you send clicks to should be fast, relevant, answer questions, and look good on a phone. This is why they get the big ZuckBucks™?
Maybe ad platforms are listening after all. Research shows Amazon uses Alexa devices to capture data for ad targeting. Quick reminder: these are Amazon controlled mics, your phone mic is not controlled by Meta.
Algorithms
So, uh, Facebook doesn't really know much about their user data, like where it goes or what they do with it. Which is probably fine, right? It will definitely make privacy and regulatory compliance fun. I think this quote from the leaked doc sums it up well:
"Imagine you hold a bottle of ink in your hand. This bottle of ink is a mixture of all kinds of user data (3PD, 1PD, SCD, Europe, etc.) You pour that ink into a lake of water (our open data systems; our open culture) … and it flows … everywhere. How do you put that ink back in the bottle? How do you organize it again, such that it only flows to the allowed places in the lake?”
Zuck's money printer has jammed, some short form videos and a pesky war got stuck in the gears.
Speaking of not knowing, Twitter apparently doesn’t know how to count? They miscounted their daily active users by millions, MILLIONS, (ok, up to 1.9 million, but still) for the past 3 years. THREE YEARS! And this isn’t the first time they forgot how math works, on this same metric!
Amazon is making a metaverse move with...some form of hardware yet to be developed that has to be both "magical" and "useful" while being "new-to-world", which I guess means not ripping off an existing idea and slapping your own logo on it? This would explain why they're not in Silicon Valley still still be a novel concept for them (see: Basics).
Prime, not just for Amazon anymore. Buy with Prime is becoming a logistics platform (as-a-service?).
Snap is betting it all (or at least most of it) on AR, including thinking of themselves as "a leading augmented reality platform."
The tide appears to be turning in social media, and the new guard is ascending. The latest illustration is the news that Snapchat is now growing faster than Facebook and Twitter (when talking users). The Ghost’s™ DAU’s increased 18% year-over-year and revenue increased 38%
Traffic lights, they're just like digital marketing! In that they will soon run on AI.
Attack Vectors
There's a cadre of (mostly-Israeli?) hacks-for-hire spyware companies that are typified by the headline-making NSO Group. They sell exploits to governments, but definitely not the bad ones, and definitely not just anyone with money. Well, EU officials have been targeted using their "solutions". I'm sure it's fine.
We’ve talked about the UK’s Online Safety Bill before, a portion of it that has been newly brought to my attention is the bit about identity verification. It’s an effort to curb, well, the nasty bits of internet use like Elon Twitter trolls and YouTube comments. It’s opt-in for users, which feels like the better approach compared to forcible collection of personal information (yeah, yeah, I know). Yet another European regulation that will be interesting to watch play out.
More rules from the EU! The newest addition to their Digital [insert word here] Act series is here and S is for Services. One of the requirements of the DSA is that platforms disclose the workings of their algorithms (mostly recommender / profiling related) to the public. Should be thrilling reading that everyone will definitely understand.
New research suggests that YouTube doesn't radicalize people, it just amplifies extremism in people that were already leaning that way and enter the "rabbit hole" (this one doesn't get 3 b's, I don't want it) via external links and subscribing to channels.
Google’s added some new policies for search removals that will let you request to not be doxxed. That’s nice.