Data-backed sales coaching means a manager chooses what to coach from evidence in the rep's deals and calls, and checks the result in the same evidence afterwards.
Two kinds of evidence
Two kinds of evidence do most of the work. The first is the deal: what the buyer has said, who has been on the calls, and what the team's method says should be there by this stage. The useful question is which forecast deals lack a dated next step or a second contact. "What's your gut feel on this deal?" gets the rep's mood. The second is the call: how the rep handled discovery, objections or the close, scored against the team's own method. A score on a generic rubric coaches a method the team does not sell with.
- 01Read the deal evidenceWhat the buyer said, who joined the calls
- 02Read the call scoresAgainst the team's own method
- 03Check the data firstDoes it order calls like your best manager?
- 04Coach on the live dealPick the deal and the moment
- 05Check the resultIn the same evidence afterwards
The trap is trusting a score that would change if you ran it again. In our research on AI call scoring, more than 250,000 scoring decisions across 15 models, one recent model changed 15% to 17% of its scores between identical runs. Trained human experts disagreed with each other on 18% of scores. A rep coached on a 3 that should have been a 4 is coached on noise.
So check the data before coaching from it. Ask whether it puts calls in the same order your own managers would. Average error is a poor test, because a scorer that never read the transcript can do well on it. Then use the data to pick the deal and the moment, and coach the rep on the live deal, where transfer research says the work environment decides whether skills get used (Saks and Belcourt 2006). This is the debrief step of a sales coaching framework with better inputs.
A session picked by the data (illustrative)
A manager looks across a rep's pipeline before their weekly session. Three deals are forecast to close this month. Two have no contact above director level on any call. The rep's last four discovery calls score low on quantifying the problem, and that pattern holds when the calls are scored again.
The session takes one of the two deals. The manager and rep agree the next call has to produce a cost figure and a name in finance. A week later both show up in the deal record, and the rep's next discovery call scores higher on the same skill. The data picked the deal and showed whether the coaching worked.
Turning the data into a coaching conversation
Data-driven sales coaching still ends in a conversation, and the data decides where it starts. A useful order:
- Open on the deal the data flagged, and say why it was flagged: a missing next step, one contact on a late-stage deal, a low score on a skill.
- Ask the rep what they see before showing the score. Reps who find the gap themselves tend to act on it.
- Dig into one moment on the call, in the buyer's words.
- Agree one change for the next buyer meeting, and check the record afterward.
In a method-specific team the flag reads in the method's own terms. A Gap Selling team would see "no current state in the buyer's numbers"; a MEDDPICC team would see "Economic Buyer still rep-inferred". That gives the manager a question to ask on the call, which a generic activity score cannot.
How much data, over what period
A coaching decision needs enough data to show a pattern and recent enough to still be true. For one rep, that usually means the last four or five calls of the same type and the deals currently in their pipeline. A single call is an anecdote. A quarter of calls mixes in habits the rep may already have changed.
Keep the data where the manager already works. If deal evidence sits in the CRM and call scores sit in a separate tool nobody opens before the one-on-one, the session falls back on memory. Put the flagged deal and the call score in the same place as the meeting notes.
Where data-driven coaching goes wrong
Coaching the dashboard. The manager reads out scores, the rep nods, nothing changes. Data picks the moment; the conversation about the live deal does the coaching.
Too many flags. A rep shown ten red flags fixes none of them. Pick one deal and one skill.
Counting activity as evidence. Calls made and emails sent show effort. They say little about whether the buyer has given the deal what it needs.
Putting the evidence to work
The sales coaching framework covers the full session. How to measure sales coaching effectiveness sets out the measures in detail, and what is sales coaching covers the basics.
Our AI sales experts start from the deal evidence. Each deal is checked against what the buyer has said, Bryce shows which deals carry buyer evidence, and the rep takes the gap to Caty on a call (or to Keenan or Val on a Gap Selling or ValueSelling Framework® team) to prepare the next conversation. Bryce gives managers the read across the team (for leaders). If you want to see the deal side on one of yours, talk to us.
FAQ
What data should sales managers use for coaching? Start with deal evidence: what the buyer has said, who has joined calls, and what the team's method expects by this stage. Add call scores against that method. Activity counts help spot effort problems. They do little to show whether a deal has what it needs to close.
Can AI call scores be trusted for coaching? Some AI call scores can be trusted for coaching, once the scorer is tested. In our call scoring research, one recent model changed 15% to 17% of its scores between identical runs, while the final scoring method changed 1.7%, and the full setup matched the experts' ranking at 0.89. Test any scorer by re-running it and checking it ranks calls the way your own managers would.
Is data-driven coaching the same as conversation intelligence? Data-driven coaching uses conversation intelligence as one source of data. Conversation intelligence records and analyzes calls. Data-backed coaching is what a manager or coach does with that data: choose the deal and the moment, agree a change with the rep, and check the next call for it.