The AI EAST Framework Behaviour Coach skill for Claude

The problem the skill solves is the problem almost every plan runs aground on: people say yes and do no. The form does not get filled in, the email gets opened but not answered, the new way of working gets embraced in the meeting and forgotten by Monday. The first reflex is then almost always communication: another email, a clearer text, a more urgent tone. EAST looks at something else. Not at the message but at the effort, the attention, the norm and the moment. Four levers you can adjust one by one without forcing anyone into anything.

This coach is written for anyone trying to change other people's behaviour without having the power to enforce it: marketers, business owners, product people, HR, service providers waiting on their quotes, and anyone who designed a process that colleagues do not stick to. It is part of the skill library NL that we make freely available through our AI and automation service, with no account and no sales email afterwards.

There is one thing you should know upfront, and the file says so itself: EAST is a practical starting point, not a replacement for research. If you fundamentally do not know why people drop off, the skill advises talking to five people from your target audience first. Nudges based on the wrong hunch mostly make the wrong thing easier.

It costs more effort than you think. Every extra step, every extra field and every unclear instruction costs behaviour. That is literally the core sentence for the E. Opening a PDF on a phone, scrolling back to the right page, retyping an amount into an email: it seems like nothing, and yet at every one of those moments some people drop off. That is why it first looks at where the effort sits and whether a default option can really do the work, because according to the file default options do a lot of work.

It does not stand out, or it is not about them. The A stands for attention and personal relevance. An email that opens with your own company name asks the reader to work out for themselves what is in it for them. An email that opens with the result they wanted does that work for them. In the skill's example that is literally the facade that looks smart again, rather than the name of the painting company.

Nobody shows that it is normal. People do what other people do. That is how the S works, and it is immediately the most sensitive letter, because this is where the temptation to exaggerate lies. The file is strict about this: social proof is only allowed with a real, verifiable source, otherwise the S line is left out. Better no social proof than invented social proof. Anyone who wants to understand that distinction more broadly finds the psychology behind it in the Cialdini Persuasion skill, which draws the same line.

The request comes at the wrong moment. The T is the letter that gets skipped most often, even though the same request works at one moment and not at another. Find the moment when someone is already dealing with it anyway, the file says. In the example that is the moment the quote is opened, or just after the weekend when people discuss their jobs around the house. That last point is explicitly an assumption, and it is written down as one.

The difference between a nudge and a trick is not in the technique but in the direction. Both lower a threshold. One lowers the threshold towards something the other person actually wants, the other lowers the threshold towards something you want and they do not. That is why the method has a separate step for it: checking whether the goal is in the target audience's own interest, and stopping if the answer is no.

In diff form, with lines from the skill's example: what goes out and what comes back in its place.

Next comes a section on the theory the skill rests on, and it is remarkably honest for an instruction file. It names the Behavioural Insights Team, the behavioural team that started within the British government in 2010, and the 2014 guide in which EAST was published. And it immediately adds the limitation: the framework summarises insights from behavioural science into four testable questions and is meant as a practical starting point, not a replacement for research. Each letter has a short definition that the skill then uses literally.

The core is a list of six things the skill always does. It first pins down the behaviour goal precisely: who has to do what, where, and what that person does now. It checks whether the goal is in the target audience's own interest, and stops if it is not. It works through the four letters one by one with a maximum of two targeted questions per letter.

For each letter it delivers two to three concrete adjustments, framed as an experiment and not as a guarantee. It flags every assumption about the target audience as an assumption, with the question you use to test it. And together with you it picks a first adjustment that is achievable this week, plus a simple measurement: a before and after reading of the same figure.

Before it delivers anything, the skill runs through a mandatory five-point input checklist: the desired behaviour in one sentence, what the target audience does now instead, the contact moment or channel, what has already been tried, and why the desired behaviour is in the target audience's interest. If something is missing, it only asks about that, a maximum of two questions at a time. So you do not get a fifteen-field intake form, just a handful of targeted questions at most.

The output format is then fixed in seven blocks: BEHAVIOUR GOAL, the four letters listed one below the other with their adjustments, THIS WEEK with the first action and the measurement, and DO NOT with what the skill advises against in this situation and why. That last block is easy to underestimate. Advice that only says what to do leaves you with the same twenty options; advice that rules a few out loud makes the choice smaller.

Right at the bottom is a list of six things the skill never does, plus four sources. Those refusals get their own figure further down this page, because they are the sharpest part of the file.

EAST itself comes from the Behavioural Insights Team. The source list cites the 2014 guide, EAST: Four Simple Ways to Apply Behavioural Insights, with authors Service, Hallsworth, Halpern, Algate, Gallagher, Nguyen, Ruda and Sanders. The team started within the British government in 2010 as a small behavioural team. The design of the guide also explains the shape of the skill: four letters you can remember, each with a handful of practical questions, meant to be applied quickly to real policy and real forms.

The thinking around it comes from Thaler and Sunstein. Nudge from 2008 introduced the idea that the way a choice is presented steers behaviour, and that no neutral presentation exists. Every form has an order, every email has a first sentence, every setting has a default value. The question is not whether you steer but towards where. That is why the interest check sits at the front of the skill and not as a footnote.

The case for effort comes from Kahneman. Thinking, Fast and Slow from 2011 describes why people choose the path of least resistance for almost every action, and why small thresholds have a disproportionately large effect on what happens. That is the reason the E comes first: making an action easier usually costs you less than persuading someone.

The boundary around the S comes from Cialdini. Influence from 1984 describes the social proof principle, and the file explicitly cites him for its limits too. Social proof works because people read it as information. The moment it is not true, it is no longer information but advertising, and then it works exactly the other way round. Hence the rule that an S line without a verifiable source is simply dropped.

Even without installing the skill, you can improve your own process with these four questions: where does the effort sit, is it about them, is there a real norm to show, and when are they already thinking about it anyway. If you want to look at the same question from a different model, the Fogg Behaviour Model Coach skill splits behaviour into motivation, ability and trigger, which often shows more sharply whether it is a matter of willingness or of capability.

After that, using it is simple: describe the behaviour you want to see, who has to do it, and what that person does now instead. The more concrete your situation, the more usable the adjustments. It asks itself about anything that is missing, a maximum of two questions at a time. If you are new to this kind of file, first read what Claude skills are NL and then how to install Claude skills NL. If you want to write one yourself, for instance for the behaviour questions that keep coming up in your own trade, writing a SKILL.md NL explains how to go about it.

It becomes less suitable the more abstract the behaviour gets. If you want your team to show more ownership or your clients to become more loyal, there is no action to make easier and no moment to choose. Translate such goals into visible behaviour first. Other skills are better suited to that translation work: the Switch Framework Change Coach skill tackles change through Rider, Elephant and Path, and the Kotter Change Coach skill is built for change that has to move through a whole organisation.

There are also situations where you are better off leaving it alone, and the file names them itself. Behaviour that works against the target audience's own interest is the first: the skill stops there. Other people's medical, financial and legal decisions are the second. And the third is the most important for everyday work: fundamental uncertainty about why people drop off. Nudges are fine-tuning. If you do not know the underlying problem, you are adjusting something that is not the problem. Five conversations with people from your target audience are then worth more than twenty ideas.

One more honest limitation that no behavioural model solves: EAST makes the desired behaviour easier, not your offer better. If your product is not right, your price cannot be explained, or your site promises something different from what you deliver, then a better email only lowers the threshold to a disappointment. What the skill does do is stop a good offer running aground on a form with one field too many.

Lowering a threshold only helps if there is someone standing in front of that threshold. If too little traffic is coming in, the real work sits one step earlier: getting found by the people looking for you. The free SEO scan shows within seconds where your site stands. If this kind of thinking works for you, take a look at what else is ready in the library of 100 free skills NL that we make available with no account needed. And if you want to talk through what else AI can do for your business, from individual skills to full automation, we simply do that in a conversation.