Meta announced Q4 2022 earnings yesterday: it beat analyst estimates on revenue, generating $32.2BN in revenue versus consensus of $31.5 billion. The company also registered 4% growth in global DAU for the Facebook Blue app, reaching 2BN DAU for the first time and the Family of Apps Daily Active People (DAP) metric nearing 3BN. Ad impressions increased across the family of apps by 23% on a year-over-year basis, although the average price per ad decreased by 22% on the same timeline. The company also stated that it will engage in a $40BN stock buyback. Meta’s stock is up nearly 19% in after-hours trading at the time that I write this.
On December 19th, perhaps at the height of the holiday shopping season, I posited on Twitter that, were Meta to beat analyst expectations on revenue (which, at the time, predicted a 6.5% decline), it would be because Meta’s Advantage+ advertising optimization engine, paired with its recently-reintroduced modeled 28-click attribution reporting (which was removed when ATT went into effect), delivered unforeseen efficiency gains for advertisers.
Meta speaks frequently about its commercial applications of AI, but I think they are generally misunderstood. To my mind, Meta is developing and leveraging its AI / machine learning infrastructure in pursuit of two goals:
- To boost engagement by transitioning the core product experiences across both the Facebook Blue and Instagram apps to short-form video content, propagated across an open graph. I discuss this in more detail in Unpacking Meta’s pivot to an open graph and short-form video;
- To boost advertising campaign efficiency by transitioning its media buying tools to automated, machine-learning-driven, “black box” systems like Advantage+. I discuss this in more detail in Google’s PMax, Meta’s Advantage+, and the logic of total advertising automation.
I believe that both of these initiatives were either initiated or accelerated as a result of the App Tracking Transparency Recession: increased engagement is required because CPMs deteriorated as a result of ATT’s adverse impact on targeting efficiency, and automation efficiency is required for the same reason. Regarding the ongoing revenue drag presented by ATT as well as Meta’s strategy for mitigating that drag, Susan Li, Meta’s recently-appointed CFO, said on the earnings call:
On ATT, I think what I would say is there is still certainly an absolute headwind to our revenue number. That is the impact of the ATT changes being in place. Having said that, we are lapping its rollout and adoption and we are making progress in mitigating the impact due to a lot of the work that both Javi and I just talked about, including the different advertiser tools, including ad formats that bring conversions on-site and including the longer term AI investments in privacy-enhancing technologies.
Reels, which is Meta’s short-form video product, is clearly boosting engagement — albeit at the cost of CPM. But Advantage+ is likely doing a better job of matching advertiser ads with the right impressions (of which there are more as a result of Reels), increasing efficiency. Note that in the earnings call, Javier Olivan, Meta’s COO, also addressed some of its mechanisms for collecting additional conversion signals, such as the Conversions API and on-site conversion observation (a concept that I have termed the Content Fortress):
I think if you look at the strategy on ads, we really have two parts, which is continue investing in AI and that’s where we are seeing a lot of the improvement in ads relevance. And as Mark was saying, we saw over 20% more conversions than in the prior year, which, combined with the decline in cost per acquisition, results on high ROIs. We also use it for automated experiences for the advertisers, improvements on measurement, which allow advertisers to do better decisions. But the second part is bringing more conversions onsite, which are also obviously helping offsetting this Signal loss.
As I describe in the aforementioned piece about Meta’s transition to an open graph, four non-mutually exclusive opportunities exist for a problem to increase advertising revenue:
- Increase “ad load,” or the ratio of ads shown to each user per session relative to organic content;
- Increase reach, or the number of users that engage with a product and thus are exposed to ads;
- Increase the value generated by ads through higher-quality formats or better targeting, which improves the general price paid for ad inventory through increased bids from advertisers in the ad auction;
- Increase time spent on site, which provides more opportunities for ads to be served.
Reels achieves number four on this list; Advantage+ achieves number three by reducing the waste of manual human oversight (and also, potentially, by maximizing total per-advertiser spend).
So despite Meta’s year-over-year revenue growth declining fairly substantially, its strategy with AI across both consumer-facing and advertiser-facing products is clearly working. And this, combined with a generous stock repurchase, led markets to rejoice.
And in addition to beating guidance and seeing its share price pop, Meta has another reason to celebrate: the FTC’s request for an injunction to block the company from acquiring Within Unlimited, the developer of a VR fitness app, was rejected by a judge yesterday.
Edit February 3, 2023: I had mistakenly claimed that Family of Apps reached 2BN DAU, when it was the Facebook Blue app that reached 2BN DAU in Q4 2022. The Daily Active People (DAP) metric for Meta’s Family of Apps stood at nearly 3BN in Q4 2022. This has been fixed.