By Ryan Richardson · Published 8 October 2026
A scoring pass catches weak ads before they burn budget, rather than after. The rubric is applied to every ad in a pool, not just the ones someone has a gut feeling about, and a pool needs a minimum spread across hook angles and audience-scene variants before it's considered ready, so one strong ad can't hide five weak ones.
Each dimension is scored 1, 3 or 5 (with those as anchor points, not a strict 1-5 scale):
Activity specificity: does the ad name a specific commitment moment (an invoice, a quote, a ticket) with a specific cadence, or does it describe a problem in general terms?
Emotional register match: does the copy's emotional temperature match what the real audience actually expresses, rather than a softened, more polite version of it?
Verbatim alignment: does the copy use 2 or more phrases that trace back to real quotes from the mined corpus, or is it tone-matched guesswork?
Forensic specificity: does the copy carry specific numbers, cadences and named failure shapes, or does it generalise on the details that would make it feel real?
Offer clarity: is the price stated, and is the value-to-price ratio made explicit (a specific dollar comparison), or is the price vague or absent?
Disqualification power: does the copy filter out people who don't have this exact problem within the first sentence, or could a much broader audience read it and feel mildly interested?
25-30 out of 30: launch immediately, at a higher initial budget allocation than the rest of the pool. 18-24: launch as part of the pool, monitor closely. 12-17: don't launch yet; identify the single lowest-scoring dimension and fix that specifically before it runs. Below 12: don't launch at all; the fix isn't a copy edit, it's going back to the research step and re-mining the corpus.
The most common failure the rubric catches is an ad that reads well in isolation but would be indistinguishable from a competitor's ad with the brand name swapped out. An ad that scores low specifically on disqualification power is usually the one that feels safest to write and performs the most generically once it's running.
| Claim | Value | Source |
|---|---|---|
| Number of scoring dimensions | 6 | Copy and creative.md, 'Step 5: 6 dimension scoring rubric' section |
| Maximum possible score | 30 (6 dimensions x 1-5 each) | Copy and creative.md, 'Step 5: 6 dimension scoring rubric' section |
| Launch-immediately threshold | 25-30 | Copy and creative.md, 'Scoring thresholds' section |
| Launch-and-monitor threshold | 18-24 | Copy and creative.md, 'Scoring thresholds' section |
| Rewrite-before-launch threshold | 12-17 | Copy and creative.md, 'Scoring thresholds' section |
| Do-not-launch threshold | below 12 | Copy and creative.md, 'Scoring thresholds' section |
| Minimum distinct hook angles required in a pool before launch | 3 | Copy and creative.md, 'Verification' checklist |
| Minimum distinct role-scene variants required in a pool before launch | 2 | Copy and creative.md, 'Verification' checklist |