Last updated: 10 November 2026
Sales managers can use AI to coach more people with less preparation. An assistant can turn call notes into a coaching focus for each salesperson, play a buyer in short practice rounds, spot patterns across the team and draft feedback that you edit before you give it. The manager still decides what good looks like and has the conversation.
Written for sales managers and team leads who coach several people at once, the prompts below work in ChatGPT, Claude, Google Gemini and Microsoft Copilot. The examples come from an invented sales team at a company that sells inventory software to small retailers. If you want to build the coach itself, start with our guide to building an AI sales coach for objection handling and deal strategy. For the design of the session, see our guide to conducting an effective sales training. For the full sales process, see our practical guide to AI for sales.
Why do sales managers struggle to find time to coach?
Most managers spend their week on forecasts, pipeline reviews and customer escalations, so coaching is the first job to be dropped. When coaching does happen, it often follows the last deal that went wrong, and each salesperson gets different advice depending on what the manager remembers. AI cannot give you more hours. It can cut the preparation for each conversation, so that coaching fits into the time you already have.
What do you need before you start?
Four things need to be in place first.
- A scorecard. A one-page list of the skills that matter in your sales process, each with a plain description of what good looks like. For example: opens the call with a clear purpose, asks about the buyer’s situation before presenting, handles an objection without arguing, and agrees a specific next step.
- A shared context pack. The product facts, customer types, common objections and the best answers your team has used. Our coach guide shows how to build one.
- An agreement with the team. Tell your salespeople what you will use AI for, what data goes in and what you will not use it for. Coaching works better when people trust the tool.
- A regular slot. A fixed time each week, even a short one.
How do you prepare a one-to-one with AI?
Give the assistant the material about one salesperson and ask for a coaching plan for your next conversation. Use call notes written by the rep, a deal summary or a call transcript that you are allowed to use, and remove customer names and contact details first.
You are an experienced sales coach. Here is our scorecard: [paste scorecard]. Below are notes from [rep]’s last [number] calls, with customer names removed. For each scorecard skill, say what the notes show and quote the line that supports it, or say there is not enough evidence. Then suggest one skill to work on in our next one-to-one, two questions I can ask [rep] so that they find the issue themselves, and one practice exercise. Do not rate the person, and do not guess at things the notes do not show.
[Paste the call notes here]
Read the plan before the meeting and cut anything you disagree with. The plan helps you arrive prepared, and your own judgement of the person comes first.
How do you run a practice round?
Short rounds work better than long workshops. A ten-minute round fits into a team meeting or a one-to-one.
| Step | Time | What happens |
|---|---|---|
| Set up | 1 minute | Pick one objection or situation from this week’s deals |
| Role play | 5 minutes | The rep sells to the AI, which plays a buyer, using the context pack |
| Score | 2 minutes | The rep scores themselves against the scorecard, then the AI scores against the same card |
| Debrief | 2 minutes | The manager picks one thing to repeat next time |
You are playing the [buyer role] at a [type of company] in Singapore. I am a salesperson at [company]. Context: [paste context pack]. Raise this objection: [objection]. Stay in character and be sceptical, as a real buyer would. After five exchanges, stop and score me against this scorecard: [paste scorecard]. Give one thing I did well and one thing to change.
Let colleagues play the buyer in some rounds, so that people hear how real buyers phrase an objection. The AI tends to be friendlier than most buyers.
How do you spot patterns across the team?
One-to-ones show one person’s gaps. Looking across the team shows what to train together. Collect short call summaries or scorecard results from several reps, remove names and customer details, and ask the assistant for patterns.
Below are scorecard results and short call summaries from our team, with names and customer details removed. Which skills are most often weak? Which objections come up most? For each pattern, say how many calls it appears in and how confident you are, given the small sample. Suggest one team exercise for the most common gap.
[Paste the summaries here]
Check the counts yourself, because assistants can miscount. Use the result to choose a team exercise, and avoid using it to rank individuals. For the pipeline side of a manager’s week, see our guide to AI sales forecasting for small teams.
How do you give feedback that people accept?
Good feedback names a behaviour, shows the evidence, says what effect it had, suggests one action and sets a time to check. An assistant can draft this in your voice, and you then edit it so that it sounds like you and says only what you believe.
Draft feedback for [rep] in my voice, which is [direct and warm]. Behaviour: [what happened]. Evidence: [the line or moment]. Effect: [what it did to the call]. One action to try: [action]. Follow-up: [date]. Keep it under 120 words and do not exaggerate.
Deliver the feedback in person or by voice where you can. A written draft is a starting point for a conversation.
How can AI help new salespeople ramp up?
New hires can rehearse the main objections and the standard discovery questions before they speak to a customer. Build a short practice plan from the scorecard and the context pack, and have the new hire complete a round each day in the first weeks. Our guide to conducting an effective sales training covers onboarding handbooks and buddies, and AI practice rounds fit alongside them.
What should managers keep for themselves?
Decisions about people stay with the manager. Performance ratings, promotions and pay, hard conversations and motivation need a human who knows the person and the context. The AI does not know that a rep had a difficult quarter for reasons outside their control, or that a deal was lost to a price war. Use its output to inform your view, and do not use it as the only basis for judging someone.
How do you handle data and trust?
Coaching material can contain customer details and information about your own staff. Remove names and contact details before you paste anything. Check the tool’s data terms and your company’s policy on what may go into an AI tool. If you record or transcribe calls, check that customers and staff have been told, and that your use fits the purposes you told them. Our PDPA checklist for AI and customer data covers the main points. This is general information and not legal advice.
What are the limits of AI coaching for managers?
- Scores can differ between runs. Use the same scorecard each time and read the evidence behind a score.
- The assistant only knows what you give it. Thin notes lead to thin coaching.
- It reads text, so tone of voice, pauses and body language need your own judgement.
- It can miscount patterns across calls. Check the numbers before you act on them.
- It cannot replace the relationship. A rep who trusts you will take feedback that a tool cannot deliver.
Frequently asked questions
Can AI replace sales coaching by a manager?
No. AI can prepare your one-to-ones, give reps something to practise against and show patterns across calls. A rep still needs a person who knows them and the market, gives honest feedback and decides what good looks like.
How often should a manager run practice rounds?
A short round once or twice a week is enough to build a habit, for example ten minutes at the start of a team meeting. Regularity matters more than length, so choose a rhythm your team can keep.
Is it fair to score salespeople with AI?
Use the AI score as a practice aid and not as a performance rating. Share the scorecard in advance, use the same card for everyone, read the evidence behind each score and make performance decisions yourself.
What data should I give the assistant?
Give it your scorecard, your context pack and call notes with customer names and contact details removed. Leave out anything you would not be comfortable showing to the rep or the customer, and check the tool’s data terms first.
Which AI tool is best for sales coaching?
Any of the main assistants, such as ChatGPT, Claude, Google Gemini or Microsoft Copilot, can handle the prompts above. Choose one your company allows, check its data settings and compare two tools by giving each the same call notes and the same prompt.
Where does this guide fit in our AI for sales series?
Coaching is the last stage of selling with AI, and this article looks at it from the manager’s side. Start with the practical guide to AI for sales, build the coach itself with our guide to an AI sales coach for objection handling and deal strategy, and plan the messages your team sends with our follow-up cadence guide. We also teach AI-assisted coaching in our two-day Mastering AI for Sales Growth and Team Dynamics course.
Coach your team with AI in class
QD Academy’s two-day Mastering AI for Sales Growth and Team Dynamics course covers AI-assisted coaching and deal acceleration, taught by practitioners and funded for eligible learners. Check the course page for the latest dates, fees and funding.