How does the RoastIQ diagnostic work?

RoastIQ uses a multimodal AI pipeline combining visual attention models, OCR, transcript analysis, and platform benchmark context to score your creative across five KPIs: Beat the Skip, Get Noticed, Brand Impact, Sell Proposition, and Build Brand.

Do I need an audience panel to run an analysis?

No. RoastIQ generates diagnostics from the creative asset itself using AI analysis — no consumer panel or eye-tracking study required.

What are the platform benchmarks based on?

Benchmarks provide platform-first, category-aware norm context so scores can be interpreted relative to your specific format, region, and objective.

How is my creative data protected?

All uploads are encrypted in transit and at rest. Your creatives are private to your workspace and are never used to train shared models.

How is a synthetic panel different from asking a single chatbot?

SaliencyLab synthetic panels run multiple personas in one auditable workflow, then apply critic QA and synthesis. A single chat response cannot provide persona spread, confidence gating, or ranked cross-persona edits in the same way.

What data drives synthetic behavior in SaliencyLab?

Synthetic behavior is built from selected persona attributes, campaign context, platform objectives, and model-guided interview logic. Where proprietary enrichment is available, it can be incorporated as controlled context instead of blind prompting.

When should we use synthetic users vs real users?

Use synthetic panels first for speed, exploration, and pre-spend pressure-testing. Use real users for final validation, high-risk decisions, and cases where legal or market sensitivity requires direct human confirmation.

How should we interpret confidence gating?

Confidence gating indicates how reliable a synthetic output appears for action. High confidence supports immediate iteration decisions; low confidence means treat the signal as directional and validate with additional tests or real-user checks.

How is proprietary data handled in synthetic analysis?

Proprietary inputs are handled as controlled analysis context with access control and auditability. They are used to improve relevance for your workspace outputs, not as unrestricted public training artifacts.

What audit and provenance details are exposed in reports?

Reports retain run identifiers, confidence outcomes, model/version timestamps, evidence anchors, and interview-turn lineage so teams can review decisions, trace recommendations, and support internal governance.

NewDiagnostics, synthetic audience feedback, and benchmark context now work together in a single workflow.

Explore audience simulation

Creative Diagnostics

Five KPIs. One upload.
Under 15 minutes.

RoastIQ scores your ad across Beat the Skip, Get Noticed, Brand Impact, Sell Proposition, and Build Brand — with platform benchmark context for your format and category. No panel. No waiting weeks.

RoastIQ Workflow

RoastIQ is designed to feel like one continuous project flow.

Start in the workspace, upload the asset with just the details needed for benchmark matching, and move directly into the results surface. The Coca-Cola example below shows the kind of report structure the flow leads into.

Start in workspace Jump to example

Step 01

Pick the tool

Open RoastIQ from the workspace left rail and start a fresh analysis without leaving the app shell.

Step 02

Upload one asset

Drop a video or image, confirm market and platform details, and send it straight into analysis.

Step 03

Review results

Land in a results-led report with KPI scores, benchmark context, evidence, and export actions.

Featured Demo Creative

Coca-Cola Ad Walkthrough

Quadrant view: media, info, KPI diagnostics, and delivery decision.

Demo Example

9-box style view

YouTube 1:01

81th percentile • High confidence

Top Left · Video

Coca-Cola · Coca-Cola Zero SugarYouTube1:01

Top Right · Detailed KPI Scores

Get Noticed

Strong

Sell Proposition

Strong

Build Brand

Moderate

KPI Components

Get Noticed Components

Branding86

Enjoyment73

Av. % Ad Viewed84

Sell Proposition Components

Persuasion89

Clarity76

Build Brand Components

Meaningful77

Different75

Salient85

Additional Diagnosis

Hidden by default to keep the demo clean. Expand when you need deeper diagnostic signals.

Bottom Left · Ad Info + Overall KPI Results

BrandCoca-Cola

ProductCoca-Cola Zero Sugar

CampaignSummer Spark

PlatformYouTube

CountryUnited States

Test Date2026-02

Ad Length1:01

Norm Versionv2.4

Prompted Key MessageCoca-Cola Zero Sugar keeps the full Coca-Cola taste with zero sugar.

RoastIQ

Beat the Skip

Brand Impact

Prompted Key Msg

RoastIQ explainer: weighted decision score from Beat the Skip, Brand Impact, Get Noticed, Sell Proposition, and Build Brand.

Bottom Right · Delivery Decision

AD SUCCESSFULLY DELIVERS TARGET PERFORMANCE

Get Noticed →

Sell Proposition ↑

Build the Brand

The creative shows strong get-noticed performance (85) supported by branding (86) and ad viewed (84). Enjoyment (73) remains the main lever to improve opening resonance.

Proposition delivery is strong (89) with clear messaging (76), CTA strength (75), and prompted key message (35). The current gap to the strongest KPI is 12 points, so persuasion cues should surface earlier.

Brand impact is moderate (77). Overall classification is Scale (high-high), with scale-ready recommendation for this benchmark context.

Platform Capabilities

Three capabilities. One decision loop.

Diagnose the creative, stress-test the message with a synthetic audience, then turn the result into a channel-ready activation plan before you commit media budget.

Creative Diagnostics

Synthetic Panels

Activation Planning

Capability rail

Select the decision layer to inspect

Diagnostics

Know what's working before you spend

Upload your creative, add channel context, and get a scored diagnostic in under 15 minutes — with evidence for every verdict.

Five KPI scores covering attention, message clarity, brand cues, and platform fit.
Timeline view shows exactly where attention holds and where it drops.
One-page summary ready to share with creative, media, and leadership teams.

Five KPI scoresEvidence timelineShareable summary

Workspace preview

Evidence-linked

Creative diagnosticBenchmark matched for format and category

Ready to review

Timeline

0-3sHook strength

4-9sAttention dip

10-15sBrand cue

Decision matrix

Used by creative strategists, media planners, and brand teams making pre-launch decisions.

Workflow Journey

From upload to deployment in one operating loop

Four stages, one continuous loop. Upload a creative, run diagnostics, pressure-test with synthetic audiences, and export clear recommendations — all without switching tools.

Stage 01

2-3 min

Intake and Objective Framing

Upload creative, add channel objective, and lock test scope before any model run.

Creative scoped and ready to testChannel and objective lockedDiagnostic queued

Decision gate: Creative is ready for structured analysis

Stage 02

3-4 min

Evidence-Led Diagnostics

Map attention flow, message friction, and Skip x Impact positioning with transparent evidence anchors.

Scored evidence timelineKPI strengths and gapsHow you compare to benchmarks

Decision gate: Promote, revise, or hold the creative

Stage 03

10 min

Synthetic Audience Pressure Test

Run persona interviews across segments to surface how different audiences react to the same creative before committing media budget.

Persona spread across segmentsAuditable interview turnsRanked edit suggestions

Decision gate: Signals strong enough to act on

Stage 04

5 min

Export Recommendations

Package diagnostic findings and audience signals into channel-specific recommendations ready to share with your team or brief into production.

Channel action cardsPrioritised edit listShareable summary report

Decision gate: Findings ready to brief into the next creative cycle

Synthetic Audience Panel

Hear how three different buyers react before you buy the media.

Amira, Carlos, and Yuki are not generic prompts. Each persona holds a full psychographic profile, answers the same structured interview, and cites the moments that changed their reaction so the output is clear enough to act on.

Structured interview workspaceEvidence-linked response view

Ready to review

In-depth interview

Moment-level evidence

Brand memory is strong. Product relevance lands too late.

Amira gives you a useful read on whether your ad earns trust early enough to justify a premium brand position.

3s brand cueMusic shiftRoutine fitLate product payoff

What was the first thing that caught your attention in this ad?

The music change in the first two seconds pulled me in. I was not looking for it but it shifted the tone immediately. The visual did not confirm the brand until much later though.

Did the message feel relevant to your daily routine?

Partially. The lifestyle framing resonated but the product felt added on, not central. It would work better if the product solved something I already care about.

What would make this feel more premium instead of just polished?

Show me how the product earns its place in the story earlier. Premium for me is confidence with purpose, not just beautiful atmosphere.

10 min panel runFast enough to use before the media brief hardens.

Moment-linked feedbackHooks, drop-offs, and brand cues tied to exact beats.

Decision-ready outputsCreative moves and follow-up prompts, not just sentiment.

Platform-aware readingYouTube, Meta, and TikTok implications stay visible beside the interview.

Explore synthetic users

The rest of the suite

RoastIQ handles the diagnose step. Use the others when the question changes.

SaliencyLab works best as one decision system. Start with the product that matches the uncertainty in front of the team, then move to the next layer only when the decision calls for it.

Current tool

Live

RoastIQ

Know if the creative will land before you spend.

Current stepDiagnose

Best whenI have an ad

5 KPIsBenchmark contextDecision trace

Pressure-test

Live MVP

Synthetic Users

Hear how different buyers react before you buy the media.

36 personasStructured interviewsMoment-linked evidence

Use it when the creative looks close, but objections, channel fit, or segment tension still feel unresolved.

Explore Synthetic Users

Track the market

Planned

SalientDigital

Catch category drift before the brief gets stale.

Search driftTheme clustersBrief implications

Use it before the next brief when the real question is where the market is moving, not how one ad already performs.

Screen concept routes before funding full research.

Parallel reviewShortlist logicNext-brief output

Use it when too many routes are still alive and the room is arguing from taste instead of the same criteria.

View SalientConceptAI

Skip + Impact

Primary decision framework

Evidence Timeline

Frame-linked diagnostics

Confidence-Labeled

Benchmark provenance visible

Fast Iteration

Move from review to re-cut in one workflow

Actionable Output

Every suggestion ties to a measurable weak point

Trust Layer

Model version and confidence preserved in reports

Platform-Native Benchmarks

Provenance Metadata

Real Analysis Flow

Pricing

Start free. Upgrade when your team needs more analyses, audience panels, or collaboration.

Free

€0/month

Try a full diagnostic at no cost. Most teams upgrade within the first month.

3 RoastIQ analyses per month
Five KPI scores per creative
Benchmark context included
Report export (PDF)
Email support

Start Free

Pro

€99/month

For in-house teams and lean agencies

Unlimited RoastIQ analyses
Full synthetic audience panel (3/day)
Platform benchmark context
Activation plan per report
History and comparison view
Priority support

Start Pro

Enterprise

€299/month

For multi-brand agencies and large teams

Everything in Pro
Unlimited synthetic panel runs
Team workspace and collaboration
Custom benchmark categories
Dedicated account support
Volume and annual discounts

Talk to us

Free plan includes 3 analyses. No credit card required.

Creative Intelligence Glossary

Browse full glossary

Learn the language before you spend.

The fastest way to misread creative diagnostics is to skip the vocabulary. These are the six terms most likely to shape a go, hold, or edit decision before budget moves.

Build Brand Beat the Skip Sell Proposition Evidence and Diagnostics

Core metric 1Primary KPI

RoastIQ

creative readiness index

Composite score that summarizes creative quality and action readiness.

Why it matters: Gives teams one top-line number before drilling into drivers.

Core metric 2Primary KPI

Beat the Skip

opening attention signal

Measures whether the opening moments reduce early drop-off risk.

Why it matters: Protects paid reach by keeping viewers through the first seconds.

Core metric 3Primary KPI

Sell Proposition

message clarity metric

Primary persuasion KPI for value-message clarity and believability.

Why it matters: Shows whether people understand why the product is worth choosing.

Evidence layerEvidence

Evidence Timeline

scene-linked proof trail

Timestamped evidence linking scenes, transcript lines, and detected cues to claims.

Why it matters: Converts insights into concrete edit instructions instead of generic advice.

Benchmark layerBenchmark

Percentile Rank

relative market ranking

Relative performance ranking against benchmark peers.

Why it matters: Shows whether a creative is underperforming or outperforming market context.

Synthetic layerSynthetic

Synthetic Panel Confidence

simulation quality gate

Quality gate for simulated audience output based on specificity and persona adherence.

Why it matters: Prevents synthetic feedback from being used when interview quality is weak.

Frequently Asked Questions

Clarifications on synthetic methodology, confidence, and governance.

Synthetic Workflow Data Sources Method Choice Confidence Privacy Governance

Five KPIs. One upload.Under 15 minutes.

RoastIQ is designed to feel like one continuous project flow.

Pick the tool

Upload one asset

Review results

Coca-Cola Ad Walkthrough

Top Left · Video

Top Right · Detailed KPI Scores

Bottom Left · Ad Info + Overall KPI Results

Bottom Right · Delivery Decision

Three capabilities. One decision loop.

Know what's working before you spend

From upload to deployment in one operating loop

Intake and Objective Framing

Evidence-Led Diagnostics

Synthetic Audience Pressure Test

Export Recommendations

Hear how three different buyers react before you buy the media.

Brand memory is strong. Product relevance lands too late.

RoastIQ handles the diagnose step. Use the others when the question changes.

Fast Iteration

Actionable Output

Trust Layer

Pricing

Free

Pro

Enterprise

Upload one ad. Leave with a decision-ready diagnostic.

Learn the language before you spend.

RoastIQ

Beat the Skip

Sell Proposition

Evidence Timeline

Percentile Rank

Synthetic Panel Confidence

Frequently Asked Questions

Synthetic WorkflowHow is a synthetic panel different from asking a single chatbot?

Data SourcesWhat data drives synthetic behavior in SaliencyLab?

Method ChoiceWhen should we use synthetic users vs real users?

ConfidenceHow should we interpret confidence gating?

PrivacyHow is proprietary data handled in synthetic analysis?

GovernanceWhat audit and provenance details are exposed in reports?

Five KPIs. One upload.
Under 15 minutes.