How to run a sales roleplay session

Roleplay is the most common piece of sales training that never actually happens. Mindtickle's State of Sales Readiness 2022, built on activity from more than a million users at 350 companies, found that reps submitted two to three recorded roleplays a year on average, that top-performing companies managed four, and that 85 percent of reps were coached on open deals while only 24 percent were coached on long-term skills. Two years later the same company reported that top performers get coached four times as much as everyone else. The exercise works. It just rarely happens, and when it does, it is often run in a way that puts people off coming back.

Two conditions decide whether the session teaches anything

In 1993 Anders Ericsson and his colleagues set out what separates deliberate practice from work, play and watching others: a task designed around a specific weakness, immediate and informative feedback, and repetition of the same or a similar task. Ericsson later disowned the popular 10,000-hour rule that had been built on his data. It is not the hours that count but what happens inside them. For a roleplay session that means three things most sessions lack: one moment at a time, feedback within a minute of the round ending, and the same scenario once more before moving on. Our article on deliberate practice covers why the conditions work. This one is about how to run the room.

The second condition is the room itself. In 1999 Amy Edmondson defined psychological safety as a shared belief within a team that it is safe to take interpersonal risks, and showed in a study of 51 work teams that safety went hand in hand with team learning, with the leader's coaching among the structures that shaped it. Google's Project Aristotle, which studied 180 teams, 65 of them sales pods, ranked psychological safety first among five dynamics. A roleplay is an interpersonal risk: you make a fool of yourself in front of colleagues. Without rules for how feedback is given, the session becomes a reason to schedule customer meetings at the same time.

The protocol: 25 minutes, three roles

The session has one rep who practises, one colleague who plays the buyer from a written brief, and one observer who counts a single behaviour. The manager keeps time, leads the feedback and does not play the buyer. The order never changes.

1. Two minutes: the manager names the moment being practised today and the behaviour being counted, for example 'ask for the meeting before you summarise'.
2. Two minutes: the buyer reads the brief silently (mandate, mood, hidden goal, opening line). The rep is told the situation only, never the brief.
3. Eight minutes: the round. The manager stops it at eight wherever the conversation is. Nobody interrupts, nobody laughs out loud, the observer takes notes.
4. Three minutes: the rep goes first. 'What did you do well, what would you do differently?' Then the observer with the counted behaviour, then the buyer with one sentence on how the call felt from the other side.
5. Eight minutes: the same scenario again, same rep, with one behaviour changed. This is the round that teaches something, and the one most often skipped.
6. Two minutes: the manager writes down the behaviour, the result in round two and what the rep will work on before next time.

Five rules for the feedback

• The rep assesses themselves first. It takes the edge off the defensiveness and shows the manager how self-aware they are.
• One strength and one change, never a list. Nobody learns from more than two points in three minutes.
• Behaviour, not person: 'you gave the price before the buyer asked', not 'you seemed unsure'.
• No scoring in front of the group until trust exists. For the first months the observer counts, nobody grades.
• The buyer plays it reasonable. An impossible buyer teaches the rep nothing except how to give up, and the brief should say how hard the buyer is allowed to be.

The cadence that keeps it alive

One session a week, 25 minutes, same time slot, with reps rotating, gives every person in a team of six a round of their own every sixth week and five rounds as observer or buyer in between. Add one recorded assessment per quarter, with the rep choosing the scenario, and the team already sits above Mindtickle's benchmark of four roleplays a year at the best companies. Highspot reported in 2025 that managers themselves claim 13 hours of coaching a week, while the Sales Management Association's 2015 estimate put actual coaching at around 36 minutes per rep per week. The gap between those figures comes down to what counts as coaching. Twenty-five minutes of roleplay is coaching in Ericsson's sense. The pipeline review is not.

There is data on the payoff. Korn Ferry reported in May 2026 that organisations with strong coaching processes achieve 14 percent higher quota attainment, 15 percent higher win rates and nearly 20 percent lower voluntary turnover. Salesforce's seventh State of Sales finds that 75 percent of reps say they are more likely to hit their targets with a coach or mentor. And in Sweden there is one more reason. Gallup puts engagement among Swedish employees at 25 percent, against 12 percent for Europe, and estimates that managers account for at least 70 percent of the variance in engagement. A weekly session where the manager leads rather than demonstrates is one of the few things that shows up in that number.

Four ways to kill the session

The manager plays the buyer and wins, or the buyer gets told to 'be difficult' instead of getting a brief. Feedback turns into a rundown of everything the rep did wrong, and round two gets cut because time ran out. All four feel reasonable in the moment, and all four make sure nobody wants to come back next week. Anyone who wants to demonstrate can do it in a round of their own, against a colleague, under the same feedback rules as everyone else.

The weekly session belongs to the team. Between sessions the rep needs to be able to run the same scenario alone, without booking a colleague, and that is where AI roleplay fits. In Salesprep the rep meets an AI buyer that plays from a brief, and every call is scored on six or seven metrics drawn from a library of ten, with a written comment per score that the rep can bring to the next team session. Three calls in the Cold call module are free, no card required.