Most people start a coaching conversation well and lose their way in the middle. Across 825 scored coaching practice sessions on VirtualSpeech, people scored 7.6 out of 10 on setting the goal and exploring the situation, but 6.6 on generating options. Two-thirds scored 8 or higher on the opening stages; only a third did on options.
The GROW model is the most widely used structure for coaching conversations, and it is simple to learn. The hard part is using it in a live conversation, especially the moment you have to stop exploring the problem and help the other person decide what to do.
This guide explains the four stages of GROW, gives you questions for each one, walks through a worked example, and then shows what our practice data reveals about where people get stuck and what helps.
What is the GROW model?
The GROW model is a four-stage structure for coaching conversations: Goal, Reality, Options, and Way forward. It was developed in the UK in the 1980s by Graham Alexander, Alan Fine and Sir John Whitmore, and popularised in Whitmore’s book Coaching for Performance.
Its core idea is that the coach asks rather than tells. The person being coached does most of the thinking, so they own the outcome.
- Goal. Agree on what the person wants from this conversation and from the situation. A clear goal stops the conversation drifting.
- Reality. Explore where things stand now: what has happened, what they have tried, and what is getting in the way. This is about facts and perspective, not judgement.
- Options. Help the person generate several possible ways forward, including ones they had dismissed. The coach resists offering the answer.
- Way forward. Agree a specific next step: what they will do, by when, what support they need, and how committed they are.
The stages are not always linear. Good coaches loop back, for example, from Options to Reality when a new obstacle appears. But every good GROW conversation ends with a Way forward the person has chosen.
GROW model questions for each stage
Open questions do the work in GROW. Start with “what” and “how”, and avoid questions that hide a suggestion, such as “Have you thought about…?”
Goal
- What would you like to get out of this conversation?
- What would a good outcome look like in three months?
- How will you know you have achieved it?
- Why does this matter to you now?
Reality
- What is happening at the moment?
- What have you tried so far, and what happened?
- What is getting in the way?
- How would the other people involved describe the situation?
Options
- What could you do?
- What else? (Ask it twice. The second or third answer is often the best one.)
- What would you do if you knew it would work?
- What would someone you admire do here?
- What are the pros and cons of each option?
Way forward
- Which option will you go with?
- What is the first step, and when will you take it?
- What might stop you, and how will you handle that?
- On a scale of 1 to 10, how committed are you? What would make it a 10?
- What support do you need, and from whom?
A GROW model example
Priya is a team lead. Sam, one of her analysts, has missed two report deadlines and seems overwhelmed. Here is a short GROW conversation, with notes on what Priya is doing at each stage.
Goal
Priya: What would you like to get out of our chat today?
Sam: I want to stop missing deadlines. It’s stressing me out.
Priya: What would “on top of it” look like by the end of the month?
Sam: Reports in on time, without working every evening.
Priya turns a vague worry into a goal Sam can measure.
Reality
Priya: Talk me through a typical week. Where does the time go?
Sam: Monday and Tuesday disappear into ad hoc requests from sales. I start the reports on Wednesday and I’m always behind.
Priya: What have you tried so far?
Sam: Saying yes faster, honestly.
She asks about facts and stays neutral. No judgment about the missed deadlines.
Options
Priya: What could you do differently?
Sam: Block Monday mornings for reports.
Priya: What else?
Sam: Ask sales to send requests through a form, so I can batch them. Or agree with you which requests can wait.
Priya: Any others, even ones that feel unrealistic?
Sam: Hand the weekly summary to the new analyst.
This is where most people stall. Priya keeps asking “what else?” instead of picking an option for Sam.
Way forward
Priya: Which of those will you start with?
Sam: Blocking Monday mornings, and I’ll draft the request form by Friday.
Priya: What might get in the way?
Sam: Urgent requests from the sales director.
Priya: How would you like to handle those?
Sam: I’ll ask you to check with her if it’s a real emergency. Can we review it in two weeks?
The conversation ends with Sam’s own plan, a date, a likely obstacle, and a check-in.
Where people get stuck: Options
Options is the hardest stage of GROW. In 86% of the 825 practice sessions we analysed, people scored lower on Options than on Goal and Reality. Where one stage was clearly the weakest, it was Options 60% of the time, Way forward 37%, and Goal or Reality just 3%.
| Skill scored | Average score (out of 10) | Sessions scoring 8+ | Sessions scoring 5 or below |
|---|---|---|---|
| Goal + Reality | 7.6 | 66% | 7% |
| Options | 6.6 | 34% | 17% |
| Way forward | 6.8 | 46% | 17% |
| Active listening | 6.9 | 33% | 15% |
| Staying non-directive | 6.8 | 43% | 27% |
The pattern held in every scenario. In open-ended coaching, people averaged 7.3 on Goal and Reality and 6.3 on Options. In a conversation about giving difficult feedback, it was 7.8 and 6.7. In mentoring, 8.0 and 7.1.
Why Options? Goal and Reality feel natural: you ask questions and listen. Options asks you to hold back the answer you can already see and help the other person find their own. Staying non-directive has the most low scores of any skill (27% of sessions at 5 or below), which suggests many people slip into giving advice at exactly this point. When they do, the other person gets one option, the coach’s, rather than several of their own.
What helps: practise the same conversation twice
A second attempt improves the weakest stages most. Among the 141 people who practised the same coaching roleplay twice, Options rose from 5.2 to 6.9 and Way forward from 5.2 to 6.9. Three in four improved on Options.
One caution: people who choose to retry tend to have scored lower the first time, so part of the gain is a natural rebound. Even so, the largest jumps came on the two stages people find hardest, which is what you would expect if specific feedback is doing its job.
The catch is that most people never get a second go. Only 22% of people repeated a coaching scenario. If you run coaching skills training, building in a required second attempt is one of the simplest changes you can make.
Three habits also help with Options and Way forward:
- Ask “what else?” at least twice before discussing any option. It slows you down and widens the list.
- Notice the urge to advise. If you have an idea, ask permission and offer it as one option among several: “Would it help if I shared something I’ve seen work?”
- Make the Way forward concrete. A good close names the action, the date, a likely obstacle and when you will check in. “I’ll think about it” is not a Way forward.
Using the GROW model as a manager, and teaching it
For managers
- Use GROW for development, not discipline. It works when the person owns the problem and the solution: a stretch goal, a stuck project, a career question. If you need a specific behaviour to change, start with clear feedback first, then use GROW to plan the change.
- Share the structure. Telling someone “Let’s agree on what you want, look at where things are, then explore options” makes the conversation feel less like an interrogation.
- Spend longer in Options than feels comfortable. It is the stage most people rush.
- Follow up. A Way forward without a check-in date rarely happens.
For L&D teams
- Teach the model in minutes, then practise for hours. The structure is easy to explain. The skill, especially holding back advice, only develops with repetition.
- Score the stages separately. A single overall score hides the pattern above. Someone with a strong opening and a weak Options stage needs different help from someone who can’t set a goal.
- Require a second attempt. Our data suggests it lifts the hardest stages most, yet few people retry unless it is built into the programme.
Frequently asked questions
What does GROW stand for?
Goal, Reality, Options and Way forward. Some versions use “Will” or “Wrap-up” for the final stage. All mean agreeing a committed next step.
Who created the GROW model?
Graham Alexander, Alan Fine and Sir John Whitmore developed it in the UK in the 1980s. Whitmore’s book Coaching for Performance made it widely known.
Which stage of the GROW model is hardest?
Options. In 825 VirtualSpeech practice sessions, people averaged 6.6 out of 10 on Options, against 7.6 on Goal and Reality. Only 34% scored 8 or above on Options.
How long should a GROW coaching conversation take?
Anything from 10 minutes for a quick check-in to an hour for a bigger decision. The stages stay the same; what changes is how long you spend in Reality and Options.
Can you use the GROW model for performance conversations?
Yes, once the issue is clear. Use direct feedback to name the problem, then use GROW to help the person plan how to fix it. GROW works poorly if the person doesn’t yet accept there is a problem.
How do you get better at GROW coaching?
Practise the full conversation, get feedback on each stage, and repeat it. In our data, a second attempt lifted Options scores by 1.7 points on average.
How we did this analysis
We analysed 825 completed coaching roleplays by 273 people on VirtualSpeech between November 2025 and October 2026. Each roleplay was a voice conversation with an AI character, scored automatically against the GROW model.
- Scenarios: open-ended coaching, giving difficult feedback, mentoring, and coaching a disillusioned staff member.
- Who: people at seven organisations. Most were business school students practising coaching skills; the rest were managers and staff at enterprise and public-sector clients.
- Scoring: each session received a score from 0 to 10 on five criteria: Goal and Reality (scored together), Options, Way forward, active listening, and staying non-directive. Sessions in Norwegian, Indonesian, French, and Japanese were mapped to the same criteria.
- Second attempts: we compared each person’s first and second attempt at the same scenario. 141 people repeated a scenario.
- Limits: scores come from AI, not human assessors. We are running a study comparing our AI scores with certified coaches and will publish the results. Most participants were students, and the people who retried chose to, so their results may not represent everyone.
All data was anonymised and aggregated. No individual results are shown.
Practise GROW coaching conversations
VirtualSpeech lets managers and students practise coaching conversations with AI characters, in the browser or in VR. Each session is scored on every GROW stage, so people can see exactly where they get stuck and try again.
Related reading:
