RoastIQ creative diagnosis

RoastIQ analyzes an ad across visual, audio, copy, brand, and platform context. It returns a positional diagnosis: where the creative appears to sit in the relevant distribution and why.

Use it before committing media budget, during creative development, or when choosing which variant deserves a live test. It is most useful when the analysis is attached to a specific decision rather than requested as a generic score.

The five KPIs

KPI	Question it helps answer
Beat the Skip	Does the opening earn enough attention to continue?
Get Noticed	Is the execution likely to register in a crowded feed?
Brand Impact	Is the attention clearly attached to the brand?
Sell Proposition	Is the offer or benefit understandable and persuasive?
Build Brand	Does the execution reinforce durable brand meaning and assets?

The dimensions interact. A visually striking opening can get noticed without transferring attention to the brand. A clear proposition can still underperform its communication role if it arrives too late. Read the profile across KPIs before optimizing one number in isolation.

What goes into the result

The analysis uses the uploaded creative, the metadata you confirm, available project context, and the relevant benchmark slice. A score is therefore not a universal grade that can be compared carelessly across every format or platform.

For video, the diagnostic can draw on changes across scenes, visible text, spoken or audio information, product and brand cues, pacing, and the relationship between the opening and the later payoff. For static creative, composition, hierarchy, legibility, brand attribution, proposition, and format context are especially important.

Prepare the creative

• Upload the version closest to the planned media execution.
• Keep captions, audio, end cards, and overlays if they will appear in market.
• Confirm platform, market, category, and format rather than relying on defaults.
• Add the relevant project context before analysis.
• If comparing variants, keep inputs consistent and label the files clearly.

Read evidence, not only scores

A useful diagnosis should connect a KPI to something observable in the asset: when a cue appears, what dominates the frame, whether the proposition is legible, or how brand and product information are linked. If a recommendation cannot be tied back to evidence in the creative, treat it as a weak basis for an edit.

Use the result in order

1. Read the verdict and summary.
2. Inspect the evidence supporting the strongest and weakest dimensions.
3. Check benchmark support and confidence.
4. Choose the single edit most likely to change the decision.
5. Re-test a meaningful variant rather than polishing without a hypothesis.

Turn a recommendation into a test

Write the proposed change in three parts:

1. Observed problem: what the current creative does.
2. Edit: the concrete change you will make.
3. Expected diagnostic effect: which communication problem the change should address.

Example: “The brand first appears after the opening demonstration. Add a distinctive brand cue within the opening without interrupting the product action, then check whether Brand Impact strengthens while Beat the Skip remains stable.”

Comparing variants

Compare variants as evidence about a hypothesis, not as a leaderboard. Small score differences can be less important than a material change in the diagnosis, evidence, or action verdict. Preserve the same project and comparison context, then validate consequential choices with a real-world experiment.

Modeled diagnosis: RoastIQ scores are model-derived. They should guide iteration and structured debate, not be presented as guaranteed campaign outcomes.

Next, read Understanding results for verdict, benchmark, and confidence guidance.