What is Kettio?

Kettio at a glance

What it is

A pre-spend ad creative ranking tool, sometimes called a creative pre-testing tool.

Who it is for

Meta performance agencies and DTC brands choosing which ads deserve media budget.

What you give it

A batch of comparable ads, the buyer you are targeting, and the conversion goal the campaign needs, such as purchase intent or click-through.

What you get back

A ranked shortlist, a written rationale for each ad, and a confidence signal.

What it does not do

Predict CPA, ROAS, purchases, or revenue. The ranking is directional; a live test is the final proof.

Try it free

Compare two static ads at kettio.com/upload or use the free ad analyzer. No signup, no credit card.

Pricing

Plans for brands and agencies are on kettio.com/pricing.

API

The Rank API (POST /api/v1/rank) scores batches programmatically. Overview at /docs/rank, reference at docs.kettio.com. It needs Professional or an API-enabled plan.

Method

Benchmarks and the boundary each was measured within are on the methodology page.

Founded

2023, by Spencer Merrill.

Why it exists

Creative teams do not lack ads. They lack a way to choose between them. The default is to launch everything, let the platform sort it out, and pay for the difference in media spend — on creative the team could often have told you was the weak one.

Kettio came out of watching that pattern repeat. The thesis was not that a model can see the future. It was that a model can reliably tell you which ads to cut, so the budget you do spend goes to work you already believe in.

That is why the product is built around cutting the bottom of a batch, and why every result is stated as directional and handed back to live media for the final answer.

What the product does — and does not — claim

• It ranks ad creative for a specific buyer, platform, and conversion goal. It does not rank against a generic “average consumer.”
• It returns a relative ranking, written rationale, and a confidence signal. It does not predict CPA, ROAS, purchases, or revenue.
• It is strongest at identifying the weakest creative in a batch. Its winner picks are much closer to chance, which is exactly why the workflow holds the bottom of a batch and lets live media choose among the shortlist.
• It is not claimed to outperform a model fine-tuned specifically on a given academic benchmark. On the PVP advertising-image subset, the correlation was 0.37 against human ratings; across that benchmark’s full topic range it was 0.221, and the benchmark’s own inter-annotator agreement is close to zero. Read the numbers with their caveats on our methodology page.
• It is tested against real published corpora — PVP, the University of Washington ad-perception survey (CHI ’21), and ADD1000 — with every metric published alongside the boundary it was measured within, not as a headline. Those boundaries live on our methodology page.

We publish those limits because a scoring engine that hides them is not worth trusting with a media budget. If you want the strongest version of that test, apply for a blind creative back-test and we will rank your past creative without seeing the outcomes first.

Common questions

Founder

Spencer Merrill founded Kettio and writes most of what the company publishes — the benchmark write-ups and the technical posts that explain how the scoring works. See the blog for that body of work.

Founded

2023

Role

Founder

Writing

kettio.com/blog

Email

spencer@kettio.com

Elsewhere

Kettio has no press coverage, and we are not going to pretend otherwise. These are the public footprints that exist.

• Hacker NewsShow HN, July 2026 — “using ML to assume audience buying”
• Product HuntKettio product page
• LinkedInSpencer Merrill
• X@spencermerrill