← Product
Product I founded

Wilow

Wilow shows what is working in your own Meta ad account, and in your competitors' too, then helps you act on it. I founded it and run the product side: mockups, feature scoping and handoffs to the developers who write the app.

88monthly active users
Visit trywilow.com

What it is

Wilow is creative analytics for Meta ads. It shows what is working in your own ad account, and in your competitors' too.

It ranks the ads in an account by performance, so the best sit at the top and the worst at the bottom. Its Ad Studio then takes an ad worth learning from and makes you your own version of it, so Wilow helps you act on what is working, not only see it.

The Wilow home page at trywilow.com.
The Wilow home page at trywilow.com.

What I did

I founded Wilow and run the product side: mockups, backlog items, product reviews, feature scoping and handoffs to the developers.

The app code is written by Wilow's developers, not by me. My side is the product and the marketing.

The marketing is run day to day by a fleet of agents I built. They handle research, content, SEO, outbound email and competitor reports.

How it works

Sign-in runs through Clerk, and signup asks only for an email, a first name and a last name.

A new user lands on the Ad Library first and connects a Meta ad account once they have seen value, not before.

When an account syncs, its ad creatives are labelled automatically by Google's Gemini models. Gemini 2.5 Flash reads the images and Gemini 3.5 Flash reads the videos.

The Creative Leaderboard then ranks the ads by performance, split by campaign objective. From there, Ad Studio can turn an ad worth learning from into the user's own image ad.

Billing runs through Polar. A successful payment updates the user's plan straight away, through a webhook.

Decisions that shaped it

A video ad is not judged from its thumbnail. A thumbnail is one frame, so the file is sampled across its whole length, at one frame every two seconds, and the spoken audio is transcribed and used as evidence for the labels.

The leaderboard uses MAD scoring rather than a simple mean. A simple average can be dragged a long way by one result, and this method keeps one freak result from distorting the ranking.

Prices are pulled live from Polar rather than written into the code. A price change takes effect without a code change or a redeploy.

Onboarding was rebuilt trust-first. A cautious user is no longer blocked behind connecting a Meta account before seeing any value, because the Ad Library comes first and the connection comes after.

The numbers

88monthly active users

This counts people who did more than load a page in the 30 days to 5 October 2026.

Built with

ClerkMetaGeminiOpenAIPolar

Skills

  • Agentic workflows