Snap announced its Q4 2022 earnings yesterday: the company reported revenue of $1.3BN, below consensus analyst estimates of $1.31BN and representing just 0.15% growth over Q4 2021. Snap’s global DAU grew by 17% to 375MM, although DAU growth in North America registered 3% growth on a year-over-year basis, versus 4% last quarter and 5% in Q4 2021. Snap’s global ARPU decreased by nearly 15% on a year-over-year basis, with North American ARPU down by nearly 9%. The company’s stock was down 15% in after-hours trading following the earnings release.
The company revealed some interesting statistics in its shareholder letter:
- Snap’s Direct Response business grew by 4% on a year-over-year basis;
- Snap’s Brand Advertising business shrank by 11% on a year-over-year basis;
- Global CPM’s declined by 9% on a year-over-year basis.
While 4% growth in the direct response business is promising, the company’s DAU grew by 18%; the CPM decline explains that delta, as do the ARPU metrics graphed above. In the earnings call, Derek Andersen, Snap’s CFO, clarified that the company is “demand constrained and not supply constrained,” having seen 8% impression growth in Q4 (which is partly attributable to Spotlight, Snap’s short-form video product, but also the company’s DAU growth) against a 9% decrease in CPM.
Spiegel walked through a laundry list of initiatives that the company is pursuing in support of “re-architecting” the company’s advertising infrastructure:
- Conversion API;
- Data cleanrooms;
- Multi-party Computation;
- Pixel for eCommerce;
- In-app WebView Performance.
These are all fairly consequential and strategic projects; Spiegel spent time in the call discussing how these programs are scaling, and why it might take time for them to contribute meaningful improvements to advertiser efficiency (emphasis mine):
So overall, obviously, the results are early, but we’re excited about these changes we are making. And then, what we’re doing is, we’re taking our machine learning models, and we’re training them on these higher value conversions, which will hopefully help us scale the overall number of those conversions over time. But the net-net, the impact in the short-term is really, that advertisers are experiencing these higher value conversions, but there are fewer of them as our models retrain, and hopefully, as we progress through the quarter, we can improve and increase that volume overall. And obviously, in addition to larger advertisers, these changes really benefits smaller advertisers who are much more reliant on last click conversions for measurement, maybe because they haven’t implemented our conversions API, or don’t have a data cleanroom, for example.
I think, at a high level on the DR business, as I mentioned, the key here is that we’re really improving the overall value of those conversions. But, as a result of volume of those conversions has decreased as our models relearn on the conversions that we’re driving, and hopefully, obviously, we can expand that volume over time. But it’s also requiring advertisers to adapt, for example, so they need to see…that increased value show up in their third-party measurement tools, for example. And then, go in and increase their bids to reflect that increased value.
Jerry Hunter, Snap’s newly-appointed COO, also shared some clarifying insight on the work the company is doing to improve advertising efficiency:
Yes. Thanks, Derek. Brian, thanks for the question. Let me just give you a little about how we think about this data. So we have a bunch of ways that we’re collecting data. So conversion API, pixel, data cleanrooms. It’s like Evan talked about earlier, multi-party computation, SKAN and MMP. And all of these signals feed into our system and give us a better view of what’s happening with customers and conversions. Add to that the changes that we made to the WebView and to the ad format, so we get better signal about how our customers are interacting with our product. These all come together to train our ML. And that gives us better targeting over time. So the way we think about this is sort of a circle where there’s constantly information that’s coming in. We make changes in each of these products, as well as making changes — customers make changes to their campaigns, and then we make changes to the ad format, we feed it into the ML and this sort of circle gives us better and better targeting over time, which we think still leads to better CPMs across the board and better ROAS for our customers.
While the company did indicate that it expects to see a revenue decline in Q1 2023 on a year-over-year basis, the earnings call was reassuring. Snap had demurred and equivocated in past quarters regarding the source of its commercial ailments and the specific remedies it was pursuing to address them.
Perhaps as a result of the promotion of Jerry Hunter to the COO role, the company feels much more clear-eyed and resolute about what it needs to achieve in the near term to ameliorate the commercial obstructions of ATT. The advertising infrastructure projects that Evan Spiegel enumerated are perfectly appropriate for this juncture. This is encouraging; the ATT recession is a reality, and the most productive path forward runs not around but through it.