AI ad testing
Kantar LINK AI
AI-based creative assessment. Kantar also offers separate research with respondents; LINK AI does not field a new recall survey for each asset.
Ask before buying: Which product and evidence are included in the proposal?
Creative testing, compared
Choose your next creative test.
AI diagnostics, consumer research or expert review? Start with the decision you need to make.
By Oussama Nakhil, SaliencyLab
Reviewed 7 October 2026. Official product sources linked below.
Different tools. Different evidence.
Compare the exact product and scope. One provider may offer several kinds of research.
AI screening or consumer research
System1 Test Your Ad
Test Your Ad Screen provides AI screening. Test Your Ad Pro measures emotional responses with real consumers. Choose the product that fits your brief.
Ask before buying: Are we commissioning Screen or Pro, and for which format?
Consumer research or AI screening
Zappi
Zappi offers consumer advertising research and Amplify AI, which uses predictions and synthetic respondents. Confirm the product, audience and evidence.
Ask before buying: Which audience, exposure conditions and norms will be used?
Qualitative research
Focus groups
Facilitated discussion with real people. Explore unexpected language, reactions and cultural interpretation that model diagnostics cannot establish.
Ask before buying: Who will we recruit, and what do we need to understand?
Written by SaliencyLab using the linked provider descriptions. This is not an independent head-to-head study. Request current pricing, timing and sample details for your brief.
For small DTC teams
How to test Meta and TikTok creatives before launch.
Choose a method for the uncertainty you need to resolve. SaliencyLab helps with the next edit; real-person feedback helps with comprehension and preferences. Neither replaces a controlled campaign test when the question is performance. You do not need to buy every method or follow a fixed vendor sequence.
What should we change before launch?
SaliencyLab: creative diagnostics and optional human review
Five-KPI diagnosis, predicted attention, revision priorities and synthetic resistance hypotheses where needed.
Boundary: Meta diagnostics are directional. No new human sample, sales forecast or proven campaign uplift.
Service scope and software pricingWhich option do people prefer, and why?
PickFu: audience surveys
Feedback from respondents on the alternatives or questions you supply. Pricing starts at $1 per response; audience, complexity and sample affect the total.
Boundary: A preference in a survey does not establish conversion performance. Check who was recruited and how the creative was shown.
PickFu pricing and audience scopeWhich design communicates most clearly?
Lyssna: preference testing
Participant choices between designs, with follow-up questions to understand their reasoning.
Boundary: Match the stimulus to the question. A design preference test is not a platform-native measure of video effectiveness.
Lyssna preference testingWhat happened after the ads ran?
Motion: creative performance analysis
Creative reporting for Meta and TikTok using campaign metrics to inspect hooks, viewing, clicks and conversion.
Boundary: Needs campaign data. Observational differences can reflect audience, spend and delivery as well as the creative.
Motion metric definitionsProvider descriptions checked against the linked official sources on 7 October 2026. This is SaliencyLab’s editorial guide, not independent testing, a complete market ranking or an integration list. Confirm current pricing and study scope.
A useful brief starts with one uncertainty.
For example, a skincare team may be unsure whether viewers understand the product benefit before the offer appears. Keep the original asset, identify the exact passage and propose one revision. Ask real target buyers what they understood if comprehension is the decision. A synthetic answer can suggest a question; it cannot stand in for their responses.
If the decision is whether the revision improves conversion, predefine the outcome, allocation, duration and stopping rule for a campaign experiment. Keep the original diagnosis and the observed result separate, including a null or inconclusive result. TikTok’s split-test guidance specifies one variable per test. Use the corresponding experiment setup in the platform where you will run the campaign.
Build a creative test brief or inspect the CeraVe worked example. Its proposed test has not been run; it is not evidence of uplift.
For agency creative strategists
What to request before choosing a Kantar or System1 alternative.
Compare the same brief across suppliers: assets, audience, market, platform, decision and deadline. Specify whether you need modeled screening, fresh consumer responses or an expert revision recommendation. A faster report that answers a different question is not an equivalent study.
- Inspect the actual output. Request an annotated asset, prioritized edits, evidence labels, rationale and next-test plan. Check the illustrative RoastIQ report and public worked case; neither is a client outcome study.
- Match the evidence to the claim. Ask for the validation target, unseen test set, relevant cohort and known failure conditions. Predicted attention, simulated testimony, real responses and observed campaign results must be labeled separately.
- Agree the commercial scope. Specify brands, assets, revision rounds, review ownership, delivery format and handoff date. SaliencyLab’s published service packages do not establish unlimited multi-client delivery; agree any additional scope before work begins.
- Choose a different method when needed. Commission real-participant research for measured audience reactions, recruitment-specific cultural questions or population estimates. Use a properly designed campaign experiment for in-market effects. SaliencyLab has no common head-to-head study proving superiority over another supplier.
Check the exact products: Kantar LINK AI, System1 Screen and Pro, and Zappi consumer research and Amplify AI. For our evidence and limitations, read the methodology.
Two decisions to settle before commissioning.
Can a €500 budget cover three TikTok videos?
For creative diagnosis and a human-reviewed editing brief, our €500 sprint covers one brand, one campaign, one audience and one platform, with up to six creatives. Three videos fit that asset limit when each is at most 60 seconds. Each edit or language version counts separately. VAT is additional where applicable; a budget inclusive of tax needs a confirmed quote.
You receive annotated priorities, one 30-minute decision call and one review of revised assets. Initial recommendations are due within 10 business days of the agreed start after payment and complete materials. Your team produces the edits. Respondent recruitment, media spend and video production are outside this service.
If the budget must also buy real-person feedback or advertising delivery, compare those costs first. A respondent study can investigate comprehension and preference; controlled delivery tests campaign effects. Neither our model scores nor simulated BuyerLens responses establish a winning ad or expected ROAS.
Check the exact scope and revision terms · Inspect a labeled worked example
What should an agency agree before sharing unreleased client ads?
Confirm the client permits processing by the proposed supplier and its subprocessors. Read the current data-use terms, processing locations, retention and deletion arrangements, access controls and incident contact. Agree any NDA or client-specific restrictions in writing before uploading. Do not infer zero retention or permission for confidential work from a generic AI label.
For SaliencyLab, start with our security overview, privacy policy and terms. Resolve any client requirement those pages do not answer before supplying the assets. A multi-brand agency engagement needs separately agreed scope.
Ask for a handoff that connects each proposed edit to a frame or passage, the reasoning, the evidence type and the next check. Keep model predictions, simulated comments, real respondent findings and measured campaign results separate. Our review helps plan revisions; it does not certify regulatory compliance or replace a consumer study.
Product-by-product · source check 2026-10-07
What the agency actually receives.
Dated research snapshot by SaliencyLab, based on the linked official product pages. Sources are not monitored continuously and offers may have changed since the check date. Vendor descriptions are not independently verified performance results. This is a selected comparison, not a ranking or common head-to-head test. Reconfirm scope and pricing before buying.
SaliencyLab creative review
AI diagnostics with human review in service engagements; BuyerLens feedback is simulated.
- Deliverable
- Five-KPI diagnosis, evidence-linked concerns, revision priorities and a next-test plan.
- Scope
- Agree brands, assets, review ownership and delivery date using the published service packages. Software access is a separate offer.
- Commercial check
- Published packages; confirm the applicable scope on the pricing page.
- Decision boundary
- Useful for deciding what to revise. No recruited consumer sample or validated sales/ROAS forecast.
Kantar LINK AI
AI predictions trained on consumer research; not fresh respondents for each AI test.
- Deliverable
- Ad-effectiveness predictions in a dashboard with channel-specific norms.
- Scope
- Creator, digital, TV, print and outdoor content. Confirm the exact market and format.
- Commercial check
- No comparable agency quote verified in the linked source; request a scoped quote.
- Decision boundary
- Choose this for modeled screening; specify LINK+ separately if the brief requires survey responses.
Kantar LINK+
Survey-based testing with consumer responses and behavioral measures.
- Deliverable
- Configurable creative diagnostics from storyboard through finished ad.
- Scope
- Digital, TV, audio, print and outdoor; agree audience recruitment and study design.
- Commercial check
- No comparable agency quote verified in the linked source; request a scoped quote.
- Decision boundary
- A consumer-study brief differs from an AI screening brief. Confirm sample, questions and exposure conditions.
- Deliverable
- Low, medium or high ranges for Star, Spike and Fluency.
- Scope
- Finished long-form TV, BVOD and YouTube ads; US and UK scope on the cited page.
- Commercial check
- No comparable agency quote verified in the linked source; request a scoped quote.
- Decision boundary
- Do not assume this covers short-form Meta or TikTok. Confirm eligibility for the actual asset.
- Deliverable
- Star, Spike and Fluency plus emotional-response diagnostics; downloadable reporting.
- Scope
- TV, social, audio, print and outdoor; concepts through finished ads.
- Commercial check
- No comparable agency quote verified in the linked source; request a scoped quote.
- Decision boundary
- Request the target sample, questions and stimulus format. Pro and Screen have different evidence bases.
- Deliverable
- AI-Quick Report: Creative Power, Reach, Resonance and Response; strengths, weaknesses and asset comparisons.
- Scope
- Social video and YouTube; one project can group assets for the same channel, category and market.
- Commercial check
- No comparable agency quote verified in the linked source; request a scoped quote.
- Decision boundary
- Simulated reactions are not fresh human testimony. Ask for a human-respondent study when that is the decision requirement.
Ipsos Creative|Spark AI
AI predictions based on relationships with Creative|Spark survey KPIs.
- Deliverable
- Dashboard of predicted KPIs, predicted attention heatmaps and creative features for video.
- Scope
- Linear, social-newsfeed, skippable pre-roll and short-form video; digital, outdoor and print static ads.
- Commercial check
- Official page advertises “From $470”; confirm assets, market, taxes and included deliverables.
- Decision boundary
- The AI offer does not recruit a new respondent sample for the supplied creative. A starting price is not an equivalent-study quote.
Adverteyes creative intelligence
AI creative scoring; Human Measurement is a separate research offer.
- Deliverable
- Dashboard organized around Sales Impact, Brand Awareness and Interactions; these are vendor metric names.
- Scope
- Confirm format, platform and market. Human Measurement uses an audience sample for attention and emotion, with an optional brand survey.
- Commercial check
- No comparable agency quote verified in the linked source; request a scoped quote.
- Decision boundary
- Specify which offer is included. An AI score must not be presented as an observed campaign result or fresh participant response.
For comparable quotes, send the same assets, audience, market, platform, deadline and required output to each supplier. Ask separately about agency access, client sharing, reuse rights and review support; the product descriptions above do not establish those commercial terms.
Before the next edit round
What should your creative team receive?
A useful review connects a specific creative moment to a decision your editor can act on. Bring the original assets, audience, offer, platform, brand constraints and the question you need to resolve.
- Keep
- The passage, image or brand cue worth preserving, with the reason.
- Change
- The exact frame, timestamp or wording to revise; the proposed edit and its priority.
- Explain
- What is visible in the creative, what the model infers, and what remains uncertain. BuyerLens objections are simulated hypotheses.
- Decide and test
- Who accepts, adapts or rejects the recommendation, which version gets made, and what real evidence would support the next decision.
Why use SaliencyLab if we already use ChatGPT or Claude?
A general assistant can help review an ad when you provide the assets and context. SaliencyLab brings brand context, eligible market references, creative versions, diagnostics and recorded decisions together in a project. The review service adds a human-owned recommendation and revision review; a software subscription is a separate offer.
That workflow is a reason to evaluate the product, not proof of better advice. Compare the same assets and brief, check whether recommendations are supported and usable, and record the time your team needs to reach a decision. We have not completed an independent head-to-head quality study against ChatGPT or Claude.
Read the comparison protocol v2Choose real participant research when you need actual buyer responses. Use a controlled campaign test for performance effects. Meta diagnostics remain directional, and BuyerLens cannot independently validate RoastIQ.
Before you send an asset
Formats, scope and what you receive.
- Images
- JPEG, PNG and WebP, up to 50 MB per file. Image analysis excludes audio and motion-specific stages.
- Video software intake
- MP4 and MOV, up to 180 seconds and 100 MB per file. Missing audio, transcript or scenes limit what a report can conclude.
- Expert review packages
- One image or video up to 60 seconds per creative. Each edit, language version or placement adaptation counts separately. Longer videos need a separate quote.
- Comparisons
- Use the same objective, market, platform and audience. A creative with incomplete evidence cannot support a confident winner. Scores from different providers are not directly comparable.
- Reports and exports
- RoastIQ diagnostics and revision priorities; separate BuyerLens synthetic studies. PDF/JSON report export is a paid software feature. Service deliverables are included in the agreed scope.
GIF, HEIC, SVG, TIFF, AVI, MKV and WebM are not supported by this intake. AI diagnostics and simulated interviews do not measure consumer responses or campaign lift.
Where SaliencyLab fits
From a creative question to a clear next edit.
Five KPI diagnostics, AI-simulated resistance themes where needed, and a human review. We connect the evidence to what your team should keep, change and test.
For a DTC team
Bring your Meta or TikTok assets, offer and target buyer. Leave with annotated revision priorities your team can act on. Meta diagnostics remain directional; actual conversion effects need a campaign test.
For an agency
Build a client recommendation with clear reasoning between edit rounds. When the brief requires measured audience responses, commission research with real participants.
Inside a worked creative review
Keep the explanation.
Test the brand cue.
Follow the CeraVe ad from its original frames to an archived diagnosis, a proposed edit and a controlled test plan.
Editorial analysis of a public ad. No customer endorsement or measured improvement is claimed.
Explore the worked caseWant to see the report layout? Open the RoastIQ demo Its sample numbers are illustrative.
A scope you can plan around.
Expert review and deliverables are included in these service packages. Software subscriptions are separate. Prices exclude VAT where applicable.
Creative review sprint
Launch with Confidence
€500/ sprint
One brand · one campaign · up to 6 creatives
- A prioritized creative shortlist, with the reasoning
- Annotated frames and visual editing recommendations
- One 30-minute decision call
- One review of up to 6 revised creatives
Initial recommendations within 10 business days. The window starts on the agreed date after payment and complete materials.
Discuss your campaignOngoing creative review
Creative Decision Partner
€1,000/ month
One brand · 2 cycles · up to 6 creatives per cycle
- The complete sprint deliverables in each cycle
- Two reserved review slots on an agreed calendar
- Two campaign briefs, with relevant category references
- A shared decision log that carries learning forward
Initial recommendations within 10 business days per cycle. The window starts on the agreed date after payment and complete materials.
Discuss a partnershipAsset limits, editing and workspace access
One image or one video up to 60 seconds. Each edit, language version or placement adaptation counts separately. Each cycle covers one campaign brief, one audience and one platform for one brand. Longer videos, additional markets or multiple brands receive a separate quote before work starts.
Your designer, editor or agency makes the changes. We provide annotated frames and priorities for the opening, sequencing, product visibility and benefit demonstration. Production, video editing and media buying are outside this review service.
The workspace access and analysis usage needed for the agreed review are included, with no additional software subscription required for that work. You receive downloadable deliverables to keep. Ongoing software use after the engagement requires a separate workspace plan; access and data-retention arrangements are confirmed in your proposal.
Full service termsKeep the evidence clear.
Use real interviews to explore buyer language, quantitative research for population estimates, and controlled campaign tests for in-market effects.
BuyerLens produces hypotheses, not independent confirmation of RoastIQ. Separate attribute-model validation does not validate every KPI or predict sales. Provider scores need a common study before they can be compared.
Our methodologyWhich creative testing tool should I use?
Start with your research question. Compare AI diagnostics, synthetic interviews and actual consumer research as distinct kinds of evidence.
How should a small DTC brand test Meta and TikTok ads before launch?
Choose diagnostics for revision priorities, real-person research for comprehension and preferences, and controlled campaign testing for performance. SaliencyLab supports creative diagnosis and optional human review; Meta diagnostics and synthetic feedback remain directional. PickFu and Lyssna offer different forms of participant feedback. Match the method to the decision rather than treating the outputs as interchangeable.
What should an agency compare when choosing a Kantar or System1 alternative?
Match the exact product, brief, audience, market, deliverables, evidence and commercial scope. Kantar LINK AI, System1 Screen, System1 Pro, Zappi consumer research and Amplify AI use different methods. SaliencyLab fits diagnosis and revision planning; it does not replace a fresh consumer study or prove campaign uplift.
Are these tools mutually exclusive?
No. Your workflow may combine diagnostics and research. Choose the scope, budget and evidence you need; no universal provider sequence is required.
How were these comparisons checked?
Product descriptions were checked against the linked official provider pages on 7 October 2026. Unverified price ranges, panel sizes and timing claims have been removed.