Eric Seufert Profile picture
Oct 1, 2020 13 tweets 3 min read Read on X
I've heard people claim that using IDFV + IP Address will allow ad platforms to profile and target users as efficiently as they can now once the IDFA is deprecated. I dont think this is the case;.I'll explain why in this thread (1/X)
2/ First, a primer on IDFA deprecation and an explanation of the IDFV: mobiledevmemo.com/mobile-adverti…
3/ The IDFV + IP Address argument is: an ad platform will "see" IDFV + ip address combinations from apps a user has open in the background. If the time interval across which those combinations are polled is short enough (eg. 30 seconds), then changing IP address is mitigated
4/ The idea is: in real time the ad platform knows that this unique IDFV has this IP address from background app data, so if that IP address shows up in bid stream, it belongs to the IDFV from that app, and the ad platform has aggregated in-app data for that user in that app
5/ But IP addresses aren't unique. If I'm in an office with 99 other people, and a quarter of us are on iPhone 11 with the latest iOS version, then there's only a 1/25 chance that the IP address being seen is me
6/ But let's say its higher than that: a 50% chance the IP address is me specifically. Two problems: 1st, how does the ad platform have any knowledge of my spending habits? SDK networks could conceivable ingest IDFV-indexed in-app events, but why would developers pass them?
7/ 2nd: in-app purchasing is extremely rare. Across all apps, globally, maybe 5% of users make purchases? And app advertisers tend to run very tight margins on their ad spend. If I target the wrong user in the above scenario, there's only a 5% chance they'll monetize at all
8/ Which prompts the question: what is the benefit of the IDFA in the first place? It is that ad platforms can create IDFA-indexed device graphs via in-app event streams and profile spenders / engagers on a spectrum. Contextual targeting is also an option but is less efficient
9/ Put aside the question of IDFV-indexed event streams and assume every platform can do that (big, questionable assumption but not core to my point): how much lower can efficiency of IDFV+IP targeting be relative to IDFA-targeting and still outperform contextual?
10/ What's the average global IP overlap in that 30-second poll of IDFV+IP combos? Is it 25%? 50%? Are most ad impressions surfaced for people that are at the office with 100 other identical device specs, or are they surfaced for people at home with none other?
11/ And now one has to question: why did Facebook claim that FAN will be severely impaired by IDFA deprecation if they could simply use this IDFV+IP profiling method? If any company could execute this incredibly difficult technical feat, Facebook could.
12/ I've learned to never underestimate the ingenuity of ad tech companies. But my sense is that IDFV+IP address is being billed as a panacea by ad tech vendors because otherwise they'd need to acknowledge the immense damage that IDFA deprecation will inflict on their businesses.
13/ That said, I'm open to being wrong about this. Feedback welcome.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Eric Seufert

Eric Seufert Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @eric_seufert

Nov 8, 2023
The black box inside the black box: Google announced yesterday the availability of its Generative AI-based creative tools in Performance Max campaigns. What considerations should marketing teams make in expanding total campaign automation to creative production? (1/X) Image
2/ First, dispelling two myths. The first: marketing teams view Generative AI as a novelty or a toy that is not yet practically useful. This simply isn't true: I've seen marketing teams that have radically improved their workflow with Generative AI tools already.
3/ Second, wholly automated campaign optimization tools like Advantage+ and PMax are naturally hostile to advertiser goals. This isn't true, either. These tools can present competing incentives, but many advertisers benefit materially from their use. mobiledevmemo.com/google-pmax-me…
Read 9 tweets
Nov 1, 2023
Apple and Google race to attribute the internet

The control exerted by Apple & Google over the consumer internet is often expressed in terms of content discovery / distribution & payments. But a more subtle and esoteric form of control is emerging: advertising attribution. (1/X)
Image
2/ Both Apple & Google have launched native advertising attribution frameworks for their mobile platforms & browsers. These dictate how and, crucially, how accurately digital advertising can be evaluated, based on rules set by these companies.
Image
Image
3/ These frameworks have been introduced alongside, or as components of, privacy policies that were authored by the platforms themselves. Moreover, it seems that the platforms' privacy restrictions don't consistently apply to their own advertising products.
Read 18 tweets
Oct 25, 2023
Meta announced changes to its Aggregated Event Measurement (AEM) protocol this May. Meta introduced AEM a few months after Apple revealed (but before it rolled out) the App Tracking Transparency (ATT) privacy policy. (1/X) Image
2/ AEM was initially modeled on Private Click Measurement, Apple's own privacy-focused attribution framework for web-to-web and web-to-app advertising campaigns. Meta stated as much in an early version of its documentation for AEM. Image
3/ But I noted when Meta first announced the changes coming to AEM that the reference to PCM had been removed from its documentation. I interpreted this as meaning that AEM would no longer be tethered to the PCM design imperative. Image
Read 9 tweets
Jun 30, 2023
Yesterday, The Verge reported that Meta will introduce a direct-to-install advertising product on Android in the EU once the DMA goes into effect next year. Some thoughts on the efficacy of such a product and its impact. (1/X)
2/ First, I believe the DMA will be systemically disruptive (in the EU). It has broad implications for all "gatekeepers" / large platform operators, not just on mobile. To my mind, the DMA represents a fundamental reset on competition in consumer tech. mobiledevmemo.com/a-deep-dive-on…
3/ Meta says that its ad product will allow consumers to install apps on Android directly from an ad click, sidestepping the intermediate step of visiting Google Play. This has the potential to meaningfully improve conversion rates (and thus decrease acquisition costs).
Read 11 tweets
Jun 6, 2023
Yesterday, Apple announced its new Privacy Manifests feature, which takes direct aim at device fingerprinting on iOS. Privacy Manifests will hold SDK publishers and app developers accountable for how user data is collected and utilized. (1/X) Image
2/ Apple explicitly stated in its blog post announcing Privacy Manifests that their intended purpose is to disrupt device fingerprinting to force app developers to indicate a legitimate use case for data collection by potentially non-compliant SDKs. From the post (emphasis mine): Image
3/ Apple's approach here is, to my mind, ingenious: by effectively segmenting SDK permissions from general app permissions and forcing developers to certify that SDKs are behaving in accordance with App Tracking Transparency, Apple places the onus of compliance on developers.
Read 14 tweets
Jun 5, 2023
Per this blog post, published today: Apple is taking direct aim at fingerprinting with Privacy Manifests. developer.apple.com/news/?id=av1ne… ImageImage
Apple seems to be saying that app developers will be held liable for the validity of SDK data use attestations through the privacy manifest system. Will a BigCo legal team be willing to sign off on data usage claims by a third party that it knows to be practicing fingerprinting?
I’d characterize this list of SDKs as “commercially sensitive.” Image
Read 4 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(