Record every discovery call
You cannot review what you did not capture. Record every discovery call and tell the prospect in the first minute: "I'll record this so I can take proper notes. Is that fine?" Check the consent rules for the countries you sell into, because they differ.
A call recorder such as Fathom, Fireflies or Gong gives you the recording and a transcript. Store them where the whole team can open them, and link each recording to the deal in your CRM, such as HubSpot or Pipedrive.
Pick the sample by outcome
Do not listen to everything and do not let reps pick their own best calls. Each month, pull 8 to 10 calls from the CRM in three groups:
- Calls that became a proposal.
- Calls that were dropped after discovery.
- Calls that stalled with no next step.
Make sure every rep appears at least once. Choose randomly inside each group. Listening only to wins teaches you nothing about where deals leak.
Score against a short scorecard
Build the scorecard from the framework you already use, see Structure your discovery framework and Design the call flow. Six questions is enough. Score each 0, 1 or 2:
- Did the rep set an agenda in the first minute.
- Did the prospect name the problem in their own words.
- Was the cost of the problem put into numbers.
- Were the decision makers and the timeline established.
- Was a budget range discussed.
- Did the call end with a next step booked in the calendar.
Add the talk ratio as a seventh number. The prospect should speak for most of the call, see Common failures. A score out of 12 is crude, and that is fine. You are looking for patterns across calls, not a grade for a person.
Review clips, not whole calls
Whole calls take too long and the habit dies by month three. Prepare each call in ten minutes: note the timestamp of the opening, the moment the problem was discussed and the close. Play those three clips of two to three minutes each.
Run one session a month of 60 to 90 minutes. The rep whose call it is speaks first and says what they would change. Then the group names one thing that worked and one thing to change. Stop there.
Agree one coaching action per rep
Every rep leaves with one action, written down, small enough to check. For example: "On every call this month, ask what happens if nothing changes." Not five actions. People change one behaviour at a time.
Put the action in a shared document or in the CRM. At the next review, listen for it in the new sample. If it is there, pick a new one. If not, keep it for another month.
Send patterns to the right place
Single calls are anecdotes. Patterns are material for other chapters. Keep a simple tally sheet in Google Sheets with one row per call. When the same objection appears three times, move it to Handle objections in discovery and the practice sessions in Train on common objection patterns. When a question predicts which calls become deals, feed it into Refine scoring criteria.
Common mistakes
- Reviewing only the calls that were won.
- Listening to full calls, so the review takes a day and gets cancelled.
- Scoring the person instead of the behaviour.
- Giving five pieces of feedback at once.
- Skipping a month because it was busy. The cadence is the point.
- Letting the manager be the only voice in the room.
How you know it works
After three monthly reviews, check four numbers. The average scorecard score has gone up. The prospect's share of talk time is closer to 70 per cent. More calls end with a booked next step. And the same objection now gets a better answer than it did in month one.
If the scorecard rises but the share of calls that become proposals does not, your scorecard measures the wrong things. Change it. On Monday morning, pick eight calls from last month and book the first session.