The Five Levers That Actually Move Behaviour
Not the ones we talk about most. The ones that change the conditions.
There’s a strange asymmetry in how the behaviour change industry talks about its own work.
The interventions that get the most airtime — motivational talks, mindset shifts, education, and various discipline-focused programs — are not the levers with the strongest and most reliable effects on behaviour over time. The interventions that actually move behaviour at scale are sitting quietly in the background, well-studied, well-validated, and consistently underused.
It’s worth knowing the difference. Both because most people trying to change something are working with the weakest tools by default, and because many practitioners are recommending them.
What follows isn’t a complete list. It’s the five levers that show up most reliably in the research, ranked by how much they actually move behaviour over time.
The ranking matters. Number one isn’t a little better than number five — it’s substantially better. But number five is probably just as far ahead of most of what didn’t make the list at all. The gap between “strong lever” and “everything else” is bigger than the gap within the top five.
None of this is a law or a final verdict. It’s information for you to work with. If you find a lever stronger than these five for your situation or your client’s, use it. The point isn’t which specific tools ended up on this list. The point is to stop and check what you’re actually leaning on. Most of the time, the tools people default to are not the ones doing real work.
Environment
If you take one thing from this article, take this one.
When you change the context a person operates in, behaviour follows almost without negotiation. The clearest demonstration on the population level comes from Iceland, which reduced teen alcohol use from 42% to 5% over eighteen years not by lecturing teenagers about substances, but by redesigning the system around them — what was available, where they spent their time, who they spent it with. They didn’t win the argument about drugs. They made the argument less relevant.
The same principle works on the individual level. The food in your kitchen, the route you walk to work, the apps on your phone’s home screen, the people you see most often — these aren’t background details. They are continuously voting on your behaviour, every single day. The person trying to lose weight while living next to a bakery and working a high-stress job is not on equal footing with the person doing the same thing in a different environment. We tend to talk about that difference as if it’s an excuse. It isn’t. It’s the most important variable in the equation.
The reason this is the strongest lever is that it doesn’t require willpower, motivation, or memory. The environment does the work whether you’re paying attention or not.
The catch is that this is also the hardest lever to actually pull.
Full environment change — moving, changing jobs, restructuring who you spend your life with — is rarely available. And even when it is, it’s not obvious that it would help, because the problem doesn’t feel like it’s about the environment. It feels like it’s about you. The environment is invisible from inside it. The bakery on the corner, the vending machine at work, the friend group where drinking is the default — these things become background rather than pressure. You stop registering them as forces acting on you.
Even when the environment is recognised as part of the problem, the exits often aren’t there. Someone with obesity can’t easily move to a food environment that isn’t engineered against them, because that environment barely exists anymore. Someone with a sedentary job can’t necessarily change professions. Someone whose social circle drinks heavily can’t just walk away from the people they care about.
But the picture isn’t binary. Between “keep everything the same” and “restructure your entire life” there’s a large middle ground where useful things happen. Reorganising the kitchen. Changing which route you take to work. Walking instead of driving on short trips. Moving to a different neighbourhood, if that’s ever possible. Adjusting the rules of engagement in a relationship, even without ending it. Most people don’t do any of this — not because they can’t, but because environmental interventions don’t feel like the answer to what feels like a personal problem.
The takeaway isn’t “change your environment or fail.” It’s that environment is worth looking at first, even when it doesn’t seem like the source of the problem. Especially then.
Choice architecture
If you can’t change the whole environment, you can almost always change how choices are arranged within it.
This is the lever that quietly runs most of the modern world. The placement of food at eye level in a cafeteria changes what people eat without changing anything they’re told to do. The default option on an organ donation form changes how many people donate. The friction between you and a behaviour — how many steps it takes, how many decisions, how much willpower at the point of action — determines whether you do it.
The principle is to design the path. Make the desired behaviour the easiest thing in the environment to do. Add friction in front of the behaviours you want to reduce. Set defaults so that not making a decision still produces a reasonable outcome.
This is less radical than full environment change, but it works on the same logic — and the research base is solid. It’s also infinitely flexible. Almost any situation can be redesigned at the choice-architecture level: the order things appear on your phone, what you have to walk past to get to the kitchen, what you have to remove from your fridge before you can reach the alternative.
The reason this lever underperforms the first one is that the surrounding context is still pulling. But within any context, choice architecture is the difference between a system that helps you and one that quietly works against you.
Implementation intentions
The biggest gap between intention and action sits in a specific place: the moment when the situation arises and a decision has to be made.
“I’ll exercise more” doesn’t survive that moment. “When I finish my morning coffee, I will put on my running shoes” often does. The reason is mechanical, not motivational. The first sentence is a general intention with no execution path. The second is a pre-made decision that fires when its trigger appears. The decision is made by the calm version of you, before the situation creates a different vote.
Peter Gollwitzer’s research on implementation intentions has shown this effect across dozens of studies and behaviours. The effect size isn’t massive, but it’s reliable, and the cost of using it is essentially zero — you just write the sentence in the right format.
The format matters: when [specific situation occurs], I will [specific behaviour].
This lever works best for specific, concrete behaviours that have an identifiable trigger. Taking medication. Doing a small routine. Saying no in a predictable social situation. It works less well for vague behaviours and large lifestyle shifts, where there’s no single trigger to anchor to.
It’s third on the list not because it’s weaker than environment, but because its scope is narrower. Within that scope, it’s one of the cheapest and most reliable tools available.
External structure
Working with someone whose job is to think about your situation produces results that working alone usually doesn’t. Not because they know more. Because they’re outside the system you’re trying to change.
This was the subject of an earlier article, so I won’t repeat it. The short version: behaviour change requires self-observation, and self-observation is unreliable under exactly the conditions where it’s most needed. An external observer — coach, mentor, therapist, peer in a structured arrangement — provides a perspective the person inside the situation literally cannot access.
The catch with this lever is quality. The research on coaching and therapy outcomes consistently shows that the alliance — the working relationship — predicts results more than the specific method used. Bad external support is worse than no external support. The lever is real, but it depends on the human on the other end of it.
This is fourth because it’s not always available, it costs money, and the quality varies. When it works well, it amplifies everything else on this list. When it works badly, it’s neutral at best.
Self-monitoring
The simplest lever, and one of the most consistently validated.
People who track their behaviour — food, movement, sleep, money, time — change it. Not because tracking is magic. Because most behaviour runs unobserved from the inside. Self-monitoring works by turning a vague experience into a visible system. You can’t change a pattern you only experience from inside.
This is the lever that fits anyone, anywhere, at almost no cost. It’s also the one most people quietly drop after a week because it doesn’t feel like it’s doing anything. The change it produces is small per day. The change it produces over three months is often significant.
It’s fifth because the effect is modest compared to environment or choice architecture. But it stacks with everything else. If you can only afford one new habit, this is the highest-yield place to put it.
What didn’t make the list
This is the part that matters as much as the levers themselves.
Education about what to do is everywhere, but it’s a weak lever on its own. Most people who don’t exercise already know exercise is good for them. Information rarely fills the gap between knowledge and action.
Motivation as a target — getting yourself or someone else fired up — produces short bursts of behaviour that don’t survive contact with everyday conditions. Motivation isn’t worthless, but it’s not a lever you can pull reliably.
Willpower, in the popular sense of pushing through resistance by force of character, is one of the weakest tools in the box. It works occasionally, in short bursts, under good conditions. Build a system around it and you’ll fail.
This is the structural mistake of most behaviour change conversations. The interventions that get the most attention — motivation, education, willpower — are the ones with the weakest effects. The interventions with the strongest effects — environment, choice architecture, implementation intentions, external structure, self-monitoring — get less airtime because they’re less dramatic.
There’s nothing inspirational about rearranging your kitchen or writing “when X, I will Y” on a card. But these are what actually moves behaviour. The dramatic version is what we talk about. The unspectacular version is what works.
If you’re trying to change something — for yourself or with a client — and the strategy doesn’t include at least the top two on this list, you’re asking the weakest tools to do the work of the strongest ones.