Locked inside with a question

Everyone has stopped. The world is indoors. Workshops are cancelled. Client work is frozen. And I have a lot of time with a question I've been carrying for a while.

BehaviourKit has methods. Ways to structure conversations about behaviour. Canvases. Frameworks. Cards with named biases on them. All useful. All tested. But I keep bumping into the same problem: we have methods for thinking about behaviour, but we don't have a proper catalogue of what actually works when people try to change it.

When a team finishes a diagnosis and asks "so what should we try?" the honest answer right now is: it depends, and I'll help you figure it out. Which is fine when I'm in the room. Less fine when I'm not.

So here's what I'm going to do with all this unexpected free time. I'm going to read. Properly read. Not skim the abstract and file it. Actually sit with academic papers, field trials, case studies, programme evaluations. Anything where someone tried to change behaviour and documented what happened.

I want to know: what intervention shapes actually recur? Not in theory. In practice. What do people actually do when they're trying to get someone to start, stop, continue, or change a behaviour?

The behavioural science world is having its own moment right now, by the way. Every government on earth seems to have discovered nudging overnight. "Wash your hands" posters in every language. Behavioural Insights Teams advising on lockdown compliance. There's a surge of interest in applied behavioural science and also a surge of oversimplification. "Just nudge them" is becoming the new "just make it user-friendly." A phrase that sounds helpful and means almost nothing.

I don't want to add to the noise. I want to find the signal.

I've started with a spreadsheet. Every paper I read gets a row. Columns for: what was the problem, what was tried, what mechanism they thought would work, what actually happened, what domain it was in. I'm coding each case across multiple theoretical lenses. Not just COM-B. Not just Fogg. All of them at once. Because I've already learned that no single framework captures everything, and I want to see what recurs across theories, not just within one.

It's slow. Properly slow. Each paper takes an hour or more to code thoroughly. Some take longer because the intervention is complex or the reporting is vague. I'm reading across healthtech, sustainability, public policy, consumer behaviour, workplace change, community programmes. Anywhere someone tried to shift what people do.

I'm about 40 papers in and something is already starting to happen. I'm seeing shapes. Not individual techniques, but recurring structures. An energy company in the UK and a government health programme in Denmark are doing structurally the same thing, even though neither would recognise it.

I don't have names for these shapes yet. And I'm not going to force names too early. But I'm keeping a separate list. Candidate patterns. Things that show up more than once, in more than one domain, doing the same kind of work.

The process feels a bit like archaeology. You brush away the surface (the domain, the specific target behaviour, the cultural context) and underneath there's a structure that keeps reappearing.

I'm also keeping a graveyard. Every time I think I've found a new pattern, I compare it against every existing pattern on my list. If it's not genuinely different, one of them goes. I'd rather have 30 tight patterns than 100 loose ones.

I don't know how long this will take. But I've got nowhere to be. And the question is a good one.

Go deeper into the Building BehaviourKit series: