How does IDFA deprecation impact ad prices?

One point that has been lost in the discussion surrounding Apple’s decision to effectively deprecate the IDFA amidst the hectic scramble by developers and advertisers to engineer compliance is that ad personalization drives much of the revenue generated by mobile apps. There will be a real revenue impact felt by many advertisers when iOS 14 reaches critical mass: for many app developers, revenue is a direct function of performance advertising spend. With ad personalization becoming less efficient, ad spend — and thus revenue — will decrease.

It is difficult to model the impact of IDFA deprecation on broader, market-wide CPM prices, because the underlying market data is non-public and fragmented across different exchanges, DSPs, and ad networks. The closest official data available in trying to gauge the impact of ads personalization on underlying CPM prices is provided by Facebook, which credits 50% of the CPM prices delivered to publishers on its Facebook Audience Network to the personalization it delivers through its Facebook ID (FBID).

So while a macro analysis is not possible in a way that provides credible guidance for how the loss of ads personalization will impact market prices, a firm-level micro analysis is possible given some set of assumptions. Most freemium apps have a distribution of LTV values that is right-skewed, with the vast majority (usually 95%+) of users never spending any money and some very small proportion of users spending very large amounts of money over their lifetimes. This is especially true of freemium mobile games, the in-app economies of which can support total lifetime spend of thousands or tens of thousands of dollars.

Consider such an app: the probability density function of some set of total lifetime value amounts is exponentially decreasing, with 95% of users spending $0 in their lifetimes. The set of possible, discrete LTV amounts increases on a log scale from $1 to $10,000, with associated probabilities listed below:

Now consider some assumptions around relevant marketing metrics:

The large delta in ARPU and ARPPU is driven by the positive skew of the LTV distribution (with the mean value, or ARPU, to the right of $0, the most commonly observed value). In this case, the ARPU is $0.42 whereas the ARPPU is $8.47, or more than 20X the mean value of a user. But the cost per install for this app is $0.40, producing positive user economics: an install costs less than the average value of a user, and so advertising can be done profitably. In this case, the advertiser is earning 5% margin on ad spend (the level of spend here doesn’t matter for the example).

What happens when ads personalization becomes impossible as a result of IDFA deprecation? Two things:

  1. Because advertisers experience less precision in reaching relevant audiences, their top-of-funnel marketing metrics (CTR and IR, or install rate) decrease;
  2. Relatedly, advertisers’ down-funnel metrics degrade: advertisers are less capable of reaching very high-value users with ads, and so the LTV distribution at values above $0 decreases.

As a result of these two things, the underlying economics for the advertiser change in a way that reduces the viability of ad spend. Consider a case in which CTR and IR both decrease by 10%, that is: 10% fewer people click on the advertiser’s ads, and 10% fewer people install the app from the app store page, meaning 1 – 90% * 90% = 19% fewer people progress through the ad funnel. The economics of advertising in this case change radically: assuming that CPM stays constant, then the advertiser’s CPI increases to $0.49 and its margin decreases to -15%. In order for the advertiser to achieve the same $0.40 CPI, CPM would need to decrease by 15%. In other words, CPM needs to decrease by a larger degree than either CTR or IR do in order to retain the same economics.

This example assumes that the LTV distribution doesn’t change when CTR and IR do, but it almost certainly would: with less ad efficiency, the advertiser would likely see a degradation in its app’s monetization, and thus its monetization curve would deflate. This is why blocking ads personalization is so problematic: the aggregate impact of funnel friction — the 19% decrease in users completing the ad funnel — matches the required decrease in CPM needed to reach parity. But funnel friction isn’t the only effect of a lack of ads personalization: the monetization curve is also likely to shift downward when less relevant users are reached, meaning CPM actually has to decrease even further than the funnel metrics do in order to achieve break-even acquisition economics.

Imagine that the shape of the monetization curve changes such that 97.5% of users don’t monetize as a result of the CTR and IR decreases. The curve is still negative exponential, and each value across the discrete probability density function for LTVs above $0 decreases.

With this change to the curve (which, again, is just an example), ARPU decrease to $0.11, producing a -78% margin on ad spend with CPM remaining unchanged. Thus, in order for the advertising economics to break even, the CPM would need to decrease by 78% when funnel metrics merely decreased by 10% each.

The idea here is that if top-of-funnel marketing metrics decrease even slightly, the CPM decrease needed to compensate for the lack of conversion is disproportionately large. CTR and IR don’t exist in isolation: if ads are less personalized, not only will fewer users survive the marketing funnel, but those that do will likely monetize to a lesser degree as a result of the loss of precision in targeting.

This disproportionality will have a real impact on the app economy. Because of the lack of friction and immediacy in app store payments, and because direct response advertising is so eminently measurable, many app developers don’t advertise against an explicit budget each month but rather set performance standards and simply spend as much as they can within those constraints (eg. execute ad spend against a ROAS target and spend as much as possible until that target is met).

If ads personalization reduces the efficiency of mobile advertising, there is very little existing margin to rely on to absorb that inefficiency; a +5% margin on ad spend over the course of 180 days or even a year is not unheard of for mobile advertisers. As ad spend decreases, so does the revenue for advertisers, and it’s also important to note that this impacts the revenues of publishers, too: advertisers scale spend down, CPMs decrease, and the publishers that serve ads also see their revenues diminished.

This dynamic obviously most impacts apps that are dependent on high-monetizing users that subsidize non-spenders and low-monetizing users. Apps that have a more balanced distribution of LTVs, with less extreme monetization at the high end, will be less impacted by the loss of the IDFA for ads personalization, presumably because they are more broadly appealing and less dependent on reaching very specific segments of users.

At some point, all of these values settle to a new equilibrium. But the disproportionate decrease needed in CPMs to compensate for changes to upper-funnel metrics and for a decrease across the monetization curve is important to think about in planning for the post-IDFA future.

Photo by Lydia Gulinkina on Unsplash