Short Answer
Across 220 scored practice calls from 53 sales reps, the weakest skill was qualification: it was the lowest-scoring category on 45% of calls and averaged 40% of its possible points. Rapport was the strongest at 92%. In other words, reps are likable, listen reasonably well, and handle objections better than you'd expect, but they leave calls without knowing who decides, what the budget is, or when a decision happens. The most common piece of coaching on the scorecards, on 57% of calls, was some version of "ask the question."
Where the data comes from
Every practice call on RolePractice ends with a scorecard: ten skill categories, each with a point value, scored by AI against the transcript. For this piece I pulled every scored call from real accounts between March and October 2026, dropped anything under 60 seconds (connection drops, not conversations) and kept the 220 calls with a full category breakdown. That's 53 reps across a mix of SaaS, security integrators, medical, insurance and professional services, mostly on "standard" difficulty, practicing discovery and objection scenarios with AI buyers.
Two honest caveats. First, 220 calls is a real sample but not a large one; treat the ordering as reliable and the exact percentages as approximate. Second, these are practice calls against AI buyers, not live customer calls, so the stakes are lower. If anything, that flatters the reps: the eleven real customer calls in the set that were scored through our note-taker averaged 32 out of 100 against 51 for practice calls, though eleven is too few to lean on.
The ten skills, ranked
Average score as a share of the points available, lowest first:
| Skill | Average | |---|---| | Qualification | 40% | | Value articulation and positioning | 43% | | Closing and next steps | 46% | | Opening and first impression | 49% | | Discovery and needs analysis | 52% | | Active listening and adaptability | 55% | | Communication and professionalism | 65% | | Objection handling | 72% | | Ethics and integrity | 86% | | Rapport and trust building | 92% |
The shape is the story. The relationship skills at the bottom of the table are nearly maxed out. The deal skills at the top are below half. Reps build trust and then fail to use it.
Qualification is the problem on nearly half of all calls
For each call I also looked at which single category scored lowest. Qualification was the weakest skill on 100 of 220 calls, 45%. Nothing else came close: closing was weakest on 12%, value articulation on 11%, the opening on 8%.
What does under-qualifying sound like? Reading the transcripts, it's rarely a rep who refuses to ask. It's a rep who asks one surface question, gets a vague answer, and accepts it. "Who else would be involved in this?" "Oh, a few people." "Great, and what's your timeline?" "We're looking at next year sometime." Both answers are non-answers, and the call moves on to the demo.
The scorecards' coaching reflects it. "Question" appears in the next-call focus on 125 of the 220 calls. "Decision" (as in who makes it, or when) on 25. "Budget" on 12. "Pain" on 13. Almost nobody is told to pitch better. They're told to find out more before they pitch.
Objection handling is better than reps think
The surprise in the data is objection handling at 72%. Sales leaders tend to assume objections are where reps crumble, and most sales training is built around that assumption. In these calls, when the AI buyer pushed back on price, timing or the incumbent vendor, reps mostly held their ground, acknowledged the concern, and kept the conversation going.
The gap shows up one step earlier. Reps handle the objection they get, but because they didn't qualify, they're often handling the wrong one. A rep who never found out that the buyer's contract renews in eight months will spend three minutes answering "we're happy with our current vendor" when the real objection is "call me in six months."
The fourth call
One more pattern, from scores by call number per rep. Averages for a rep's first three practice calls were 47, 46 and 43. The fourth call jumped to 59, and the fifth and sixth held at 58. The dip on call three is partly selection (reps often choose a harder scenario once they've seen the easy one), but the jump after it is consistent.
Three calls appears to be the amount of repetition it takes before the coaching from the scorecard changes what a rep does on the call. One practice session doesn't move anything. It's why we built the Skill Trend to unlock on the third call: before that there's no trend to show.
What to practice, in order
If you run a team, the data suggests a simple order for the next month:
- Qualification, every call, before anything else. Drill the four questions that were missing most often: who else is involved, what happens if nothing changes, what the budget conversation looks like, and what date they're working back from. Practice them until they come out as conversation rather than a checklist.
- Connect value to what you heard. Value articulation was the second-weakest skill, and it depends on the first: you can't position against a problem you didn't uncover.
- End the call with a next step that has a date. Closing was weakest on 12% of calls, and almost always because the call ended with "I'll send some information."
- Leave objection handling alone for now. It's the skill most teams practice most, and it's already the strongest of the deal skills. The time is better spent upstream.
And run each scenario at least four times per rep. Based on this data, the first three are the price of admission.
Frequently asked questions
How were the calls scored? Each call is scored against the transcript across ten categories with fixed point values (qualification is worth 10, discovery 20, and so on), by the same AI scorer every time. The categories and weights are visible on every scorecard. Difficulty is accounted for: harder buyers get a small upward adjustment.
Are practice-call scores comparable to real calls? Not directly. The eleven real customer calls in this set, scored through the note-taker on the same rubric, averaged 32 against 51 for practice, which suggests reps do worse under real pressure, but the sample is far too small to say by how much.
Will you update this as the data grows? Yes. This is the first cut, at 220 calls. The plan is to re-run it quarterly and publish what changes.
If you want to see where your own team sits on these ten skills, start a free trial: five practice calls, a scorecard after each one, no credit card.