6 creatives · MYOTIDE / Meta prospecting · Purchase conversion
Predicted highest ROAS
Variant AFund this concept84
Example model output · Real MYOTIDE campaign creative
How it works
How Kettio predicts which ad is stronger.
You provide the competing ads, the buyer, and the goal. Kettio simulates likely buyer responses, compares every ad on the same scale, and returns a directional winner before you spend.
01 Upload your ads
Add the ads you are choosing between.
Kettio compares like with like: static images, copy, or full ad packages. Each creative is evaluated separately so one ad cannot influence the read on another.
02 Choose the buyer and goal
Tell Kettio who you want to persuade.
Choose the platform, conversion goal, and buyer profile. Those choices determine the context and the question the simulated audience evaluates.
03 Simulate the response
Kettio models how that buyer is likely to react.
AI respondents modeled on your chosen buyer evaluate each ad independently. Their responses are compared with calibrated anchors and combined into a directional ranking.
04 Make the decision
Get a winner, the reasons, and clear next steps.
See which ad leads, why it leads, how confident the read is, and what to improve. Kettio flags close calls; live media remains the final proof.
For high-volume creative operations
Built for teams that ship thousands of ads.
The decision layer between production and media: rank the batch, cut weak work, and launch a stronger shortlist.
1,089Real ads scored on the PVP academic benchmark (Seoul National University, ACL 2025), beating GPT-4o zero-shot on correlation.
83.45%Pairwise agreement (701 of 840 pairs) for the outcome-trained comparator on the CreativeRanking development cohort.
0.775Correlation (ρ) against the University of Washington ad-perception survey, n=160.
~30sFrom upload to a ranked read on a two-ad comparison.
Correct-prediction volume reflects predictions produced across the PVP academic benchmark (1,089 ads scored) and the CreativeRanking development cohort. The pairwise figure is a retrospective result on that cohort, not a fresh holdout, and its frozen success gate is not fully passed: ranking cleared all three control banks while click-capture cleared one of three. Predictions are directional and do not guarantee CPA, ROAS, or purchases; individual results vary by account, campaign, offer, and media conditions. Read the benchmark.
Find your fit
How does Kettio compare?
See where Kettio fits alongside the tools you already use, and which one solves your next creative decision.
The questions high-volume creative teams ask before putting a model in the launch workflow.
Kettio pre-scores ad creative for a specific audience, platform, and campaign goal. It returns a relative ranking, rationale, and confidence signal so you can cut weak work and concentrate live media on the strongest creative.
Kettio can rank for purchase intent and conversion potential, not only attention or engagement. The output is a directional winner signal—not a promise of an exact CPA or ROAS. Use it to decide what deserves live testing; use the ad platform to verify the final behavioral outcome.
Losers. On a public benchmark of marketplace product images, Kettio identified the weakest creative in a batch about 62% of the time against a 23% chance rate, while its winner picks on the same benchmark were close to chance. That is why the workflow is cut the bottom of the batch first, then let live media choose the winner from a smaller shortlist.
Yes. In a blind back-test, Kettio ranks past creatives without seeing spend, purchases, CPA, ROAS, CTR, or the winner labels. We then compare the ranking with your known outcomes so you can judge the signal on your own brands before adopting it.
That starts with the blind back-test. If you have enough clean creative, spend, and outcome data, a founding pilot can scope account-specific calibration around the patterns that actually matter for your portfolio. This is a pilot workflow, not an automatic self-serve training switch.
Images are batch-ready through the Ranking API, which accepts up to 20 images per request and 60 scoring evaluations per minute per API key. Larger image batches can be orchestrated across requests. High-volume video ingestion and turnaround are scoped with founding pilots while the dedicated bulk-video workflow is productionized.
Yes through the Ranking API: it returns structured JSON with asset IDs, rank, score, rationale, and confidence data. The request also accepts audience, goal, platform, and ad-copy context. Founding pilots map your concept, persona, angle, and vertical taxonomy into the handoff your team uses.
Individuals and small DTC teams can use the $29.99 or $139 monthly subscriptions. Agencies, multi-brand groups, and teams producing hundreds or thousands of creatives use a custom pilot based on validation scope, throughput, video needs, API workflow, and calibration requirements.
No. Kettio changes the order of operations. Your creative team still makes the work, and live media still provides the final truth. Kettio helps the team remove the obvious losers first, learn why the strongest work wins, and spend the test budget on a smaller, better shortlist.
MYOTIDE
Predicted highest ROAS