Skip to main content
Guides & How-To

AI Coaching for Sales Teams: How to Use AI to Improve Your Salespeople in 2026

AI can suggest observations from sales conversations for coaching. See how to test data quality, human review and concrete usefulness in a pilot.

Author: Agent360 editorial team
Published: 2 June 2026
Last updated: 23 October 2026
Reading time: 16 min read

How often do salespeople get coaching from their manager?

Coaching frequency varies between teams. Start by mapping which conversations the manager actually reviews and what feedback the salespeople get. AI can help select material, but the analyses must be assessed against the context of the conversation.

When the manager hears only a selection of conversations, important patterns can be overlooked. Use a limited sample to examine whether AI analysis gives relevant observations about questions, objections and follow-up in your team.

There may be learning in conversations that nobody has time to review in full. The task is to find relevant examples without turning an automatic assessment into the final verdict on the salesperson.

AI coaching can support the sales manager with suggestions for observations and focus areas. The outcome depends on data quality, lawful use, setup and the manager's review.

TL;DR:

  • AI can suggest feedback based on the conversations that lawfully form part of the specific setup
  • Assess conversation patterns such as talk time, questions, objections and next steps against the actual recordings
  • Coaching on recorded customer conversations requires a suitable and lawful data basis; simulation is another option
  • Establish a baseline and measure usefulness, errors and results in a pilot; a particular win rate improvement is not guaranteed

What is AI-based sales coaching?

AI-based sales coaching is a system that automatically analyses sales communication (primarily phone calls and video meetings) and generates structured feedback for salespeople and sales managers based on data from the conversation.

A system can use transcripts to suggest patterns and focus areas. Test the observations against the conversations and your own results; a statistical correlation is not in itself the cause of a sale.

AI analysis can supplement a manager's assessment, but it has its own conditions and limitations:

  • Data-based: Observations depend on the transcription, the model and the chosen criteria; they are not automatically objective
  • Limited in scope: Only conversations that have been lawfully recorded and processed, and that the integration actually covers, can be included
  • Resource-intensive: Higher volume requires capacity, quality control and human follow-up
  • Verifiable: Use clear criteria, and check for variation and errors across the team

When you read Gartner's material on sales, distinguish between forecasts and documented results in a particular population. Choose a pilot based on your needs, not a general expectation of adoption.

The traditional coaching problem: time and data

Here is the reality that most sales managers know but rarely admit:

The time constraint: Map how much time the manager spends preparing and delivering coaching. The number of salespeople, meetings and conversation length affect capacity; there is no single documented time norm for Danish sales managers.

Read also our guide to AI sales coaching with real-time feedback.

Record both direct coaching time and preparation over a period. Use the local baseline to decide whether AI actually saves time and whether the quality of the feedback is maintained.

The data problem: The manager may have CRM notes, selected conversations and pipeline results as data sources. They give different cross-sections and can contain errors. Compare them before an observation is used in coaching.

Examine whether relevant insight in your sales data is being overlooked, and whether that is due to time, access, data quality or other factors. That determines whether a coaching tool solves a concrete problem.

Conversation Intelligence as one option for coaching on customer conversations

For coaching on customer conversations, Conversation Intelligence can deliver transcripts and observations. It is one practical architecture, not a requirement to buy all the parts of a particular sales stack.

The chain looks like this:

Lead Intelligence: Data can be used to suggest prioritisation of prospects. Check relevance and lawful use before the suggestions are used.

Conversation Intelligence: Selected, lawfully processed conversations can be transcribed and analysed. Check what the integration actually covers, and the quality of the transcription.

Automated Admin: The system can suggest CRM notes or updates. Clarify which changes may happen automatically, and how errors are corrected.

AI Coaching: Data from Conversation Intelligence is used to generate coaching insights: what went well, what needs improving, which patterns characterise the salespeople who win the most deals?

Test coaching on the workflow and data basis you have. Compare the usefulness with the costs; no general ROI ranking between the layers is documented here.

What does AI analyse in a sales call?

AI coaching systems typically measure a combination of quantitative and qualitative indicators:

Talk ratio: The share of the conversation in which the salesperson is talking can be used as an occasion for dialogue. An appropriate ratio depends on the meeting type and the task; there is no single documented optimal percentage range for all your sales calls.

Question-to-pitch ratio: Examine which questions are asked and whether they help in understanding the customer's needs. The SPIN Selling methodology can be used as a conversation framework. AI classification of questions must be checked; a particular number does not guarantee higher conversion.

Filler words: Repetitions such as "um" or "you know" can be examined in specific excerpts. Assess whether they actually make the conversation harder to understand; a count is not a measure of the person's expertise.

Objection handling: AI can suggest categorisation of specific types of objection ("the price is too high", "we don't have budget right now", "we already use X") and provide input to a human assessment of the salesperson's strategy.

Next step: Does the call end with a clear, specific next step? "I'll get back to you" vs. "let's book 30 minutes on Tuesday at 14:00": AI can suggest recording the agreements; check the suggestions and test whether the chosen solution can support follow-up.

Themes and expressed reactions: Stick to what is actually said, and check the interpretation against the conversation. Inferring emotions from biometric data, e.g. voice, in the workplace is as a starting point prohibited under Article 5 of the EU AI Act; coaching is not in itself an exception.

Keywords and themes: What are prospects talking about? Prices, competitors, internal decision-making processes? AI can suggest patterns across calls for closer analysis. Examine the data basis and any links to sales results.

Manual coaching vs. AI coaching: a comparison

Direct comparison of traditional manual sales coaching and AI-driven coaching across key parameters

Parameter Traditional coaching AI coaching
Coverage Selected conversations and meetings The conversations that the lawful setup actually covers
Data source The manager's observation and the salesperson's information Transcription and analysis that must be verified
Basis of assessment Human judgement with possible biases Model and criteria with possible errors and biases
Feedback speed Depends on the agreed coaching rhythm Depends on processing and human review
Scalability Depends on the manager's time Depends on capacity, price and quality control
Personalisation Based on the manager's perception Based on the individual salesperson's data profile
Bias risk Must be assessed and handled Must be assessed and handled
Resource requirements Measure preparation and meeting time Measure setup, review and meeting time

AI coaching can support the manager's preparation and selection of conversation examples. Examine in the pilot whether it actually frees up time. The manager remains responsible for employee dialogue, motivation and the specific assessment of the situation.

Coaching platforms and a solution built for you

Two well-known platforms for AI coaching are Gong and Chorus:

Gong.io Gong offers Conversation Intelligence and coaching features. Test the observations and workflows on your conversations, and get a quote for the specific package. Gong's pricing page describes user licences, platform fees and the quotation process.

Gong's resources can form part of the assessment, but a supplier example is not a guarantee of your result. Measure your own baseline and subsequent development with the same definition of win rate; separate correlation from the documented effect of the implementation.

Chorus.ai (now ZoomInfo) Chorus offers conversation analysis, CRM context and deal intelligence. ZoomInfo's product description can be used to clarify the features. Compare specific quotes and workflows with Gong; no general price or feature ranking is documented here.

A coaching programme built with Agent360 We can build a workflow that gathers agreed conversation data and suggests feedback to a sales manager. The manager reviews the suggestions before they are used in coaching. Data basis, recording, language quality, integrations and storage must be clarified and tested in the specific solution.

See our comparison of Conversation Intelligence platforms for a complete walkthrough of features and prices.

How to use AI coaching data in 1:1 meetings with salespeople

AI analysis can provide conversation examples for the 1:1 meeting. Use them together with the salesperson's own explanation, and agree which observations are relevant to work on further.

The structure of an AI-informed 1:1 meeting:

1. Review AI insights before the meeting (10 min) The sales manager goes through the AI report for the salesperson: talk ratio this week vs. last week, filler word frequency, objections met and handled, deals that moved forward.

2. Choose one focus point (the most important from the data) One possible coaching approach is to choose one limited focus at a time. Prioritise together with the salesperson, and follow up on whether the exercise helps; several simultaneous focus areas can make follow-up harder.

3. Listen to a specific call excerpt Choose, for example, a couple of specific excerpts from the week's conversations. Listen to them together, and check whether the AI suggestion fits the context.

4. Coaching vs. telling Ask the salesperson how they understand the situation and what could be done differently. An AI report is a starting point that can also be wrong; the salesperson must be able to correct it.

5. Measurable commitment Agree one specific, relevant goal for the next 1:1, for example to clarify the customer's next step more clearly. Follow up on both conversation examples and the salesperson's experiences.

Cultural considerations: monitoring vs. enablement

Implementing AI coaching raises legitimate questions about sales culture and autonomy. In Denmark, where employee autonomy and psychological safety are core priorities, framing is crucial.

The wrong framing (avoid): "We are now going to monitor all your calls and put a score on you."

The right framing (use): "We want to test whether conversation analysis can give useful suggestions for coaching. You should be able to see and correct the observations, and we'll agree the purpose, access and use of the information in advance."

Three practical choices can form part of the trial. If the team uses Challenger Sale or MEDDIC, the criteria can be linked to specific behaviour examples from the method; the usefulness must still be assessed:

  1. Insight and corrections: Give the salesperson access to the relevant observations and the opportunity to explain or correct them; clarify access and rights specifically.
  2. Limited purpose: Agree how information is used in coaching, and assess any use in employee evaluation separately. Consent is not a universal shortcut to lawful processing of employment data.
  3. Voluntary start: Start with voluntary early adopters. Let them become ambassadors for the system before it is rolled out to the whole team.

Involve the salespeople in the trial and measure actual use. McKinsey's material on sales and marketing can provide perspective, but no particular adoption multiple or deadline is documented here.

Example of an implementation plan for AI coaching

Phase 1 (Weeks 1-4): Foundation

  • Clarify the lawful data basis, roles, information, access and storage; test transcription on an agreed sample
  • Integrate with the CRM so that calls are linked to deals
  • Define the coaching metrics that matter most to your team (talk ratio, next step, specific keywords)

Phase 2 (Months 2-3): Soft Launch

  • Roll out to 2-3 voluntary salespeople as a pilot group
  • Run weekly coaching sessions with the AI report as the starting point
  • Collect feedback: what is useful, what is missing, what feels irrelevant?

Phase 3 (Months 4-6): Team Rollout

  • Expand to the whole sales team based on experience from the pilot
  • Standardise the 1:1 format with AI coaching as a fixed item
  • Set team-level benchmarks: what is the average talk ratio now, and what is the target in 3 months?

Phase 4 (Ongoing): Optimisation

  • Quarterly reviews of which coaching metrics correlate with deal closing
  • Adjust the coaching focus based on what the data shows actually works
  • Identify top-performer patterns and use them as a basis for onboarding new salespeople

The timings are a planning example, not a delivery deadline or a promise of results. Expand only on the basis of the pilot's quality, use and measured usefulness.

AI coaching and onboarding new salespeople

Onboarding is one possible use of AI coaching. First define what you mean by productivity, and measure the current ramp-up period; there is no general documented reduction for all teams.

How:

Early review: Analyse agreed conversation examples and have a manager assess the observations. Combine with a Discovery call methodology if it suits your sales situation.

Personal learning plans: Use reviewed observations to suggest one focus area. A single percentage or a small sample should not decide the plan on its own.

Library of best examples: AI can suggest calls for human assessment of strong objection handling (see our guide to AI techniques for objection handling), strong openings and effective CTAs. Approved examples can be used as onboarding material once the use has been clarified.

Measure whether the duration and quality of onboarding change after the trial. Take differences in experience, market and management support into account before a change is attributed to AI.

For a complete guide to AI sales tools that support onboarding and coaching, see our guide to AI sales training: simulation vs. role-play.

Worked example: revenue and costs of AI coaching

Investment:

  • Platform: get a specific price for licences, usage, integrations and add-ons
  • Implementation: estimate data preparation, setup, testing and training
  • Ongoing operations: include quality control, administration and coaching time

Results to measure:

  • Win rate with the same definition before and after
  • Time to agreed productivity level for new salespeople
  • Deal size, segment and margin
  • Actual preparation time and feedback quality

Hypothetical worked example for a sales team of 8 salespeople:

Assume 100 opportunities per quarter, an average deal size of DKK 150,000 and a win rate of 22%. Assume, for the purposes of the example only, that the win rate becomes 27.5%; that is not a documented AI effect:

  • Current: 100 opps/quarter × 22% = 22 deals × 150,000 = DKK 3,300,000/quarter
  • Hypothetical after: 100 × 27.5% × 150,000 = DKK 4,125,000/quarter as a calculated average value
  • Calculated difference: DKK 825,000/quarter = DKK 3,300,000/year, all other assumptions unchanged
  • Assumed software cost for the example only: 8 salespeople × DKK 700/month × 12 = DKK 67,200/year; this is not a supplier quote
  • The ratio between additional revenue and software cost is approximately 49.1. This is not ROI: margin, implementation, operations and a documented cause of the change are missing.

FAQ: AI coaching for sales teams

Q: What is the difference between Conversation Intelligence and AI coaching? Conversation Intelligence can deliver transcripts and observations from customer conversations. Coaching is an application of the material, which a manager reviews. AI coaching can also be based on simulations and exercises; recorded customer conversations are not the only possible data source.

Q: May sales calls be recorded in Denmark for AI coaching purposes? Recording and AI analysis each require a specific assessment of purpose, basis, information, security and storage. The Danish Data Protection Agency describes that recording for training purposes can, under certain conditions, take place without consent, if the person called can decline the recording before or during the call. That does not determine the basis for analysing the employee's personal data or use in evaluation. Clarify these matters separately, including roles and necessary agreements; consent does not make every use lawful.

Q: Does AI coaching replace the sales manager? AI can support preparation and the selection of conversation examples. The manager remains responsible for the specific coaching and for checking the observations. Whether the solution frees up time or makes it possible to coach more people must be measured locally.

Q: How long does it take to implement AI coaching? It depends on integrations, data basis, lawful use and the need for training. Run a limited pilot with specific acceptance criteria; the example above is a planning framework, not a typical or guaranteed implementation time.

Q: What are the most important coaching metrics to start with? Start, for example, with talk time, questions and agreed next steps. Define what is relevant for different conversation types, and check the observations manually. No fixed percentage or question threshold guarantees a higher win rate.

Conclusion: from coaching as a one-off event to continuous improvement

AI coaching can form part of an ongoing coaching programme alongside conversations, exercises and human feedback. The specific way of working must suit the team.

Ongoing feedback can form part of development, but lasting behavioural change must be verified.

A trial can show whether conversation analysis gives the manager relevant examples and a better basis for coaching. Measure usefulness, errors, time spent and the salespeople's opportunity to correct the analyses.

The goal is more concrete feedback that the salesperson and manager can work with. Systematic improvement is a result that must be documented, not an automatic consequence of AI.

AI coaching can be combined with the other parts of your sales setup.

Ready to test AI coaching? Book AI consultation about your sales team.

See also our guide to win-loss analysis with AI to understand how AI coaching connects to systematic analysis of lost deals.

#AI coaching sales #sales coaching #sales training AI #conversation intelligence coaching

Want to learn to build with AI yourself?

A personal one-to-one programme on your own tasks, even if you have never coded before.

How we built it · Sales & follow-up

Sales cockpit with AI

A call list for salespeople. The call is transcribed in Danish, AI suggests the outcome (e.g. “Call again”) and drafts the email based on what the customer said.

  1. The salesperson calls
  2. AI writes transcript and email
  3. The salesperson sends
See the call and the follow-up →