Activity coaching fixes how much pipeline you build. It cannot fix a win-rate problem, because nothing about a lost deal shows up in a dial count. Win-rate problems live in what was said on the call, so the coaching has to happen there.
Activity metrics still answer a real question, and the last section says which one. A team that coaches only on activity, though, has no way to see why it loses the deals it already has.
Two conversations with the same person
Take a seller at Corvane, a made-up software company. Here are two coaching conversations the manager could have with her this week.
Activity coaching: "You made 42 dials. The target is 60. Pick it up this week."
Skill coaching: "On the calls you did make, you never asked what happens if they wait a quarter. Here is the moment on Tuesday's call where you could have."
Both are reasonable. The first is about capacity and takes a minute. The second is about what the seller says and needs the manager to have heard, or at least read, the call. Only the second can change a win rate.
Where each problem shows up
| What you see | Where it shows up | What to coach |
|---|---|---|
| Not enough meetings booked | Dials, emails and meetings held | Activity: volume, list quality, time blocked |
| Plenty of meetings, deals stall or are lost | What was and was not said | Skill: the questions and next steps on the call itself |
| Strong early stages, weak late stages | Calls with the decision maker | Skill: who signs, the decision process, the next step |
The first row is a pipeline-volume problem, so the activity numbers are the right place to look. In the other two rows, every activity number can be green. A seller can hit her dial target, book her meetings and still lose, because the dials and the meetings happened and the questions did not.
Why teams default to activity
Activity is the one thing a CRM can count without anyone listening to a call. So the weekly review becomes a dashboard review, and the calls, where deals are actually won and lost, stay unheard.
Research on managers describes the same pattern. Frank Cespedes wrote in Harvard Business Review that most sales managers overestimate how much time they spend coaching, and that even when they make time, their conversations are typically limited to discussing results and pending deals (HBR, December 16, 2021, read November 3, 2026). CSO Insights' 2018 report says sales coaching is not a matter of managing activities, and lists limiting it to discussions of opportunities or a funnel review among the narrow readings of what coaching is (CSO Insights 2018 Sales Enablement Report, page 35, read November 3, 2026).
Gartner made a related point about measurement in May 2026. It wrote that relying on lagging indicators like win rates and deal size hides the real drivers of seller performance, and proposed a measure it calls average interaction value as the missing link between activity and outcomes (Gartner, Steve Rietberg, May 4, 2026, read November 3, 2026). Activity counts the interactions and win rate counts the result. What happened inside the interaction is the part in between.
One limit on all of this: I could not find a study that compares activity coaching with call coaching on win rate. The case rests on definitions, on what managers say they discuss, and on the logic of what each metric can see.
More activity on a broken call
Push a seller from 42 to 60 dials with the same opener and you have the same opener, 18 more times a week. If the opener is the problem, you burn more of the list at the same conversion.
The same logic applies to the metrics that sound like skill. A talk ratio measures who spoke, and it can look textbook on a call where budget never came up. A count of questions rewards volume, not the one question that mattered. If you use them, treat them as a prompt to go and listen, not as the finding.
What skill coaching sounds like
Skill coaching has three parts: the moment in the call, the behavior that was missing, and the words to try next time. Compare two pieces of feedback on the same call.
- "You need to qualify the deal better." There is nothing in it to change on the next call.
- "At 22:10 the buyer said legal reviews every contract, and the call ended without a date for legal. Next time, ask when legal can see the draft."
The second one needs a call to point at, which is why the coaching has to start from the recording and not from the CRM. It also gives the seller something to practice: one question, asked in one place. A scorecard makes it repeatable, because the step it checks ("agreed a dated next step") is the same for every call of that type, so the same feedback can be given to everyone who misses it.
What it looks like when the one-on-one changes
Ray Dorsey, a junior partner at Apollo Insurance, described the shift in a Triple Session customer case study: "We're still doing one to ones, but we're coaching 100% off of the scorecards." The meeting stays, and the agenda changes from a dashboard to the calls (Apollo case study, read November 3, 2026).
Keep both, for different jobs
Use each metric for the question it can answer:
- Activity answers: is there enough pipeline? Dials, emails and meetings held are the right numbers for that.
- Call scores answer: why are we losing the pipeline we have? A scorecard step, such as "asked what happens if the buyer waits a quarter", answers that for each call.
To start, split the one-on-one. Spend the first ten minutes on activity and capacity, then pick a moment from a recent call and coach the one behavior it shows. To score calls on steps you can hear, Sales call scoring: how to build a scorecard your team uses has the templates. For what the manager does in each conversation once the scores exist, see AI sales coaching: what it is and how to run it.