Creative Pattern Analysis

It started with a simple question: what actually makes people buy? Scarcity, reviews, fear of missing out, a good photo, a limited-time offer. Every team has a favourite answer, and every answer comes with a case study.

So I went to the data. A few hundred thousand ads, and the results behind them, to see what the ones that sold had in common.

What I was looking for

The usual list of triggers and tactics: scarcity, social proof, urgency, the hook in the first seconds, the format, the offer. If any of them reliably separated the ads that sold from the ones that didn't, a dataset that size should have shown it.

Why the question is harder than it looks

• Cheap signals. The ads with the most likes were, more often than not, not the ones that sold the most. Engagement measures attention. It doesn't measure intent.
• The platform's own choices. Delivery systems push budget behind the ads they favour early, so a "winning" ad has often simply been shown more. A lot of what looks like a pattern is the algorithm's taste reflected back.
• Everything moves at once. Offer, audience, budget, season and placement change alongside the creative. It's easy to credit a headline for results it shared with a discount or a payday, and it's wrong.

What it changed

Mostly I learned why the question is harder than the industry admits. A lot of what marketing tells itself, about which tactics work and what a psychological trigger does to conversion, is convention presented as evidence.

It also changed how I work. I read customer behaviour by what each signal cost the person who sent it. I treat patterns found in ad data as ideas to test, not conclusions. And I try to keep the line between what was measured and what is being claimed in plain view.

The essays that came out of it

• Why a Customer's Silence Is More Honest Than Their Like
• The Psychology Marketing Repeats Without Checking