What the report returns
RoastIQ is built to move the conversation from opinion to one clear decision while the ad is still editable.
Live Demo
Real-time scene detection, object recognition, and KPI scoring. The interactive walkthrough loads only when you get close to it.
The Science
A score only helps if you can see what drives it, what benchmark frame it sits inside, and what it does not claim. RoastIQ is built to make a scale, sharpen, or rebuild decision while the ad is still editable, not to pretend it has already measured market outcome.
How we score
RoastIQ scores in three layers: raw perception signals from visual attention and transcript analysis, sub-KPI families that group related signals, and the five main KPIs with a weighted composite. Each layer is visible so the team can trace exactly what drives the verdict.
Read the scoring breakdownHow benchmarks frame the verdict
Every RoastIQ score is read against platform-first, category-aware norms. A 72 on Instagram Reels for DTC skincare means something different than a 72 on YouTube for automotive. The benchmark frame keeps the verdict tied to the real launch conditions.
Read how benchmarks workWhat RoastIQ does not claim
Attribute detection is ~85% accurate, not 100%. No published correlation to in-market success yet. Heatmaps predict visual attention, not measured eye tracking. We publish these limits because the tool is only useful if the team trusts what it actually does.
Read the honesty pageThe five KPI families
The composite tells you the verdict. The individual scores tell you where the fix is.
Composite formula
Verdict
Sharpen
Composite: 71 · Sell Proposition dragging. Fix the value prop before scaling.
Brand and performance teams running creative diagnostics across beauty, ecommerce, and direct-response.
I manage paid media for Erborian across France, the UK, the US and Europe. Being able to check creative effectiveness in minutes — in English, in French, with the right regional benchmarks — lets me optimise campaigns daily instead of monthly.
Paid Media Online Manager, Erborian
Paris, France
I’ve bought Meta and Google ads for six years across my Shopify stores. SaliencyLab catches things I miss — brand drift, attention drops, CTA timing — at a fraction of the price of the legacy platforms. For an indie operator, that’s a no-brainer.
Founder, Timed Post
Morocco
Running L’Oréal Paris in Morocco means balancing global brand codes with local cultural nuance. SaliencyLab’s MENA-aware benchmark finally gives brand managers here the same creative intelligence that global teams in Paris have had for years.
Brand Director, L’Oréal Paris Morocco
Casablanca, Morocco
When to go deeper
Score is clear. Buyer objection still isn't?
If the team agrees the Sell Proposition is weak but can't agree which buyer is resisting and why — that's when BuyerLens opens. Not before.
FAQ
Method
How does the RoastIQ diagnostic work?
Accuracy
How accurate are the scores?
Benchmarks
What are the platform benchmarks based on?
Privacy
How is my creative data protected?
RoastIQ is an ad testing tool that uses multimodal AI to score your creative before you commit media spend. Instead of waiting days for manual creative reviews or weeks for consumer panel results, RoastIQ delivers a scored, benchmarked verdict in under 90 seconds. Upload an image or video ad, and the pipeline runs visual attention models, transcript analysis, brand cue detection, and platform benchmark context to produce five KPI scores and a single next-move recommendation.
Every ad is scored across five KPI families. Together, they form a weighted composite that determines the verdict: Scale, Sharpen, or Rebuild. Each KPI targets a different dimension of creative performance:
RoastIQ is built for performance marketers, creative strategists, and agency teams who need a fast pre-spend creative decision. DTC brands use it to test ad variants before committing budget. Agencies use it to pressure-test client creative before the media plan goes live. In-house teams use it to settle internal debates with a scored, benchmarked readout instead of opinion.
Traditional ad testing requires assembling consumer panels, waiting days or weeks for results, and spending thousands of dollars per test. Creative scoring AI like RoastIQ compresses that cycle to under 90 seconds at a fraction of the cost. The tradeoff is clear: RoastIQ scores are model predictions, not measured consumer responses. Use RoastIQ for the fast pre-spend filter. Reserve panel-based testing for the highest-stakes creative decisions where measured validation matters.
Every RoastIQ analysis ends with one of three verdicts. The verdict is not a subjective opinion; it is derived from the composite score and KPI thresholds:
If the score is clear but the team still debates which buyer is resisting, that is when BuyerLens opens. Not before. Start with the diagnostic, then go deeper only when needed. See ad benchmarks to understand how scores compare against platform and category norms.
