Campaign testing platform

Test the campaign before you fund it.

Put two campaign concepts in front of the buyer you’re targeting. See which one they’d act on, what decided it and what to test next, then send the stronger concept into live media.

Try it on your own ads
Live testTwo concepts · one buyer · about 30 seconds
Who is the buyer?
What should the concept do?

Free. No signup or ad account. Your images are scored, not published.

Testing more than two, or a custom buyer?Open the full test

Campaign testing vs creative testing

Different question. Different unit.

Both matter. Campaign testing comes first, because polishing an asset only pays off when the idea behind it was right.

Creative testing

Which asset is stronger?

Compares versions of one idea: image A against image B, headline one against headline two. Useful once the concept is settled.

Campaign testing

Which idea deserves the budget?

Compares the ideas themselves: different hooks, angles and offers, each framed for the same buyer and the same goal. Useful before anything is funded.

The test above handles both: two different concepts, or two cuts of the same one. More on ad creative ranking →

How it works

One buyer. One question. A fair fight.

The same engine runs the test at the top of this page and the full report.

01

Hold the buyer and goal steady

Choose who the campaign is for and what it should make them do. Every concept is read by the same buyer against the same question, so the comparison is fair.

02

Each concept gets a buyer reaction

The buyer model reads each ad and says what it noticed, what it believed and whether it would act. Close calls go to a three-voter panel instead of being decided on noise.

03

See what decided it

Pins land on the hook, offer, proof, visual and call to action that moved the verdict, with one concrete change to test next. Brief the next round from that, not from a hunch.

Before media, not instead of it

Spend live budget on the finalists.

Live testing is still the proof. A pre-launch read decides which concepts earn a place in it.

Without a pre-launch read

Every concept gets a live cell. Giving each one even 10,000 impressions at a $15–$25 CPM costs $150–$250 per concept before you learn anything, and weak concepts spend at the same rate as strong ones.

With a pre-launch read

Rank the concepts first, cut the ones your buyer ignores, and fund live cells only for the two or three that earned them.

After launch

Live results remain the proof. Feed what the platform shows back into the next round of concepts.

Evidence

Measured, published and bounded.

Each figure measures something different, so they are reported separately and never rolled into one accuracy number.

83.45%
Pairwise agreement (701 of 840 pairs) for the outcome-trained comparator on the CreativeRanking development cohort.
0.775
Correlation (ρ) with the University of Washington ad-perception survey, n=160. Measures what people say about ads.
1,089
Real ads scored on the PVP academic benchmark (ACL 2025), beating GPT-4o zero-shot on correlation.
~30s
From upload to a ranked read on a two-ad comparison.

The pairwise figure is a retrospective result on the CreativeRanking development cohort, not a fresh holdout. Predictions are directional and do not guarantee CPA, ROAS or purchases; results vary by account, offer and media conditions. Methodology · Read the benchmark

Where it fits

Kettio vs Motion, Marpipe and AdCreative.ai.

Most testing tools answer after the money is spent. Kettio answers before, then hands off to them.

CapabilityKettioMotionMarpipeAdCreative.ai
When it answersBefore media spendAfter spend, from live dataDuring live multivariate testsBefore launch, by generating options
Ad account requiredNoYesYes, to serve testsFor performance scores
Reads a specific buyerYes, you choose the buyerAccount-level resultsLive audience resultsGeneric scoring
Explains the resultElement by element, per buyerPerformance reportingElement-level live resultsLimited
Hands you nextA ranked winner and the change to test nextInsight reportsA live winner, after spendMore variations

Capabilities as publicly described at the time of writing. Kettio vs Motion · Kettio vs Marpipe · AdCreative.ai alternative

Who tests this way

Built for teams that ship a lot of concepts.

Agencies

Walk into the client review with a ranked shortlist and the reasoning behind it, instead of a mood board and a hunch.

Multi-brand groups

Judge every brand’s concepts against the same buyer and the same question, so results compare across teams.

In-house brand teams

Cut the concepts your buyer ignores before they take a live cell, and spend the testing budget on finalists.

FAQ

Campaign testing questions.

Creative testing compares versions of one idea: image A against image B, headline one against headline two. Campaign testing compares the ideas themselves: different hooks, angles and offers, each framed for the same buyer and the same goal. Creative testing tells you which asset is stronger. Campaign testing tells you which concept deserves the budget.

Use the test at the top of this page. Add two concepts, choose the buyer and what the campaign should make them do, and Kettio returns which concept that buyer is more likely to act on, the elements that decided it, and one change to test next. The full test adds more concepts, custom buyers and a strategist report. No ad account or media spend is needed.

Kettio publishes its benchmarks separately because they measure different things. On the CreativeRanking development cohort, the outcome-trained comparator reached 83.45% pairwise agreement (701 of 840 pairs); that is a retrospective result, not a fresh holdout. Against the University of Washington ad-perception survey (n=160) the correlation is ρ = 0.775, which measures what people say about ads, not what they buy. Treat the result as a directional pre-launch ranking. Live media remains the final proof.

It ranks; it does not forecast. Kettio tells you which of two comparable concepts a buyer is more likely to act on, and why. It does not output a point-estimate CTR and does not guarantee CPA, ROAS or purchases.

No. Pre-launch testing is meant to give you signal before live campaign data exists, so Kettio never needs an ad account connection to score concepts.

The test on this page uses preset buyers such as cold traffic and repeat buyers. The full test lets you describe your own buyer or start from segments suggested from your ads. No real user data is used.

The test on this page compares static images. The full test also compares short videos (MP4, WebM or MOV, up to 90 seconds), where the read is about whether a viewer keeps watching.

The test on this page and your first comparisons in the full test are free, with no credit card. Paid plans add volume, larger batches and team features.

Before the next launch

Know which concept to fund.

Two concepts, one buyer, a clear call in about 30 seconds. No signup, no ad account.

Find my winning ad
  • 01Pick the buyer and goal
  • 02Get the verdict and the why
  • 03Fund the finalist