Why Difficult-Conversation Frameworks Fail Without Practice

Almost every management-development programme hands out a framework for hard conversations. SBI — Situation, Behaviour, Impact — from the Center for Creative Leadership. Radical Candor. GROW for coaching. DESC for pushing back. Nonviolent Communication for conflict. They are taught on the course, printed on a card, and dropped into the LMS. And they are, mostly, good models. Clear, memorable, built on sound thinking.

Then watch what happens three weeks later, when a manager has to tell a capable but coasting team member that their work has slipped. They open with the SBI opener they were taught. The employee gets defensive, or upset, or fires back with a reason it is not their fault — and the framework evaporates. The manager softens, retreats to vagueness, promises to "keep an eye on it," and leaves the real message unsaid. They knew the model. They could have passed a test on it that morning. It still did not survive contact with a live human being.

This is the pattern I want to take apart, because L&D keeps drawing the wrong conclusion from it. When a framework does not stick, the instinct is to find a better framework, or to teach the same one more thoroughly. Both miss the point. The framework was never the missing piece. Practice was.

Why do difficult-conversation frameworks fail in the moment?

Because a framework and a conversation are two different kinds of thing. A framework is a piece of knowledge — something you can hold in working memory and recite. A difficult conversation is a motor and emotional skill, performed in real time, while your own stress response is firing and the person in front of you refuses to follow the script. Knowing the steps is necessary. It is nowhere near sufficient.

Frameworks are also, by design, linear. SBI assumes you get to state the situation, describe the behaviour, and explain the impact, in order. Real conversations go non-linear the instant the other person reacts. They interrupt after your first sentence. They cry. They challenge your facts. They agree too readily and change nothing. The framework tells you what a good opening looks like; it says almost nothing about what to do in the ninety seconds after the opening lands badly — which is precisely where difficult conversations are won or lost.

And there is the pressure itself. Under emotional load, the very capacities a hard conversation needs — measured pace, listening, holding your point without hardening — are the first to degrade. A manager rehearsing SBI in a quiet training room is not the same person delivering it while their heart rate climbs and they can see the other person's face fall. The gap between those two states is not closed by understanding the model better. It is closed only by having been in something like that state before, more than once.

What does the evidence say about framework training and behaviour change?

The numbers around management capability are not the numbers of a population short on frameworks. They are the numbers of a population that has never practised. In the UK, the Chartered Management Institute has found that 82% of managers step into the role with no formal management or leadership training at all — the "accidental managers" who are promoted for being good at the job below and left to work out the people part live, in front of their teams.

Even where training exists, the honest finding from decades of research is that most of it does not transfer. Meta-analytic reviews of training transfer consistently find that only a modest share of what is taught shows up as changed behaviour on the job, and that the strongest influences on whether it does are not the content or the model — they are the work environment and the manager's own opportunity and motivation to use it. A better acronym does not move that. Rehearsal, reinforcement and follow-through do.

You can see the same thing in how managers behave around the conversations themselves. In a survey conducted by Harris Poll and reported in Harvard Business Review, 69% of managers said they are often uncomfortable communicating with employees, and 37% admitted avoiding giving feedback when they thought the person might react badly. These are not knowledge gaps. No one avoids a conversation because they cannot remember the framework. They avoid it because they do not trust themselves to hold the framework when the other person pushes — and they have never had a safe place to find out whether they can.

If frameworks are so limited, why are they everywhere?

Because they are teachable, and teachable is what a training budget can buy. A framework can be explained in a slide, demonstrated in a role-play the facilitator controls, and tested at the end of the day. It produces a deliverable — people learned a model — and a satisfied feedback form. Skill built through repeated practice is slower, messier and harder to schedule, so it quietly falls off the plan.

None of this makes the frameworks bad. SBI is a genuinely useful way to keep feedback specific instead of personal. Radical Candor names a real trap — caring personally without challenging directly. The mistake is not adopting them; it is mistaking model-literacy for capability, and treating the moment a manager can describe the framework as the moment they can use it. Those are months of practice apart, and most programmes ship the first and call it the second.

What actually builds the skill?

The same thing that builds any performed-under-pressure skill: deliberate practice. The psychologist K. Anders Ericsson, whose research defined the field, described it as focused, repeated practice on a specific aspect of performance, with immediate feedback and the chance to try again straight away. Not watching. Not understanding. Doing, getting told precisely what happened, and doing it again.

Every other field that depends on high-stakes human performance already accepts this. Pilots rehearse the emergency in a simulator before they meet it in the air. In medicine, simulation-based deliberate practice — running the procedure repeatedly against a realistic scenario, with feedback between attempts — is a well-established way to build competence before a real patient is ever involved. No one would let a surgeon operate having only read the framework for the operation. Yet that is exactly the standard we hold managers to for conversations that can end careers, trigger grievances, or push a good person out of the door.

The reason managers do not practise is not laziness. It is that their only available practice ground has always been the real conversation — and the real conversation is far too costly to experiment in. Try a new approach and get it wrong, and the damage is real: a broken relationship, a formal complaint, a lost employee. So managers default to what they already do, however poorly it works, because the risk of fumbling something new in front of a real person is worse than the slow cost of the status quo. Give them a place to fumble safely, repeatedly, with feedback, and the calculation changes.

How should L&D deploy frameworks so they actually stick?

Keep the framework — and stop treating it as the finish line. Treat it as the first ten minutes of a much longer process whose real work is rehearsal. For every model you teach, ask a simple question: where does a manager get to run this against someone who pushes back, as many times as they need, before they have to use it for real? If the answer is "in the actual conversation," you have not built capability. You have handed out a card and hoped.

This is the gap HumanVantage was built to close. Managers rehearse the exact conversations they dread against realistic AI role-play that reacts — gets defensive, gets emotional, goes off-script — so the framework has to survive something that behaves like a real person. They can retry as often as they like, in private, with no one watching and no real relationship on the line, and get structured feedback on what they actually did: whether they stayed specific, listened, held their point, kept the other person safe enough to hear it. The framework stops being something they know and becomes something they can do under pressure — which was always the point.

So before the next framework rollout, add one thing to the plan: a rehearsal loop. Give managers repeated, low-stakes, private practice at holding the model when the other person does not cooperate, and a way to see whether the behaviour is changing — not whether they enjoyed the session. A framework tells a manager what a good conversation sounds like. Only practice makes them able to have one when it matters, and their heart is going, and the person across the table has just gone off-script.