Meta’s Andromeda is live: Here’s what it means for your ad data

Meta’s Andromeda update has now been rolled out globally, and it’s one of the biggest shifts in how ads are chosen and delivered since the introduction of Advantage+.

Andromeda’s update redefines how Meta learns from the on-platform and off-platform signals it receives, making ad delivery faster, more relevant, and more efficient.

The richness and consistency of the tracking data you send into Meta is now more important than ever.

Let’s dive into what this means for your campaigns.

What Andromeda changes

At its core, Andromeda is a new retrieval engine - the part of Meta’s ad system that decides which few ads (out of millions) to show a person.

In the past, this process relied on a mix of targeting inputs and lookalike signals. Andromeda replaces much of that with a deep-learning model trained to predict who’s most likely to convert - before they even click.

Instead of “find people like X,” Meta now runs on “find people who will act like X.”

It’s a subtle but significant change. The retrieval model constantly updates as it sees new behaviour across Meta’s apps, getting better at connecting people with the right products.

Why it matters for marketers

Feedback quality has become everything. Andromeda’s model learns from the data it receives - purchases, add to cart, view-content, leads etc. If that conversion data is missing or incomplete, the algorithm learns more slowly - or learns the wrong lessons entirely.

In practice, that means:

• Poorer targeting and slower campaign learning

• Less personalised and poorer performing retargeting

• Inefficient optimisation, wasted ad spend and less profitable Meta ad spend

Creative signals play a bigger role too. Because Andromeda’s retrieval engine learns which visuals and messages drive conversion, variety and clarity in your ad creative help it learn faster. The more distinct, high-quality creative inputs Meta has to test against, the better its predictions become.

However, without the right conversion signals being sent back to your creative, that learning is limited - your campaigns will always underperform.

Meta’s shift to predictive targeting makes data quality an even more crucial performance lever. The brands that benefit most will be those with rich, consistent conversion data feeding back into the system.

That’s why we’re seeing a growing focus on conversion tracking accuracy and session enrichment. The question is no longer just “is tracking working?” - now it’s “is our data helping Meta learn the right things?”

Meta’s AI era

Andromeda is the clearest signal yet that Meta’s ad platform is fully entering its AI era.

Advertisers don’t control the levers in quite the same way, but they do control the quality of data the system learns from.

That’s where the biggest wins will come from in the months ahead: rich, reliable conversion tracking signals, trustworthy measurement, and a feedback loop that helps Meta’s algorithms see your business clearly.

Solutions like Leaf Signal help brands maintain that foundation by ensuring that customer behaviour data is fully and accurately captured, and consistently fed back into Meta’s systems.

The future of ad performance isn’t just about targeting smarter - it’s about teaching smarter.