Creative analysis field guide
Turn creative evidence into a decision.
Use the creative, its brief and the available evidence to decide what to keep, change and test.
By Oussama Nakhil · Updated 28 September 2026
Evidence status
This guide describes a review workflow. It is not a new empirical study or proof of a universal score-to-performance relationship.
Start with the creative and the brief
Keep the original file, brand, audience, placement and intended decision together. Record missing audio, unknown delivery context or unavailable audience evidence before interpreting the result.
Review the five RoastIQ diagnostics: Beat the Skip, Get Noticed, Brand Impact, Sell Proposition and Build Brand. Each is model-based evidence about the creative, not a measured consumer outcome.
Check whether a verdict is supported
Use the saved evidence gate. Meta, unsupported cohorts and missing lineage remain Directional only; do not force every creative into Scale, Sharpen or Rebuild. An eligible recommendation is a review input, not an instruction to buy media.
Keep the named model, validation estimate, cohort coverage and run stability distinct. Outcome-model research does not automatically validate the five-KPI composite or every market.
Open the evidence behind the recommendation
Link each proposed change to the relevant scene, attribute or saved source. Heuristic attention timelines are not observed viewer retention, and incomplete media should not be described as fully analyzed.
When the buyer why is unresolved, a separate BuyerLens study can explore synthetic resistance themes. Its output must stay labelled synthetic, carry its panel provenance and remain separate from real customer research.
Preserve human choice and continuity
Record whether the reviewer accepts, adapts, rejects or defers the recommendation, along with their reason and proposed test. Preserve human edits through review and export.
Keep the original prediction and version when the revised creative or measured outcome arrives. A later result must not overwrite what was known when the decision was made. Adequate, permissioned evidence is needed before drawing calibration conclusions.
Continue the review
Check the formats, sample deliverables and procurement facts, or read the guide en français.
Use the Hook Timer to mark creative moments, read the methodology and evidence limits, or return to the industry review guides.