Six questions every Meta account should be able to answer.

03 · Four more questions, one index

The same evidence, answering four more questions.

1. ---

   03

   What should I test next?

   ShowHide

   The AI Strategist runs nine rule checks over the account's own numbers and ranks what it finds: a fading winner ahead of a minor edge. A claim needs at least $100 of spend and 5 conversions to compare a creative attribute, or $20 of spend and 1,000 impressions in a week to judge a single ad's trend. Below that bar the card reads “not enough data yet”, with the real counts, instead of a guess dressed as insight.

   How the Strategist decides →

   03

   What should I test next?

   The AI Strategist runs nine rule checks over the account's own numbers and ranks what it finds: a fading winner ahead of a minor edge. A claim needs at least $100 of spend and 5 conversions to compare a creative attribute, or $20 of spend and 1,000 impressions in a week to judge a single ad's trend. Below that bar the card reads “not enough data yet”, with the real counts, instead of a guess dressed as insight.

   How the Strategist decides →

2. ---

   04

   Which apparently losing ads never received a fair test?

   ShowHide

   Rescue Lab finds ads whose lifetime spend stayed under $346.57 (the greater of $20 or 10% of the account's median ad spend), reached fewer than 2,000 impressions, or produced fewer than 3 conversions. In the demo workspace, 7 ads clear that bar on $57.00 of combined lifetime spend, and 2 carry a high conviction rating. “the founder’s DIY note, as a screenshot” spent $6.66 across 490 impressions and shares a third of its creative DNA with the account's proven winners. That does not make these ads losers: a second-chance test gives each one a capped budget and a stop-loss, instead of leaving it buried by algorithmic delivery.

   Rescue Lab →

   04

   Which apparently losing ads never received a fair test?

   Rescue Lab finds ads whose lifetime spend stayed under $346.57 (the greater of $20 or 10% of the account's median ad spend), reached fewer than 2,000 impressions, or produced fewer than 3 conversions. In the demo workspace, 7 ads clear that bar on $57.00 of combined lifetime spend, and 2 carry a high conviction rating. “the founder’s DIY note, as a screenshot” spent $6.66 across 490 impressions and shares a third of its creative DNA with the account's proven winners. That does not make these ads losers: a second-chance test gives each one a capped budget and a stop-loss, instead of leaving it buried by algorithmic delivery.

   Rescue Lab →

3. ---

   05

   Which winning ads can be revived?

   ShowHide

   Winner Revival separates an ad's own decline from a shift across the whole account. The demo workspace's founder-story raw photo, “The night before we shipped the first batch,” run under Maya Chen, the demo brand's fictional founder persona, fell 12% in ROAS (from 4.19x to 3.69x) over the last 7 days against the prior 7, while the account's own ROAS rose 3% over the same window, so the decline reads as the ad's own. A revival plan tests new imagery first, then the headline, then the body, and keeps the original running as the control at every stage.

   Winner Revival →

   05

   Which winning ads can be revived?

   Winner Revival separates an ad's own decline from a shift across the whole account. The demo workspace's founder-story raw photo, “The night before we shipped the first batch,” run under Maya Chen, the demo brand's fictional founder persona, fell 12% in ROAS (from 4.19x to 3.69x) over the last 7 days against the prior 7, while the account's own ROAS rose 3% over the same window, so the decline reads as the ad's own. A revival plan tests new imagery first, then the headline, then the body, and keeps the original running as the control at every stage.

   Winner Revival →

4. ---

   06

   Which hooks, stories, images, CTAs, identities and formats keep generating revenue?

   ShowHide

   Creative Lab classifies every ad by style, hook, angle, funnel stage and production zone, then measures which of those traits carried the spend. Raw executions return 5.47x ROAS in the demo workspace against 3.44x for middle production and 3.92x for high production, at a 36% lower CPA than middle production. The Creative Leaderboard ranks the specific headlines, images and CTAs behind those numbers: one headline, “Finally, a weighted blanket that does not trap heat,” carries 5 ads and an adjusted ROAS of 8.72x over the last 30 days. Its coverage mode is co-occurrence, not attribution: an ad using several winning elements counts toward every one of them.

   Creative Lab in detail →

   06

   Which hooks, stories, images, CTAs, identities and formats keep generating revenue?

   Creative Lab classifies every ad by style, hook, angle, funnel stage and production zone, then measures which of those traits carried the spend. Raw executions return 5.47x ROAS in the demo workspace against 3.44x for middle production and 3.92x for high production, at a 36% lower CPA than middle production. The Creative Leaderboard ranks the specific headlines, images and CTAs behind those numbers: one headline, “Finally, a weighted blanket that does not trap heat,” carries 5 ads and an adjusted ROAS of 8.72x over the last 30 days. Its coverage mode is co-occurrence, not attribution: an ad using several winning elements counts toward every one of them.

   Creative Lab in detail →

---

06 · Connections, and what it will not do