Most behaviour-change efforts fail because they pick the intervention first and diagnose the problem never. COM-B and the Behaviour Change Wheel are the antidote: a disciplined method that makes you work out which of three things is actually stopping the behaviour before you spend a penny trying to change it.
COM-B is the most coherent shared framework behaviour-change practice has. It holds that for any behaviour to happen, three conditions must all be met: Capability, both physical (strength, dexterity) and psychological (knowledge, skills, attention); Opportunity, both physical (time, resources, environment) and social (norms, what others do and expect); and Motivation, both reflective (goals, intentions, beliefs) and automatic (habits, impulses, emotions). Behaviour sits at the centre where the three meet, and it changes only when the one that is actually missing is supplied (Michie, van Stralen and West, 2011). The core discipline follows immediately: before designing anything, diagnose which of Capability, Opportunity, or Motivation is the binding constraint, because an intervention aimed at the wrong one does nothing.
To make that concrete, take someone trying to start cycling to work. Their Capability is whether they can ride and are fit enough for the distance (physical) and know a safe route (psychological). Their Opportunity is whether they own a bike, have somewhere to park and change, and have the time (physical), and whether cycling is normal among the people around them (social). Their Motivation is whether they want to, how they feel about arriving flustered, and whether it has become automatic (reflective and habitual). The important subtlety is that the three do not simply add up: they interact. A route that feels unsafe, an Opportunity problem, quietly drains Motivation however keen the person began, which is why supplying the genuinely missing piece matters more than strengthening the ones already present. And behaviour feeds back on its own causes: once the first few rides go well, both capability and motivation grow, which is how a change takes hold, or, if the early experience is miserable, collapses. Diagnosing the binding constraint is exactly the discipline of finding, in a specific case like this, which of those boxes is the one actually holding the behaviour back.
The Behaviour Change Wheel builds outward from that hub. Around the COM-B core sit nine intervention functions, the broad kinds of thing an intervention can do: education, persuasion, incentivisation, coercion, training, restriction, environmental restructuring, modelling, and enablement. Around those sit seven policy categories (guidelines, regulation, legislation, communication, and so on) through which the functions get delivered. The design move is to read outward from the diagnosis: a Capability problem points to training, education, or enablement; an Opportunity problem points to environmental restructuring, restriction, or enablement; a Motivation problem points to persuasion, incentivisation, modelling, or coercion. Michie, Atkins and West turned this into a step-by-step design protocol now standard in public health (Michie, Atkins and West, 2014).
Two companion tools complete the kit. The Behaviour Change Technique Taxonomy is a catalogue of 93 specific techniques grouped into 16 clusters (goal-setting, feedback and monitoring, social support, shaping knowledge, and the rest), which gives the field something it badly lacked: a shared vocabulary for reporting exactly what an intervention did, so trials can be compared and replicated (Michie et al., 2013). And the Theoretical Domains Framework offers a finer-grained diagnostic layer, fourteen domains of influence on behaviour that map onto COM-B, useful when "Capability, Opportunity, or Motivation" is too coarse to locate the real barrier (Cane, O'Connor and Michie, 2012).
The central and honest point is that COM-B is integrative, not predictive. It does not supply new causal claims about behaviour; it organises what the older, scattered models already said into one usable structure. That is a real limitation if you expect it to forecast, and a real strength if you understand what it is for. Its genuine contribution is to force an explicit diagnosis of the constraint before the intervention is chosen, which sounds obvious and is astonishingly rare in practice, where the intervention (usually an awareness campaign or an app) is chosen first and the problem reverse-engineered to fit.
The comprehensiveness that makes COM-B useful also makes it hard to falsify: because every behaviour has some Capability, Opportunity, and Motivation component, almost any intervention can be justified against the wheel after the fact, so the framework can become a veneer of rigour over undisciplined design, a diagram in the strategy deck rather than a genuine diagnosis. The safeguard is honest, evidence-led diagnosis, and the discipline matters more than the wheel. On the techniques themselves, the evidence is sobering in the familiar way: the effects of specific behaviour-change techniques are moderate and highly context-dependent, and whether particular technique combinations work reliably across settings remains open, with the subsequent meta-analytic record showing modest, conditional results (the same effect-size humility that runs through L5-08 and L5-13).
COM-B's breadth is bought at the price of predictive power, so it guides design rather than forecasting outcomes. Applying the full protocol well is effort-intensive and demands genuine behavioural data, not a workshop's guesses, about where the constraint really lies. The BCT effects are moderate and setting-specific. And the diagnosis itself is culturally contingent: what is a Capability barrier in one context is an Opportunity barrier in another, and intervention functions sit very differently across societies, since coercion and restriction that are routine in one system are unacceptable in another.
Do specific behaviour-change-technique combinations have reliable effects across contexts, or only within them? Can COM-B ever become genuinely predictive, or is it destined to remain a diagnostic scaffold? And how much of the protocol's value comes from its detailed machinery versus simply from forcing practitioners to diagnose the constraint at all?
The usable core: never pick the intervention first. Specify the behaviour precisely, diagnose which of Capability, Opportunity, or Motivation is the binding constraint, match the intervention function to that constraint, specify the techniques so you can measure them, and expect moderate effects you will need to test. What follows is the design workflow as concrete steps, then a per-component menu of what to reach for.
The ethics belong up front, because a method for changing behaviour is a method for changing behaviour whoever wields it. Two of the nine intervention functions are coercion and restriction, which is a useful reminder that this toolkit reaches well past gentle nudging into things a company or a state can do to people, so the governing questions are whose behaviour is being changed, in whose interest, and with what transparency and consent. Applied to a customer's or a citizen's own goals, with their knowledge, the method is a genuine service; applied to extract behaviour they would reject if they saw the design, it is manipulation with a diagram.
1. Specify the target behaviour precisely. Not "eat more healthily" but "add one portion of vegetables at the evening meal on weekdays", named down to who, what, when, and where. Most behaviour-change work fails at this first step, because a vague target cannot be diagnosed.
2. Diagnose the binding constraint with COM-B. Ask, with real evidence rather than assumption, whether the thing stopping the behaviour is Capability (they cannot), Opportunity (their world will not let them), or Motivation (they do not want to), and use the Theoretical Domains Framework when you need a finer cut than the three headings give (Cane, O'Connor and Michie, 2012). This is the step that earns the whole method its keep.
3. Match the intervention function to the constraint. Read outward from the diagnosis to the nine functions rather than starting from your favourite tactic (Michie, van Stralen and West, 2011), using the per-component menu below.
4. Specify the behaviour-change techniques. Choose concrete techniques from the taxonomy and write them down, both so the intervention is precise and so you can report exactly what you did and compare it with everything else (Michie et al., 2013).
5. Deliver, measure, and report. Expect a moderate, context-dependent effect, run it as a test with an honest counterfactual, and report in the taxonomy's shared vocabulary so the result adds to a comparable evidence base rather than a pile of incomparable case studies.
If the constraint is Capability (they cannot do it yet). The person lacks the knowledge, skill, or mental resource. Reach for training (build the skill through practice), education (supply the understanding), and enablement (tools, aids, and support that extend capability), for example a pill organiser and a simplified regimen for someone who intends to take medication but loses track. Caveat: capability is the constraint far less often than practitioners assume, so do not default to "educate them" (the knowing-is-not-doing lesson of L5-08); confirm the gap is genuinely capability before training for it.
If the constraint is Opportunity (their world will not let them). The environment, time, resources, or social norms block the behaviour. Reach for environmental restructuring (change defaults, layout, and availability, the strongest move of all, cross-ref L5-13 and L5-09), restriction (rules that limit the competing behaviour), and enablement (remove friction, provide resources). This is where the biggest and most reliable wins usually live, because you are removing a real barrier rather than manufacturing want. Caveat: restriction and its harder cousin coercion carry the heaviest ethical and political weight of the nine functions, and their acceptability varies sharply by society.
If the constraint is Motivation (they do not want to, yet). Reach for persuasion (reasoned or emotional argument), incentivisation (create expectation of reward), modelling (show admired others doing it), and education (where a belief is genuinely missing), with coercion (create expectation of punishment) as the last and most fraught resort. Caveat: motivation is the constraint people reach for first and it is often not the real one, and pure motivational campaigns without capability or opportunity support tend to produce intention that never becomes behaviour.
Put the discipline before the tactic: for any behaviour you want from customers or employees, run the five-step workflow and, above all, diagnose the constraint before commissioning the intervention, because the reflex to reach for an app or an awareness push is exactly the error COM-B exists to prevent. In practice the binding constraint for commercial behaviours is usually Opportunity (friction, defaults, availability) rather than motivation, so environmental restructuring and enablement will outperform another persuasion campaign more often than not. Report your interventions in the behaviour-change-technique vocabulary so you can actually compare what worked across initiatives rather than relearning each time. The serious applied home for this is Susan Michie and UCL's Centre for Behaviour Change, and the step-by-step manual is Michie, Atkins and West's The Behaviour Change Wheel: A Guide to Designing Interventions (2014), which is the practitioner's reference for doing the protocol properly.
The same diagnostic applies to civic behaviours, and it corrects the same reflex. Before assuming supporters lack motivation (and pouring resources into inspiring them), check whether the real constraint is Capability (they do not know how or when to act) or Opportunity (registration friction, timing, no one around them doing it), which is usually where the binding constraint sits (cross-ref L5-01 and the reminder evidence in L5-13). Match the function to that: enablement and environmental restructuring for an opportunity gap, clear how-to for a capability gap, and reserve persuasion for the genuine motivation gap. The honesty line matters here too, because the coercion and restriction functions have obvious authoritarian misuses that a democratic campaign should not reach for.
This is COM-B's home ground and its best-evidenced use. Build interventions on an explicit COM-B (and where needed TDF) diagnosis of the target population and behaviour, run them through the full Behaviour Change Wheel protocol, and report them in the technique taxonomy so the public evidence base compounds instead of fragmenting (Michie, Atkins and West, 2014; Michie et al., 2013). Three disciplines protect the work. First, resist the veneer-of-rigour trap: a wheel diagram in a strategy document is not a diagnosis, and the value is entirely in doing the diagnostic honestly with real data. Second, hold realistic expectations: behaviour-change-technique effects are moderate and context-dependent, so plan for modest, tested gains rather than transformation, and never let the method crowd out structural policy where the real driver is structural (the L5-13 lesson). Third, treat the coercion and restriction functions with the seriousness they deserve, since the state's use of them is a matter of legitimacy and rights, not just effectiveness. The global caveat is decisive: the diagnosis and the acceptable functions differ profoundly across societies, so a design validated in one system is a starting hypothesis, not a template, in another.
Three habits. Diagnose before you design: specify the behaviour, then work out whether Capability, Opportunity, or Motivation is the binding constraint, before choosing any intervention, because this single discipline is the framework's real gift. Match the function to the constraint using the per-component menu, rather than reaching for your favourite tactic, and remember that for most behaviours the constraint is opportunity, not motivation. And specify, test, and report in the technique vocabulary, expecting moderate, context-dependent effects, so your work adds to a comparable evidence base rather than a drawer of incomparable stories.
.png)