Prime, converge, rationalise.

What is the problem?

Most trend work fails at the first step: noticing. The interesting signals rarely arrive in a report. They arrive as a film scene, a music video, a photo that hits harder than it should. Without a method, you either ignore that moment or build a trend on it.

How does the model work?

Three moves, in this order:

1. Prime. One vivid image from pop culture sticks with you: a film scene, a music video, a red-carpet look. You have not seen a trend yet. You have been primed to notice one.
2. Converge. A second, independent instance shows up, ideally somewhere else entirely: on the street, in a shop, in a conversation. Two unrelated hits turn a stray image into a hypothesis: this seems to be coming back. Two versions of the same story do not count.
3. Rationalise. Only now do you look for structural drivers that explain why this would happen. Several drivers make the hypothesis stronger. None, and it stays a curiosity. Curiosities are allowed to stay curiosities.

The order matters: culture first, daily life second, analysis last. Do it the other way around and you end up explaining whatever you already believed.

What does it look like?

Smoking aesthetics. Prime: Dakota Johnson smoking in Materialists (2025), read as a strikingly powerful image. Converge: a photo of Charli XCX with a cigarette, in Vogue. Rationalise: the 90s coolness revival, a backlash against health-optimisation culture, smoking as a socially sanctioned pause, and a hunger to feel something real.

The example also shows where it can be stronger. Both hits come from pop culture, and the New York Times put both, and more, in one piece in June 2025. The stronger second hit would be a terrace, a shop window or a friend who picked it up again.

What can go wrong?

The frequency illusion. Once primed, you see the pattern everywhere and count only the hits. The counterweight sits in the method itself: the second instance has to be independent, and the drivers have to hold up when you test them.

What happens after the three moves?

Reflection · 4 min readRipples, not trends.A ripple is a small movement worth testing before it is a trend.Two hits make a hypothesis. A trend needs more. Before a candidate counts in my pipeline, it needs:

• Three independent sources. The people and places that started the thing, not three articles about it.
• A named driver. Something structural that explains why now.
• A counter-trend. If you cannot name what pushes back, you do not understand it well enough to use it.
• TRACE. True, relevant, actionable, counter-checked, evolving. Checked at entry and again every quarter. Trends that stop passing get retired, not defended.

Then it moves into Trend planning: cluster, validate, prioritise, plan. This method feeds that process. It never replaces it.

Why does it still matter?

Scanning is now largely automated. Agents count mentions well. They do not notice an image that hits harder than it should. That front end stays human, and it is cheap: you can run it while watching a film or walking through a city.

The method as articulated by trendwatcher Tom Palmaerts (Trendwolves) in conversation, 5 August 2026. Later overview of the smoking example: Esther Zuckerman, Pop Culture Takes Up Smoking Again, The New York Times, 11 June 2025. The explanation and the link to my own pipeline are my interpretation, not Tom’s.