Rescue Lab
Some of your losing ads never got a fair test.
Rescue Lab finds ads that were starved of delivery before they could prove anything, and suggests a capped, stop-loss test for the ones worth a second look.
01 What counts as a candidate
Too little delivery to judge fairly, not proof of a loss.
An ad qualifies when its lifetime spend sits under max($20, 10% of $3,465.72 median ad lifetime spend), it has fewer than 2,000 impressions, and fewer than 3 conversions. In the demo workspace that threshold works out to $346.57: every one of its 7 candidates spent less than that, combined lifetime spend $57.00. Northwind Sleep demo workspace, fictional data.
These ads received too little delivery to judge fairly: not enough spend, impressions or conversions to say whether they work. That does not make them losers.
Rescue Lab
A second chance test gives each one a dedicated, capped budget with a stop-loss, so it gets a real evaluation instead of staying buried by algorithmic pacing.
7
Candidates · demo workspace
$57.00
Lifetime spend · demo workspace
2
High conviction · demo workspace
Fig. 2 · Rescue Lab, the candidate stat row · Northwind Sleep demo workspace, fictional data
02 Never got anywhere near a fair budget
Never got anywhere near a fair budget.
Candidate threshold, $346.57
$9.59
$6.66
$7.79
$10.36
$5.23
$8.14
03 Scored against your own proven winners
Scored against your own proven winners, not a rule of thumb.
Each candidate is scored by Jaccard similarity of its creative DNA against the workspace’s own proven winners, combined with a 1 to 5 conviction rating you add: 60% similarity, 40% conviction. “the editorial gift-guide ad” carries 50% similar to winners and a conviction of 4, demo workspace.
| Ad | Lifetime spend | Impressions | Conversions | Similar to winners | Conviction |
|---|---|---|---|---|---|
| the editorial gift-guide ad | $9.59 | 689 | 0 | 50% similar to winners | 4 of 5 |
| the founder’s DIY note, as a screenshot | $6.66 | 490 | 0 | 33% similar to winners | 5 of 5 |
| the news-style award ad | $7.79 | 503 | 0 | 50% similar to winners | Not rated |
| the small-sample test ad | $10.36 | 798 | 1 | 33% similar to winners | Not rated |
| the raw office-nap photo | $5.23 | 390 | 1 | 33.3% similar to winners | Not rated |
| the relatable meme (partner steals the blanket) | $8.14 | 589 | 0 | 33.3% similar to winners | Not rated |
| the founder prototype, before and after | $9.23 | 680 | 0 | 33.3% similar to winners | Not rated |
04 A second chance, capped and stop-lossed
A second chance, capped and stop-lossed.
Rescue Lab suggests guardrails for the second-chance test rather than reopening full spend. Launching it is a separate, confirmed action, the same as every other experiment on SocialBuster.
Daily budget
Clamped to the account’s median daily spend, between $10 and $50.
Spend cap
The lesser of 5 times the daily budget or 10% of the account’s trailing 30 day spend.
Stop-loss
Triggers at 1.5 times the account’s own CPA or CPL, once at least 2 times that CPA or CPL has been spent.
Minimum delivery
At least 2,000 impressions and 5 conversions, lowered only when the spend cap genuinely cannot reach it.
Never launched automatically, and never claimed as causal proof: a similarity score is a lead worth testing, not evidence a style caused a win.
Creative Lab in detail →How the Strategist decides →Give your buried ads a fair test.
Connect a Meta ad account and SocialBuster syncs its campaigns, ads and daily results. Or load the Northwind Sleep demo workspace and look around first.