Good morning,
Today is going to be a full week of posting, thanks in large part to the Apple-driven changes to the mobile advertising ecosystem manifesting themselves in a major way. Today I want to cover Snap’s earnings, which I suspect are a warning sign for Facebook’s earnings, which are coming this afternoon. Google and Twitter earnings are on Tuesday; both should be hurt less than Facebook or Snap. I will also be keeping a close eye on Amazon and Shopify earnings on Thursday (Apple’s changes benefit Amazon and hurt Shopify).
There is also the rumored news about a Facebook name change, which could be announced this afternoon during earnings, or given that it is expected to be focused on the Metaverse, at Facebook Connect this Thursday.
There is also, of course, the choreographed release of internal Facebook documents happening as well; it may be difficult to tease out cause-and-effect as far as impact on the Facebook stock price goes, but there is a reason why Stratechery has been laser-focused on Apple’s App Tracking Transparency (ATT) for over a year now.
On to the update:
Snap Earnings
From the Wall Street Journal:
Snap Inc. SNAP -26.59% said Thursday it expects growth to slow in the current quarter because of recent changes to Apple Inc.’s App Store privacy rules, sending Snap’s stock tumbling more than 20% in after-hours trading and highlighting the impact of the iPhone maker’s new policy on digital advertising.
The social-media company said revenue rose 57% for the three months ended Sept. 30, slightly below analysts’ expectations and its own guidance, but it expects expansion to slow. Daily users rose 23% to 306 million. Apple’s new rules have made it difficult for advertisers to test and measure their campaigns, according to Snap. Its share price fell to about $57 after closing at roughly $75. Other technology companies, including Facebook Inc., Twitter Inc. and Alphabet Inc., also fell in after-hours trading…
To cope with the problem, Snap said it is developing additional first-party tools to help its ad partners achieve their goals. But the company also attributed its holiday-quarter outlook to another complication—supply-chain and labor-market shortages—saying those are indirectly crimping its ad business by hurting business for its ad partners.
Snap, until these earnings, had been conspicuously blasé about Apple’s changes, with CEO Evan Spiegel publicly professing support for Apple’s moves; three things appeared to catch the company by surprise.
First, the ATT changes took a long time to roll out. Apple didn’t even launch iOS 14.5 until late April, but more importantly, it didn’t start pushing users to upgrade until June; that meant that as of last quarter’s earnings calls companies only had a few weeks of data to work with.
Secondly, Snap said that SKAdNetwork (SKAN), Apple’s API for tracking conversions, worked less well than they expected; from the earnings call:
As part of these changes, Apple rolled out SKAdNetwork, or SKAN, as a proprietary solution to allow app-based advertisers to continue measuring their advertising on iOS. The initial results we observed using SKAN were generally aligned with prior industry standard solutions, and we were among the first platforms to lean into this solution and push for widespread industry adoption. However, over time, we saw SKAN measurement results diverge meaningfully from the results we observed on other first and third-party measurement solutions, making SKAN unreliable as a standalone measurement solution.
Furthermore, as our advertising partners have explored and tested SKAN’s solutions, they have surfaced a variety of concerns about its limitations. Every advertiser has their own unique, fine-tuned perspective on the optimal parameters to measure ROI for their business, but SKAN requires them to use Apple’s fixed definitions of advertiser success. For example, advertisers are no longer able to understand the impact of their unique campaigns based on things like the time between viewing an ad and taking an action or the time spent viewing an ad. Additionally, real-time campaign and creative management is hindered by extended reporting delays, and advertisers are unable to target advertising based on whether or not people have already installed their app.
Third, Snap was — and remains! — confident that their ads work; Spiegel said:
When we look at incrementality testing, when we look at first-party data like on platform swipes or installs, things like that, we see that those conversions are still happening at similar rates that they did in the past. So, I think we can work through the tooling issues; it will take time because they’re new. But the underlying performance of the advertising platform is still very strong.
In short, iOS 14 hit later than expected, it impacted measurement worse than expected, but I think the biggest reason why Snap was surprised is that the company lost touch with how its direct response advertisers actually run their business.
Attribution and Targeting
One important — and expected — takeaway from Snap’s earnings is that brand advertising is doing fine; from the call:
Total revenue for Q3 was $1,067 million, an increase of 57 percent year-over-year. This was $3 million below our guidance range entering the quarter, which primarily reflects headwinds associated with iOS platform policy changes. Direct response advertising has comprised the majority of our business for some time now and has also been growing at relatively higher rates in recent quarters. Due to the impact of changes in the iOS ecosystem that affect optimization and measurement of direct response advertising objectives, the revenue from this portion of our business was approximately flat quarter- over-quarter. Our brand focused advertising business was the primary driver of sequential growth in Q3 and remained healthiest in North America during the quarter.
It is brand advertising that is subject to that famous John Wanamaker adage, “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.” This is because the payoff from brand advertising comes over time, whereas direct response advertising — as the name helpfully makes clear — is about an action taken directly in response to an ad. Naturally, the former is much less affected by ATT; that is why I expect Twitter’s results, for example, to suffer less than Snap’s, and Facebook’s to suffer more.
What is interesting, though, is that brand advertisers are generally much more interested in the sort of demographic and psychographic data that people generally associate with Facebook-style data collection; however, as I explained in Privacy Labels and Lookalike Audiences, what direct response advertisers really carry about is very straightforward: did you install the app, or buy the item. That is the data linchpin on which everything else turns.
What is being lost with ATT is knowledge of who actually made a purchase or installed an app, which at first blush seems like an attribution problem; the reality of direct response advertising, though, is that attribution and targeting are the same thing. If you don’t understand who bought an item or why you don’t know how to find the next person to target, or how. The way this manifests itself is in plummeting returns on advertising spend (ROAS); note this bit from the earnings call:
Average eCPM increased 62 percent year-over-year in Q3. Rising eCPM relative to the prior year reflects the year-over-year rise in overall demand, a mix shift towards relatively higher eCPM products, as well as a mix shift towards relatively higher eCPM regions such as North America. In addition, we have observed sequential and year-over-year increases in the cost per action for our goal-based bidding products, which we attribute to the loss of signals that our advertising partners have previously relied on to measure the impact of their campaigns. Adoption of our new measurement solutions, such as Advanced Conversions, is ongoing but will take time to be fully adopted. In addition, the ecosystem for measuring advertising returns is continuing to shift, with more changes anticipated as part of iOS 15 that are expected to further reduce the availability of certain signals that are currently used broadly as tools for optimization and measurement. Continuing to evolve and adapt our measurement solutions amid the rapidly shifting operating environment is a top priority as we seek to deliver attractive returns on advertising spend for our advertising partners over the long term.
“Goal-based bidding” lets advertisers spend in pursuit of a specified action, like an app install; the advertiser can specify the price they are willing to pay for an app install, which is based on their understanding of the lifetime value of a customer who installs their app. The better that Snap is at finding customers likely to install said app — which is best accomplished by knowing who previously installed the app — the fewer ads Snap needs to run to accomplish the advertiser’s goal; the fewer ads Snap needs to run to accomplish any one goal the more goals it can accomplish at ever lower prices, increasing ROAS for all advertisers.
If this market is functioning properly, rising eCPM is an indicator that advertising demand outstrips the supply of ad inventory (which has happened on Facebook the last couple of years); in this case, though, eCPMs are rising because Snap is having a hard time measuring whether or not it is accomplishing the advertiser’s goals, so it is using far more ads to do so. In other words, instead of more advertisers clamoring over a limited supply of ad inventory and thus driving up prices, Snap itself is gobbling up ad inventory to deliver on goal-based bidding, driving up the cost for everyone, reducing ROAS.
This is just one symptom of the lack of visibility for direct response advertisers, whose entire business is predicated on understanding to the cent what their ROAS is; the response now will be to pull back spending, reducing demand for ads (which will shift the mix to brand advertisers); this is the exact inverse of what happened when COVID hit, when direct response advertisers snapped up advertising inventory that was cheaper because of the exit of brand advertisers. And, in the long run, the entire direct response ecosystem will, in effect, have to get much better at identifying which ads works and which ones don’t by expanding the link between attribution and targeting to multivariate analysis involving campaigns, creative, SKAN, Snap’s Advanced Conversions tool, etc. It’s going to be hard work, and it’s going to be take time, and Snap’s surprise suggests they’re off to a later start than they should be.
The Supply Chain
ATT wasn’t the only headwind; from the call:
Separate from these iOS-related issues, we have heard from advertising partners across a wide variety of industries and geographies that they are facing headwinds in their business related to disruptions in global supply chains as well as labor shortages and increasing costs. In turn, we expect this to impact advertising demand in Q4 in particular as, in many cases, their businesses do not have the inventory or operational capacity to support incremental demand. We expect that some of these clients may opt to slow their marketing spend given the diminished need to drive incremental demand at a time when their supply chains are not able to operate at peak capacity.
The supply chains for a lot of the direct-to-consumer products that advertise on platforms like Snapchat are not dissimilar to direct response advertising: price-dependent and just-in-time. D2C companies are significantly less likely to have long-term contracts with shipping companies for guaranteed container space; that means they are reliant on spot pricing which, given the fact that there are few if any containers available, have been sky-high. That’s even before getting to the question of port congestion, which is introducing delays in products actually arriving.
The silver lining for these companies — at least those that aren’t drowning in debt from recent roll-ups — is that direct response marketing can be turned on-and-off; right now, more and more of them are turning it off, which means there are that many fewer companies competing for Snap’s ads, hurting pricing and thus revenue. Non physical-good verticals like travel, streaming, and fintech are helping make up the difference, but then again, those are some of the verticals that are now much harder to measure.
The question for today is what happens with Facebook; the company already warned advertisers/investors that it was having a hard time measuring conversions, but the company also seemed to get started on its ATT preparation earlier and more aggressively than Snap did. At the same time, Snap still has a very strong brand component to its advertising mix; Facebook is almost all direct response. My bet is that the latter factor matters more; building around ATT, whether that be through new ways of measuring, or moving conversions onto Facebook itself, is going to take years, not months. Quarterly results, though, have to be delivered in nine hours or so.
This Update will be available as a podcast later today. To receive it in your podcast player, visit Stratechery.
The Stratechery Update is intended for a single recipient, but occasional forwarding is totally fine! If you would like to order multiple subscriptions for your team with a group discount (minimum 5), please contact me directly.
Thanks for being a subscriber, and have a great day!