Entropy
ESStart on EPYC

Meta in 2026: what works, and what doesn't.

We tagged and scored more than 10,000 creatives from 32 brands, across 50 ad accounts, USD 15M+ of Meta spend and 17B+ impressions in twelve months. Every ad was judged only against other ads from the same brand, campaign and month. What survived is in the guide.

What the data says.

  1. 01
    There is no winning formula.

    No hook, offer, content style or production style improved sales across every brand that used it. Each one worked for some and failed for others. The guide shows which, and how often.

  2. 02
    One lesson travels. Only one.

    Exactly one creative decision came out positive in every run, at no extra cost per impression. Everything else depended on the brand. It is the first thing we would change in your account.

  3. 03
    Video is not free, and not always worth it.

    It costs more per impression and it buys more attention. Whether that trade pays depends on what you are asking the campaign to do, and in one of those jobs static still wins.

  4. 04
    A format's value depends on the funnel stage.

    The same format loses at one stage and wins at the next, and catalog ads flip outright. Judging a format on a blended number is how budget ends up in the wrong place.

Get the guide.

Eleven pages: the findings, the six decisions we would make from them, and a five-step worksheet for your next creative cycle.

We use this to send the guide and, if it looks relevant, to follow up once. No newsletter, no sequence. How we handle your data.

This came out of Meta Optimizer.

Every figure here was produced by the tool we run on client accounts. It tags every live ad by format, production style, content, hook and offer, scores each one against the account average, and writes the brief: what to make more of, what to stop. Point it at your accounts and you get this analysis on your own creative, every day.

See Meta Optimizer