2. Meta Lattice: The Giant Library
Meta Lattice is our ad ranking architecture that allows us to generalize learnings across campaign objectives and surfaces in place of numerous, smaller ads models that have historically been optimized for individual objectives and surfaces. This is not only leading to increased efficiency as we operate fewer models, but also improving ad performance. These significantly larger models are able to learn more because they can represent steps that people take in their purchase journey across surfaces and campaigns. For a deeper understanding, read more here.
Said differently: Think of Meta Lattice as one giant library. In the past, we had many small libraries, each dedicated to a specific subject like history or art. Each library had its own set of books and librarians who were experts in that one subject. Now, with Meta Lattice, it’s like we’ve combined all those small libraries into one library, connecting information from all subjects. It doesn’t just have more books; the expert librarians can also learn from all the books it has and apply that knowledge to different subjects. This is much more efficient because we only maintain one library, and it can help people find better information faster, no matter what they’re looking for.
In the world of ads, this means we can use one powerful system to improve how ads are shown to people, making them more relevant and effective, rather than relying on many smaller systems that each focus on just one type of ad product or placement destination.
Results: Meta Lattice has increased ad quality by almost 12% and increased ad conversions by up to 6%.