The AI Fogg Behaviour Model Coach skill for Claude

The model behind it comes from BJ Fogg, founder of the Behavior Design Lab at Stanford University. The core is a three-letter formula: B = MAP. Behaviour only happens when Motivation, Ability and Prompt come together at the same moment. If someone wants to but it is too difficult, it does not happen. If it is easy and someone wants to, but a signal never arrives at the right moment, it does not happen either. If one of the three is missing, the behaviour stays away, however good the idea behind it is.

Two sources are named in the file: the 2009 paper A Behavior Model for Persuasive Design, presented at the Persuasive conference in Claremont, California, and the 2019 book Tiny Habits.

Why is that useful for an AI assistant? Because a behaviour question without a model almost always gets the same answer. Ask any random chatbot why nobody uses your new feature, and you get a list of suggestions: communicate more clearly, show the value, send a reminder, make it more attractive. That is not a diagnosis, that is a grab from the drawer. This skill forces a different route. First the behaviour is made so concrete that you can see it happening, in the form who does what at what moment. Then all three elements each get their own score. Only then may an intervention follow, and that intervention belongs to the lowest score.

Anyone who works with other people's behaviour, or struggles with their own, gets something out of it: product people who see a feature not being picked up, marketers who cannot get a conversion rate up, founders who watch their onboarding leak, coaches and team leaders who want a new way of working to land, and people trying to build a habit of their own. Three worked examples in the file show exactly that spread: a product onboarding, a personal reading habit and a sequence of cold sales emails. It is part of the library of 100 free skills NL we make available through the AI and automation service, with no account and no sales email afterwards.

One thing you should know beforehand: the skill analyses and designs, it does not carry anything out. The scores come from what you describe, and the file does not prescribe a measurement method. The sharper you describe the moment at which the behaviour should happen, the more usable the diagnosis. A vague question produces a vague score, and that is exactly why step 1 exists in the file.

The behaviour is not concrete enough to score. More engaged users is not a behaviour, that is a wish. Opening the app every morning is a behaviour, because you can see whether it happened. Hence the requirement: who does what at what moment. Without that, the skill does not go a single step further. That looks like a formality, but it is the whole analysis: without a moment you cannot judge Prompt, and without a concrete action you cannot judge Ability.

The three elements get lumped together. A user who drops out at an import screen and a user who never opened the email have both failed to do the behaviour, but their problem is the opposite. The first has an ability problem, the second a prompt problem. Anyone who puts them together into one score then picks a solution that does nothing for half the group. That is why the skill always scores all three separately, even when the answer looks obvious.

Motivation gets pulled on while the barrier is the actual problem. This is the most expensive mistake, and the file turns it into a hard rule: no intervention on Motivation while Ability is below 5. Fogg's reasoning is that motivation swings by the hour, by the day and by mood, while ability becomes stable as soon as you genuinely make the behaviour easier. A campaign that temporarily pumps up motivation just moves the problem to next week. A form that goes from eight fields to two stays shorter. Dull work, and it does hold up.

The behaviour is too big to start with. Thirty minutes of meditation, adding a lead every day, an hour of reading: all goals where ability is low at the moment it needs to happen. That is why the file shrinks every behaviour down to the smallest version that still counts, with three breaths and reading a sentence as examples. Small behaviour has much higher ability and is therefore carried out far more often, and it grows by itself once it succeeds.

The reward comes too late. Points, badges and a discount next month do little for habit formation, according to Fogg. The brain anchors behaviour through positive emotion right after the action, something he calls shine. In a product that is a micro celebration in the interface, for a personal habit a spoken yes or a smile. That celebration is a required block in the output, not a suggestion, and that is a deliberate choice.

In diff form, with lines from the examples in the file: what goes out and what comes back in its place.

Then comes the core in seven steps. Step 1 pins down the target behaviour in the form who does what at what moment, and explicitly rejects vague goals. Step 2 scores each element of MAP from 1 to 10, with an explanation per element of what the bands mean.

For Motivation, 1 to 3 stands for seeing little value, 4 to 6 for doubt, and 7 to 10 for genuinely wanting it. For Ability, 1 to 3 stands for many barriers, 4 to 6 for moderately difficult, and 7 to 10 for practically no barrier. For Prompt, 1 to 3 stands for no trigger or bad timing, 4 to 6 for a trigger that regularly misses, and 7 to 10 for the right signal at the right moment.

Under Ability, the file names Fogg's six elements of simplicity, and these turn the score into something usable rather than a feeling: time, money, physical effort, mental effort, social deviance and routine. If the behaviour costs little time and money but asks someone to deviate from what the group does, that is an ability problem, not a motivation problem. That distinction is exactly where analyses without a model fall apart.

Step 3 plots the behaviour on the action curve and forces a call on the balance: is this high motivation with low ability, like quitting smoking, is it low motivation with high ability, like doing two push-ups, or does it sit balanced above the Action Line. Step 4 names the lowest-scoring element as the bottleneck and forbids improving all three at once. That comes with Fogg's rule that colours the rest of the answer: tackling ability is almost always more effective than tackling motivation, because motivation fluctuates and ability is stable as soon as the behaviour is genuinely easier.

Step 5 designs two to three concrete interventions per element, and the file provides a repertoire list for each element. To raise ability: cut steps or fields, shorten the time, lower the cognitive load with defaults and pre-filling, make the behaviour socially normal, and attach it to an existing routine, which is called habit stacking.

To raise motivation, explicitly named as a last resort: make the short-term benefit visible, use social proof or norms, scarcity or loss framing, and identity framing along the lines of people like me do this. To design a prompt, the file distinguishes three kinds: an action prompt such as a button or an explicit question, a context prompt such as a notification or email at the right moment, and a person prompt where another human reminds you. The Tiny Habits method works mainly with anchor moments: attach new behaviour to something that already happens, with the example after I make coffee, I do two push-ups.

Step 6 shrinks the behaviour down to the smallest version that still counts. Not thirty minutes of meditation but three breaths. Not adding a lead every day but opening the lead tool once. Step 7 adds the celebration, based on the research that habits anchor mainly through positive emotion right after the behaviour. No reward later, no points system, but an immediate internal celebration that Fogg calls shine, or in a product a micro celebration in the interface at the moment of success.

After that, the file fixes the output into nine blocks with set names, from BEHAVIOUR to EXPECTATION, so that every answer looks the same and you can lay two analyses side by side. The file closes with three fully worked examples, a list of five things the skill never does, and three sources. If you want to learn to build a file like this yourself, writing a SKILL.md NL explains the structure step by step.

The second is Tiny Habits, the book from December 2019 published by Houghton Mifflin Harcourt, which translates the model into a method: shrink the behaviour, hang it on an anchor and celebrate it straight away. The third is the Stanford Behavior Design Lab, the lab Fogg founded and where the research continues.

Why ability comes before motivation. This is the point where the model clashes most with how work actually gets done in practice. Motivation is visible, easy to discuss and easy to budget for: a campaign, a mailing, an incentive. Ability is dull work: cutting fields, merging steps, setting defaults, making a choice binary. Yet according to Fogg that is where the return sits, because a lowered barrier stays in place while a motivation spike fades away. Translated into a rule the skill never breaks: no intervention on motivation while ability is below 5. That feels strict, and it is.

Why the prompt is a factor of its own. Many behaviour models treat the trigger as a detail of execution. Fogg makes it one of the three factors, and the skill follows that. A prompt can exist and still be worthless because it arrives at the wrong moment. In the third example in the file, about cold sales emails, the prompt actually scores relatively high, because the email gets opened. The behaviour still does not happen, because the action being asked for takes too much thinking. You never arrive at that kind of analysis without separate scores.

Where the model reaches its limit. The Fogg model describes the moment at which behaviour does or does not take place. It does not describe how a habit wears in over months, how an organisation moves along, or how you sell a change to a team. There are other frameworks for that, and they have their own skill in the same library.

For habits over a longer time there is the Atomic Habits Coach skill, which works with the same building blocks but places the emphasis on repetition and identity. For behaviour change at scale, where it is about an audience rather than one user, there is the EAST Framework Behaviour Coach skill. And for the hooks a product itself builds in to get people coming back, there is the Hook Model Builder skill, which pulls the trigger from this model through into a whole loop.

Even without installing the skill, you can use these principles to frame your own question better: make the behaviour concrete, give the three elements a separate score, and start with the lowest one.

If the installation gets stuck anywhere, or you want the steps with screenshots, the full explanation is at installing Claude skills NL. After that, using it is simple: describe the behaviour you want to see, what you have already tried, and where people drop off. The more concretely you describe the moment, the sharper the scores. If you are new to this kind of work and want the broader context first, you will find the explanation of working with AI in the knowledge base.

There are also situations where you are better off leaving it alone. The first is a question without a moment. More customer focus in the team is not a behaviour, so there is nothing to score, and the skill will point that out to you before going any further. Anyone who first wants to work out which behaviour actually matters at all is better off starting at the stage before this one: for a product or service that is the AARRR Pirate Metrics Coach skill, which works through the funnel and points to the stage where growth is leaking. Only once you know which step costs the most does it make sense to break that step down behaviourally.

The second limit is scale. The Fogg model looks at one person's behaviour at one moment. If your question is about an organisation that needs to get from A to B, with interests, resistance and a timeline, this is not the right instrument. That is where the change frameworks belong: the Kotter Change Coach skill for the process, and the Switch Framework Change Coach skill for the tension between reason, feeling and environment. If the resistance sits mainly in the sphere of status and certainty within a team, the SCARF Model Coach skill NL is a better way in.

The third limit is a matter of substance, and we want to say it out loud: the file contains no rule whatsoever on healthcare, addiction or clinical behavioural problems. The model is used in the smoking-cessation example only to explain the curve, not to design a treatment. If it concerns behaviour with a health risk or mental health issues, that calls for a professional and not a 1,827-word instruction file.

One more honest limit: the skill makes your analysis better, not your product. If the behaviour you want to see has no value at all for the user, a lower barrier will not change that much. What it does do is stop you pulling on the wrong element for months.

A lower barrier only helps if someone is standing in front of it. If too few people come in, the real work sits before this model: getting found by the people you are looking for. The free SEO scan shows in a few seconds where your site stands. And if you want to talk through what else AI can do for your business, from stand-alone skills to full automation, we simply do that in a conversation.