A finding should change the proposal
Campaign learning often arrives as a short sentence: a particular audience responded, a proof point mattered or a format underperformed. Feeding those sentences into AI is only useful if they change what it proposes.
For an illustrative example, evidence that buyers need to see the product in use should lead to a demonstration-led route. A fresh opening line attached to the same unsupported product promise would leave the learning unused. Ask the proposal to identify the source insight and the creative choice it changes.
Use the existing portfolio as context
The test supplies codified insights and a digest of existing creative. That combination matters. The insight suggests what might be valuable; the portfolio shows what has already been tried and where a new route would add coverage.
A gap is a place to investigate. It does not by itself justify production. A missing format may be missing because previous evidence argues against it, or because it does not suit the product. The model needs both the gap and the surrounding evidence to make a useful proposal.
Keep the evidence link in the handover
A practical proposal should make three things easy to inspect: the learning it uses, the element of the ad it changes and the response that would support or challenge the idea. This turns a list of concepts into a set of intelligible tests.
Benmark evaluates the proposal's commercial potential, creative innovation and suitability. Once a route is chosen, carry its reasoning into the brief so the production team can preserve what the test is meant to learn.
How Benmark tests this
Models receive brand context, codified campaign learning and a digest of the existing creative portfolio, then propose three next tests per prompt condition. The proposals are blind-scored for their commercial and creative quality.
Test inputs, instructions and scoring
What the model receives
The inputs combine brand guidelines for an invented brand, existing insight memory and a live-ad gap digest. Each condition asks for three concepts. This is a test of translating supplied learning into proposals; the model is not discovering that learning from a live account.
What changes between prompts
Both conditions receive the same brand context and line-specific inputs, with a 150-word concept limit. The simple prompt states the task. The scaffold adds judging criteria, an ad-quality framework and a hook/narrative/visual/conversion structure. It therefore tests that whole package of guidance, not a single wording change.
How the result is scored
Ben blind-scores commercial potential, creative innovation, and coherence/suitability on a 0–10 scale. A concept clears the test-worthy threshold only when every dimension reaches 8. The displayed mean summarises quality; it does not measure the ad's eventual conversion rate. Scoring hides model labels, although the scaffold's formatting can reveal the prompt condition.