Facebook-TikTok data changes: what marketers need to know
Facebook deprecated Advanced Mobile Measurement. Tiktok deprecated advertiser accessibility to user-level impression data. While MMPs like Singular still have access to view-through attribution on TikTok and device-level install measurement data on Facebook, tightening privacy controls means only aggregated and privacy-safe summaries can be released to advertisers.
What does that mean for the future of mobile marketing?
The answers are here, in our recent webinar.
More questions about Facebook, TikTok, and marketing data
But there were some more questions about Facebook and TikTok, marketing data, and what this all means for the future of marketing measurement. As usual, we answered as many as we could during the webinar, and had a few left over.
Here they are, with answers:
Q: Do you think this shift with providing less info to marketers will force marketers to explore other platforms including broadcast, radio, etc. Social is going to become increasingly challenging …
How do you see the evolution of contextual with in app? Or is it just a buzzword?
A: Yes, I think marketers will broaden their reach and channel selection. We’ve seen it already with mid-tier ad networks. But let’s be real: Facebook and Google and the other self-attributing networks are 75% of mobile user acquisition spend, and that’s not going away anytime soon.
Also, on contextual advertising: it works incredibly well in some verticals. Gaming is one that stands out for me. I don’t think this is just a buzzword, and I do think it will get increasingly important. The downside: targeting whales will be harder.
Is the sky falling?
Q: Moving forward in an ecosystem without user level data, how do you see publishers assessing campaign performance?
A: First off, we still have device-level data on Android. We still have device-level data for about 20% of iOS. And we still have deterministic data in SKAdNetwork (yes, aggregated data, but still deterministic). Plus there are all kinds of probabilistic methodologies that mobile marketers are going to have to learn: incrementality, MMM, etc.
To get a good primer on where Singular sees this going, check out this article on post-IDFA user acquisition by Singular’s CGO Ron Konigsberg as well as my post on Google “killing” last-click attribution.
Social, I’m outta here
Q: Would you suggest that marketers shift from social altogether and shift to SEM and/or native? We have seen an 80-85% drop in revenue and number of transactions for some of our clients and some are doing the same or better from when before the iOS update was implemented.
A: We’ve seen some of those massive swings. Typically what we’ve seen is massive dips in campaign effectiveness that turn out to be artifacts of applying yesterday’s measurement methodology on today’s mobile advertising mechanics. In other words, if you’re using an IDFA growth stack under SKAdNetwork, it’ll look like the roof caved in.
You must, must, must adapt to the new realities.
That said, if the drop is real and not just perceived, that may be due to a severe impairment of the biggest channels’ ability to find whales specifically or payers in general. Give them some time: they are working through change just like you.
And, consider how you can use SKAdNetwork and day-one predictive insights to approach your former efficiencies.
Trust the process?
Q:In light of the Facebook changes, and their driving ad buyers into utilizing automated app ads, what’s the recommendation on running/scaling budget toward Automated App Ads versus a more granular, fully segmented approach?
A:Remember the first months of Google’s algorithm-driven mobile app install campaigns? Remember how much $$$ you needed to train Google’s ad-to-prospect matching AI?
First, give them some time. I do think they’ll figure it out. Second, if you’re going to use the giant platforms, I would assume they know more than you and let them stir, mix, and add the ingredients themselves. (Also, make your campaigns too granular and you’ll lose too much data to SKAdNetwork privacy cutoffs.)
That said, if you’re Rovio or another big team with massive data science and incredible BI, give it a shot.
Everything is crashing to the ground?
Q: So will MMPs not be able to disambiguate conversions between sources eventually? Will there be pools of overlapping conversions between sources? What will the role of MMPs be in this landscape?
A: MMPs are still able to disambiguate sources. And get access to a ton of granular data they cannot contractually share with advertisers. MMPs will not only start using multiple models to generate a better picture of reality; they’ll also need to be able to allow advertisers to run models in the MMP’s cloud to generate insights based on data that can’t be shared, and then export the resulting insights back to advertisers’ BI systems.
For more detail on what that might look like, check out Singular CEO Gadi Eliashiv’s blog post on exactly that topic.
Be specific, please
Q: When you say “We can’t share the device-level data, but we can share measurement insights derived — at least partially — from it …” what are the insights that will still be available?
A: Check out that blog post I linked to in the question above, specifically the section titled “How you can still do granular analysis.”
What’s the impact?
Q:What is the impact on small/medium-sized advertisers and publishers? Do you think it will be more difficult for them to acquire sales or generate revenue?
A: Larger advertisers can certainly use more campaigns via SKAdNetwork-based measurement, because they can feed enough volume into each to avoid privacy censorship levels from destroying their ability to optimize. But we also see plenty of smaller players being super-nimble in web-based acquisition flows that are much cheaper than in-app ad flows, and offer some other measurement advantages as well.
Q: Instead of sending users directly to the App/Play Store, could directing users to a first-party website and using MMP tracking links solve this hole?
A: That’s certainly an option. You might (see above in #7) try models that originate on the web as well, but there’s plenty of in-app inventory that you’d want to access, of course. Going app to web to app is an extra hop, but could enable a sign-up phase or a more detailed introduction to your app which could result in higher-value users/customers making it all the way through the funnel.
So it’s all one big melting pot now, huh?
Q: How will you measure the performance of a TikTok campaign vs Facebook in the future? If the data is in “one pot?”
A: It’s not in one pot. Sources are still very distinct — there are lots of pots — and there’s additional data about conversions and performance within each source. The changes just mean there’s a little less granularity in each pot.
Aggregation and consolidation
Q:Do you see data aggregators starting to offer media buying? or vice versa? Through acquisition of teams or growing their own?
A: Hmmm … you wouldn’t mean like Applovin acquiring Adjust, or InMobi acquiring Appsumer, would you?
Q: What are the opt-in rates currently being observed for getting access to FPD?
A: It really, really, really depends on what kind of first-party data you’re referring to. If you’re talking IDFA, it’s about 20%. For more on first-party data, check out this iOS 15 app marketing webinar.
Top funnel versus bottom funnel
Q: Engagement with creative doesn’t necessarily correlate with conversions or ROAS … with this in mind, why the emphasis on changing the way we develop creative? Will creative benefit from being made for more broad audiences?
A: Good creative gets attention. Bad creative does not. After that, there are two questions. First, did you get the right kind of attention, and second, what are you going to do with that attention? High CTR is good, but if it’s the wrong kind of attention, it won’t hit your bottom line. Bad CTR is irredeemably bad … unless the one in a thousand person who clicks on it also becomes a whale.
More questions? More answers …
There are always more questions.
If you have more that are not answered here or in the webinar (check it out here) then there’s a good time and place to get them: in a demo with some of our experts.