There’s ample discussion about Meta’s AI integrations in ad creation and campaign workflows. But we rarely get a glimpse into how those integrations and infusions are happening. Last week at Street Fight Live, we were given just that. Meta’s Brody Karmenzind presented all the ways AI enters the picture.
Before getting into those particulars, first a bit of context. AI is now the driving force behind ad creation, targeting, and optimization at Meta. It’s the reason the company has been able to accomplish a massive rebound in revenue, profitability, and correlative confidence on Wall Street (more on that in a bit).
Karmenzind further put things into perspective regarding Meta’s scale of AI operations. Among other things, its data centers that power AI have a footprint of 1,000 football fields worth of servers. This level of cost and compute capacity starts to signal the magnitude of its investment in AI, with more coming.
“AI is powering ad engagement because it brings people back,” said Karmenzind with a home-field advantage on the Street Fight Live stage at Meta’s Chicago offices.
Breaking it Down
With that backdrop, how is AI working its way into the campaign creation and management workflows in Meta’s products – primarily Facebook and Instagram? Karmenzind walked us through several of them in no particular order. We’ve broken it down below, with top takeaways for Localogy Insider readers…
Audience Automation: Meta now lets advertisers on Facebook and Instagram target audiences they might not have reached or even known about. They first enter or upload their primary audience (via first-party data or other inputs). Meta will then find adjacent audience groups that are likely to be receptive and performant. This is all done through its extensive interest graph.
Creative Automation: As we’ve examined, Meta (and a few others including Amazon, Google, and eBay) are bringing generative AI to ad creative. In Meta’s case, it offers tools to differentiate product images with things like custom backgrounds via text prompts (think: seasonal variations). It can also create image animation or transform still images into video so that they can work as Instagram Reels. This is all done to stand out in Facebook and Instagram’s feed-based UI with multimedia that’s “thumb-stopping.” It’s already done this for a million+ advertisers and 15 million+ ads. For example, Unilever changed its image backgrounds, which resulted in a 13 percent CTR boost.
Destination Automation: Where should ads drive people for maximum impact? Meta automates and optimizes calls to action. For local/physical businesses, that includes driving people to physical locations via Google and Apple Maps links. And when relevant, it drives users to AI customer service agents to answer questions. These are operational for 24 hours per-day and don’t take vacations. Karmenzid says this has cut down on customer service costs by 25 percent in pre-alpha trials.
Placement Automation: Meta also automates where ads show up to perform best. Use any ad format and Meta will take care of the rest, says Karmenzind. This includes the growing slate of options Meta offers: Instagram Reels, Facebook Shops, etc. By doing this, Karmenzind says that cost per acquisition goes down because the right users and contexts are targeted. And like all of the above, the goal is to boost performance while also automating processes to save time.
Budget Automation: You get the theme by now… things are automated. That applies to budget allocation as well. Advertisers can toggle this option to activate the most optimal allocation. “Just put in all your ads and let it rip,” said Karmenzind,
Lastly, it’s worth noting that these tools can be mixed & matched to taste. Advertisers can select none, some, or all of them. The most impact can be seen when they’re used in combination, said Karmenzind, which makes sense as they build on each other. The results can be up to 14x return on ad spend, he said.
Necessity Births Invention
Back to Meta’s operational and financial performance, you likely remember how the company’s revenues and stock price nosedived about two years ago. That happened as Apple dropped the hammer on behavioral tracking and other data collection in iOS… on which apps like Facebook and Instagram relied.
Forced to innovate its way around these restrictions, Meta got to work on finding alternatives. Fortunately, the timing was right as this coincided with recent inflections in AI and large-language models. This became the basis of its turnaround, up to and including all the above AI infusions.
Now it’s in a technical position where its ad-serving capabilities are in better shape than before Apple’s restrictions such as ATT. It’s a lesson in necessity being the mother of invention – something we also learned in the Covid era. Now, with AI’s impact glaringly evident, expect Meta to continue doubling down.
“When I started at Meta, AI was this distant thing,” said Karmenzind. “Now it’s built into everything we do.”
