Let’s start with a theoretical framework. For simplification purposes, assume all all a company’s customers have the same lifetime value (LTV), equal to the present value of incremental profit they generate after becoming a customer, and the company has an effectively infinite pool of customers it can acquire.
Again for simplification purposes, assume a company can only use one marketing platform. Also assume each ad (impression) bought acquires one customer. The price of ads varies.
Simplified Pricing Algorithm
To maximize the net present value of ad spend, the company is willing to buy all ads with a price less than the LTV to acquire a customer:
Buy ad if price < LTV
Total Ad Spend Formula
So the company creates a campaign that automatically buys ads if the price is less than LTV. The total marketing spend is the average price spent per ad, X, which is less than LTV, multiplied by the number of ads bought, Y. The number of ads bought, Y, is based on the number of ads available to the company with a price less than LTV. The more ads with a price less than LTV, the more ads bought.
Total Ad Spend = X * Y
X = Average Price Paid per Ad, Y = # Ads Available with Price < LTV
where X < LTV
Impact from Average Price in the Platform Increasing
What happens to a company’s total ad spend (which in this stylized example equals CAC, since each ad acquires a customer) when the average price of ads in the platform as a whole increases? Assuming the distribution of ad prices is the same, the number of ads available with a price less than LTV decreases. Therefore, if the average price of ads in the platform increases, the number of ads bought will decrease, and the company’s total marketing spend should decrease as well. While the company’s average price paid per ad should also increase, since the company was already purchasing all ads available with a price < LTV, the reduction in number of ads bought should outweigh the increase in average price paid per ad, resulting in an overall decrease in total ad spend.
Supply-Side Example: Facebook
Outside of Google, Facebook is the biggest advertising platform in the world. And, relative to last year when ad prices were depressed due to the pandemic, ad prices on Facebook are increasing big time. From Facebook’s most recent earnings release:
Advertising revenue growth in the second quarter of 2021 was driven by a 47% year-over-year increase in the average price per ad and a 6% increase in the number of ads delivered. Similar to the second quarter, we expect that advertising revenue growth will be driven primarily by year-over-year advertising price increases during the rest of 2021.
Demand-Side Examples: Root, Wish, and Zynga
On the demand-side, over the past week, we saw this dynamic play out in the earnings reports of three companies: Root, Wish, and Zynga.
Here’s Root, clearly saying that ad price increases are causing them to spend less on ads:
This quarter, competition and cost for space in performance marketing channels increased significantly, resulting in unacceptably low returns on our investments in those channels. As we position Root for long-term success, we are conscious of bringing the right risks onto our books at the right cost. With the significant increase in online ad pricing this year we have decided to significantly reduce our spend in the performance marketing channel.
And Wish, saying basically the same thing (by “lower marketing efficiency,” I think Wish means that the higher ad prices resulted in fewer customers available to acquire at an acceptable price):
At the same time that engagement was declining, the cost of digital advertising on leading ad platforms, which we historically have used to drive demand and conversion on our app, increased more than we expected. In addition, the recent privacy changes for iOS have caused more advertisers to shift spend to Android devices, creating more competition for a limited supply of impressions. Ultimately, this drove up competition for advertising bids, restrained our ability to reach more users and increased advertising costs for Wish since most of our growth marketing has been focused on Android, the preferred device for the majority of our users.
These rising digital advertising costs contributed to lower marketing efficiency. Therefore, during the quarter, Wish’s proprietary data science algorithms reduced our digital advertising spend.
And finally, Zynga, highlighting how the increase in cost to acquire players on Apple meant less players available with a CAC less than LTV (target returns):
… the adoption of Apple’s privacy changes resulted in a higher cost to acquire our players. In response, we scaled back our UA [user acquisition] spend to maintain target returns, resulting in fewer players installing our games during this period.
Why Are Customer Acquisition Costs and Ad Prices Increasing?
An interesting secondary question is what is causing customer acquisition costs and ad prices to increase in the first place.
Increase in CAC - iOS
Let’s look at customer acquisition costs on iOS first, since that seems to be the root of all of this. Apple’s targeting changes have made ads less effective by increasing the number of ads needed to acquire a customer. For example, if before it took 10 impressions to acquire a customer and now it takes 15 because the ads are less targeted, the number of impressions needed to acquire a customer increases. Assuming no fully offsetting decrease in the cost per impression, the cost to acquire a customer (# of impressions bought multiplied by cost per impression) will increase. However, since each impression is less targeted, then all else equal the cost per impression should go down, despite CAC overall going up. However, since demand overall for digital ads is increasing, the cost per impression (price per ad) can still go up even as targeting gets worse.
Increase in Prices - Facebook
Given these ad targeting headwinds, it’s interesting that Facebook’s prices increased by 47% year-over-year last quarter. Without the ad-targeting headwinds, this increase would have been even higher. Eric Seufert of MobileDevMemo has estimated the impact at 7%. So assuming he’s right, without Apple’s privacy changes, Facebook’s average price per had would have increased by 54% (47% + 7%).
Of course, a large part of the 47% increase is Facebook lapping a quarter in which the average price per ad decreased by 21%, which results in a two-year annualized growth rate of 7.8%. For context, per Facebook’s 2019 10-K, Facebook’s average price per ad decreased by 5% in 2019 and increased by 13% in 2018 resulting in a two-year annualized growth rate of 3.6%. So the 7.8% growth over the past two years is quite a bit higher. Overall I think it’s safe to conclude that prices went up a lot year-over-year despite the privacy headwinds.
On the earnings call, Facebook’s CFO Dave Wehner explained the increase in ad prices as benefiting from “broad-based strength in advertiser demand” and the depressed prices in the prior year’s quarter. Wehner also indicated that the targeting headwinds were decreasing ad prices relative to where they would be otherwise:
When viewing growth on a two-year basis to exclude the impacts from lapping the Covid recovery, we expect year-over-two-year total revenue growth rates to decelerate modestly in the second half compared to the second quarter rate. We continue to expect increased ad targeting headwinds in 2021 from regulatory and platform changes, notably the recent iOS updates, which we expect to have a more significant impact in the third quarter compared to the second quarter.
In general, for the price of ads to increase, either the inventory of ads has to decrease, shifting the supply curve to the left (lower quantity available at each price) or the demand for ads has to increase, shifting the demand curve upwards (higher willingness to pay at each given price). For the inventory of digital ads to decrease, people would have to be spending less time online versus last year, which is possible due to the “reopening” from the pandemic. However, Facebook saw a 6% increase in ads delivered, meaning inventory increased. Therefore, it seems likely that the culprit of the increase in ad prices is an increase in demand.
Alternatively, for ad prices to decrease, either supply has to increase (people spend more time online) or demand for ads has to decrease. Facebook has explained the 21% decrease in prices in Q2 2020 as a result of both of this factors, with Wehner citing “more engagement related to the lockdowns” and Facebook’s 10-Q attributing most of the price decline to “a decrease in advertising demand globally due to the COVID-19 pandemic.”
However, for the current quarter and 2021, Facebook’s prices are increasing, and they would be increasing even more if not Apple’s targeting changes. The overwhelming reason is the secular shift towards digital advertising, with dollars flowing away from linear TV and other legacy forms of advertising and towards Facebook, YouTube, Snapchat, etc. This secular shift means the demand curve for Facebook advertising will continue to increase for many years.
Increase in Prices - Android
Because Apple’s targeting changes have made ads less effective on iOS, advertisers are apparently shifting budgets to Android.
Theoretically advertisers should have already been spending as much as possible on Android, as long as price < LTV. However, in reality budgets may have been constrained before this point. Or it’s possible the shift is just a reflection of test budgets and therefore temporary).
In any case, across advertising platforms on Android, ad prices are, at least temporarily, higher than they would be otherwise. Therefore, Apple’s targeting changes not only hurt advertisers on iOS, but also advertisers on Android.
Summary
For an individual company, if the average price of ads on a platform increases, the number of ads available with price < LTV will decrease. Therefore, the company will buy fewer ads and likely spend less overall on the advertising platform (total ad spend = ads bought * price paid per ad). Recent examples of companies reducing overall ad spend in response to platform-wide price increases are Root, Wish, and Zynga.
Apple’s targeting changes are making ads less effective, which increases CAC on iOS by increasing the number of impressions required to acquire a customer (CAC = impressions * cost per impression). Because Apple’s targeting changes make each impression less effective, digital ad prices (cost per impression) are lower than otherwise, which partially offsets the increase in impressions required.
However, the secular increase in demand for digital ads means digital ad prices overall are still going up for platforms such as Facebook. In addition, the worsened targeting on iOS has caused, at least temporarily, increased demand for ads on Android, making ad prices on Android higher than otherwise.