Try Wilow Today
Get Started Free
Learn
Meta's algorithm reads every ad across six dimensions: Persona, Offer, Surface, Type, Emotion, and Readiness. Here is how to use the POSTER framework to audit your own ads.
Meta's ad algorithm reads every ad you upload across six dimensions before it ever shows that ad to anyone. Those six dimensions spell POSTER: Persona, Offer, Surface, Type, Emotion, and Readiness. Meta builds this picture privately and never shows it to you. This guide walks through each dimension so you can read your own ads the same way the algorithm does.
When you upload an ad to Facebook or Instagram, Meta's GEM system reads it in full: the text, the images, the audio, the video. It works out what the ad is and who it suits, writes a summary, and stores it for retrieval. Later, when someone opens the app, Meta's Andromeda engine scans those summaries and narrows tens of millions of candidates to a few thousand for that person. Then GEM reads the shortlist in detail, ranking each ad on six aspects before the auction picks the winner.
Meta has publicly described GEM as a generative model that understands ads the way ChatGPT understands language. It has described Andromeda as a retrieval system built to find relevant ads at the scale of billions of users. And it has disclosed that it tracks more than 150 emotion categories, naming humour and nostalgia as two examples. But none of this is visible in Ads Manager. There is no screen that tells you Meta has tagged your ad as humour for busy parents or nostalgia for athletes.
That is the gap POSTER closes. Each letter is a dimension the algorithm reads. Understanding all six turns the system from a black box into something you can work with.
Meta's system does not just look at your targeting settings. It reads your creative to determine who the ad speaks to. A product shot of work boots with a warehouse in the background signals a different person than the same boots on a hiking trail. A testimonial from a founder in a suit signals a different person than a UGC clip filmed on a phone.
The practical question: if someone who has never heard of your brand saw this ad, would they immediately know it was meant for them? Every visual choice, every word, every voiceover accent is a signal Meta reads to file your ad under a persona. When an account runs ads that all signal the same persona, Andromeda only retrieves from one shelf, however many ads are sitting there.
The offer is the message or value proposition you use to convince someone. It is not your discount or your price. It is the reason the ad gives for acting now.
Meta reads this from your headline, primary text, and visuals together. One ad might lead with social proof ("join 10,000 teams"). Another might lead with a benefit ("ship orders twice as fast"). Another might lead with a problem ("still tracking ad spend in spreadsheets?"). Each frames the same product differently, and each signals a different reason to act. Meta treats these as distinct offers, not variations of one.
Most accounts run one or two offers across every ad. That means the algorithm sees one or two reasons to show your ad to anyone. Running five offers means your ad can be the right match for five different kinds of motivation.
Surface covers the format your ad takes and the placement it appears in. A square static image in Feed is a different surface from a vertical Reel, even if they carry the same message. A carousel is different from a single image. A Collection ad that opens into a fullscreen experience is different from a link ad that sends someone to a website.
Meta reads each surface as a distinct creative experience because each one performs differently for different people. Some people scroll Reels and never stop on static images. Others read carousel captions. Running multiple surfaces gives Andromeda more formats to retrieve for each person, rather than hoping one format works for everyone.
Type is what kind of ad it is: a product shot, a testimonial, a demo, a founder story, a user-generated clip, a competitor comparison, a how-to. Meta reads these as fundamentally different creative experiences and groups them separately.
An account that runs nothing but product shots is invisible to the person who only responds to social proof. An account that runs nothing but UGC is invisible to the person who wants to see the product in detail. Type diversity is one of the cheapest ways to widen reach because different types cost the same to produce but open access to different audiences.
Meta has publicly stated that it works with more than 150 emotion categories to decide which ads match which people. Humour and nostalgia are two examples Meta has named directly. Others include urgency, curiosity, aspiration, belonging, and surprise.
Most advertisers never think about emotion at all. They write copy around features and leave the emotional register to chance. But Meta is measuring this whether you intend it or not. An ad that opens with a joke signals humour. An ad that opens with a before-and-after signals aspiration. An ad that opens with a countdown signals urgency. Each gets retrieved for a different emotional profile.
The question to ask of every ad: what does someone feel in the first three seconds? If the answer is "nothing in particular," Meta is picking an emotion for you, and it may not be the one you want.
Readiness is the awareness stage your ad is built for. It ranges from people who have never heard of your brand or even your category (problem-unaware) to people who know your brand and are deciding whether to buy (solution-aware, brand-aware).
Meta reads readiness from how your ad is framed. An ad that explains what the product category even is signals low-readiness. An ad that compares your product to a competitor signals high-readiness. An ad that opens with "the best alternative to X" assumes the viewer already knows X. An ad that opens with "there is a faster way to do Y" assumes the viewer already feels the pain of doing Y the slow way.
Different ads built for different readiness stages mean your account can meet someone wherever they are in the buying process, instead of only speaking to the people who are already about to convert.
Open your Ads Manager and list your active ads. For each one, ask the six questions:
Now look at the distribution. If eight of your top ten ads have the same persona and the same offer, your account is concentrated. Meta's Andromeda engine is retrieving from a single shelf. The fix is not to make one ad better. It is to make ads that land in different parts of the POSTER grid, so Andromeda has more shelves to pull from.
This is what creative diversity means in practice. It is not about running more ads. It is about running ads that are genuinely different from each other across these six dimensions. Our guide on how to measure and improve creative diversity covers the workflow that keeps that spread wide.
Meta builds this whole picture behind the scenes and never shows it to you. Wilow rebuilds it from the outside. AI Creative Tagging labels every ad in your account across these same six dimensions, mirroring the read Meta's system performs privately. The Creative Leaderboard then ranks every ad by performance so you can filter by tag and see your distribution in one view.
The Ad Library tool surfaces POSTER diversity for any brand you study. And the Weekly Competitive Report maps your competitors' creative across the same six dimensions, so you can see which parts of the grid they own and which are open for you.
Creative is the one lever the advertiser fully controls. Covering the POSTER space well, across all six dimensions, is the edge over advertisers who keep making one ad and hoping it works for everyone. Wilow measures that coverage so you can see the gaps before the algorithm makes the point for you in rising costs.
Start free, no credit card required. Connect an ad account at trywilow.com and see the profile the algorithm has built for your account.
For the full theory behind the three steps Meta runs on every ad, read our guide on how the Meta ads algorithm works. For the deeper engineering dive, see the original breakdown on The Keyword: how the Meta algorithm decides which ads you see.