How do you coach reps to ask better questions using call data?

Two ways to coach question quality

There are really only two mechanisms for turning "ask better questions" from a vague directive into a trainable skill, and they pull in different directions. The first is manual transcript review: a manager pulls one real recording, sits with the rep, and hand-counts questions into open and closed buckets while they both listen. The second is automated conversation-intelligence scoring: a platform like Gong or Chorus ingests every call, auto-classifies each question, and surfaces trend lines — question count per call, open-to-closed ratio, talk-to-listen ratio — across a rep's entire call history without a human doing the counting.

Manual review wins on depth and trust. When a manager and rep sit down with one transcript, they can dig into a single exchange — "at 7:40 you asked a closed question right after the buyer said renewals were a headache, what could you have asked instead?" — and that specificity is what actually changes behavior. The rep hears their own voice, sees the missed opening, and generates the better question themselves instead of being handed a rule. The cost is scale: a manager with eight reps cannot manually review more than one or two calls per rep per week without the review queue backing up, and manual counting is slow enough that it is easy to skip when the week gets busy.

Automated scoring wins on coverage and consistency. A platform tags every question on every call, so a manager sees that a rep's open-question ratio dropped from 70% to 40% over the last two weeks, or that question count per call fell from 12 to 6 after a new sales sequence launched — patterns a manual spot-check would never catch because it only samples one call in twenty. The tool never gets tired, never misses a call, and produces the same categorization logic across the whole team, so comparisons between reps are apples-to-apples. The cost is nuance: automated classifiers are decent at flagging "is this a question" and "does it end in a question mark or interrogative phrasing" but weaker at judging whether a question was actually *good* — a technically open question can still be shallow ("tell me about your company") while a technically closed one can be sharp ("so renewals slipping is costing you the Q3 forecast number, is that right?").

The practical answer for most RevOps and sales-enablement teams is not to pick one — it's to sequence them. Use the automated scorecard as the triage layer: it tells you which reps and which weeks need attention, at a scale no manager could sustain by hand. Use manual review as the coaching layer: once the data flags a rep whose open-question ratio is sliding, that is the call you sit down and manually dissect together. Trying to run pure manual review across a team of any size collapses under its own time cost; trying to run pure automated scoring without ever sitting with a rep on a real transcript produces dashboards nobody acts on.

How to decide between them

The decision usually comes down to three factors: team size, tooling budget, and how much coaching bandwidth managers actually have in a given week. A five-person team with a hands-on manager can lean more heavily on manual review because the reachable call volume is small enough to review meaningfully every week. A twenty-five-person team spread across two managers cannot — the math does not work without an automated layer doing the first pass.

Budget is the second constraint. Full conversation-intelligence platforms carry real per-seat costs, and if the org does not already have Gong, Chorus, or a similar tool deployed, that is a build-or-buy decision above the individual manager's pay grade. Where the tooling does not exist yet, the fallback is not "skip data-driven coaching" — it is manual review of raw call recordings or transcripts, which still beats coaching from memory even without automated tagging.

The third factor is coaching bandwidth: how many hours per week can a manager actually spend in 1:1 coaching sessions versus pipeline reviews, forecasting, and their own selling. If bandwidth is thin, automated scoring should carry more of the weekly workload — flag the two or three reps whose numbers moved the most, and spend the scarce manual-review time only on those.

Concrete numbers behind each option

The numbers matter because "ask better questions" without a target is not coachable — reps need a line to clear. Gong Labs' research on discovery calls is the most cited benchmark here: strong discovery calls cluster around 11 to 14 questions, weighted heavily toward open, problem-centered phrasing rather than closed yes/no confirmations. That is not a hard ceiling — a call that runs long can carry more — but it is a useful floor: a discovery call landing at 4 or 5 total questions is very likely under-probing, regardless of channel or deal size.

For the open-to-closed ratio, a workable target is roughly 70% open to 30% closed, mirroring the pattern Gong's benchmark data shows among top performers. Closed questions still earn a place — confirming a fact, locking a commitment, gaining explicit agreement to a next step — but a call where closed questions dominate usually means the rep never dug past surface-level pain.

If a manager is building a scorecard, a simple point system works well in practice: 1 point per open-ended question, 0.5 points per closed question, 0 points for leading questions ("don't you think our solution would help?" counts as leading, not open, because it pre-supplies the answer). A strong 30-to-45-minute discovery call should score somewhere in the 10-to-15-point range. That single number gives a rep something to track week over week without requiring a manager to relitigate every individual question.

Talk-to-listen ratio is the adjacent metric worth tracking alongside question count, because the two move together: Gong's research on talk ratio consistently links calls where the buyer does the majority of the talking to higher win rates, and a rep who is asking strong open questions naturally talks less because they are listening to the answer instead of pitching past it.

On timelines, expect the leading behavior — the actual question count and ratio on reviewed calls — to shift within two to four weeks of consistent coaching, because it is one of the more mechanically coachable sales skills; it does not require new market knowledge, just repeated, specific practice. The lagging metrics — discovery-to-demo conversion and win rate — take longer, typically a full quarter, to show a clean signal, because deal cycles and other variables dilute the effect of any single skill improvement. A 30/60/90 framing helps set expectations: by day 30, question count and open ratio should be trending up on every reviewed call; by day 60, the rep should be self-scoring their own calls before the 1:1 even happens; by day 90, the behavior should hold without prompting and start showing up in the conversion numbers.

Implementation details and sequencing

Rolling this out well is less about the diagnostic tools and more about the weekly cadence a manager actually runs. Start with a baseline week: pull two of the rep's most recent discovery calls, whether through an automated platform or a manual transcript pull, and build the initial scorecard — question count, open-versus-closed split, whether a cost-of-inaction question was asked, whether the call ended with a confirmed next step. This baseline is the number every future week gets compared against, so it needs to be a real, representative call rather than the rep's best performance.

In week two, shift to modeling and drilling. Watch a top performer's call together so the rep has a concrete bar to aim for rather than an abstract standard, then role-play the rep's next one or two live deals before those calls actually happen. This is where the manual-review depth pays off — a role-play lets the manager stop mid-conversation and ask "what could you have asked there?" in a way a dashboard never can.

Weeks three and four move into a steady review-and-adjust loop: one real call reviewed against the scorecard each week, with the manager deliberately calling out the open questions the rep already nailed before naming the single thing to improve next. Piling on multiple critiques per session dilutes the message; picking one specific, measurable target per week keeps the feedback loop from overwhelming the rep.

The actual coaching conversation inside that loop should follow the GROW model — Goal, Reality, Options, Will — because it keeps the manager from just handing the rep better questions to memorize. In the Goal step, tie the session to the rep's own number (their discovery-to-demo conversion rate, for instance) rather than a company-wide mandate. In Reality, open the transcript and let the data — not the manager's memory of the call — be the mirror: "you asked four questions here, three were closed." In Options, make the rep generate three to five better questions themselves rather than supplying a list; if they stall, model exactly one example and hand the pen back. In Will, close with a specific, measurable commitment tied to a date and a recording — "on your next three discovery calls, ask at least ten questions, at least seven open, measured off the next three call recordings, reviewed together Friday."

Layer in short recurring drills between formal review sessions to keep the skill live: a transcript audit where the rep highlights their own questions in two colors and counts them before the manager says anything; a question-only role-play where the manager plays the buyer and the rep is barred from pitching for three full minutes; a "why" ladder drill that takes one piece of stated pain and pushes five layers of follow-up questions deeper; a silence drill where the manager stays quiet for five seconds after the rep asks a strong question, training the rep not to rescue the silence with a weak closed follow-up; and a "steal from the best" exercise where the rep adapts a 90-second clip of a top performer's sharpest discovery question to their own active deal.

For a RevOps or enablement function standing this whole system up, the sequencing above is the operating model to formalize — baseline, model, drill, review, verify — not a one-time training session. The biggest implementation failure is coaching from memory instead of the recording ("you should ask more questions" is an opinion the rep can dispute; "you asked four, three closed" is data they cannot), and the second-biggest is a manager solving one stuck deal instead of coaching the reusable questioning pattern that shows up across every deal that rep will ever run. Build the cadence once, using the real call data as the shared source of truth, and it keeps producing better reps without needing to be reinvented every quarter.

Related questions

How many questions should a rep ask on a discovery call?

Gong Labs' benchmark data points to strong discovery calls landing around 11 to 14 total questions, weighted toward open, problem-centered phrasing. Treat it as a guardrail, not a hard quota — diversity and depth of questions matter more than the raw count alone.

Should a manager share the call data with the rep before giving feedback?

Yes — share the transcript and let the rep count their own open-versus-closed questions first. Self-diagnosis from real data turns the gap into the rep's own conclusion rather than the manager's criticism, which makes the fix stick faster.

What's the fastest way to break a rep's habit of asking only safe, closed questions?

Pull three calls where a closed question got a one-word answer, then drill rewriting each one as an open follow-up in real time — for example, turning "is this a priority?" into "what makes this a priority now?" — and track the reduction in closed-question clusters over the next ten calls.

Is more questions always better?

No. A rep who asks strong, diverse questions but still loses deals may be facing a territory-fit, pricing, or process problem rather than a questioning skill gap — more coaching on questions will not fix a broken deal structure or the wrong target account.

FAQ

What conversation-intelligence tools support this kind of question coaching? Gong and Chorus are the two most widely used platforms for auto-tagging questions, tracking talk-to-listen ratio, and surfacing trend lines across a rep's call history. Teams without either can still coach effectively using manually pulled transcripts or call recordings — the scoring is just done by hand instead of automatically.

How do I stop a rep from gaming the question count instead of actually improving? Score for quality, not volume — weight open, problem-centered questions higher than closed ones in the scorecard, and flag any pattern of throwaway closed questions stacked just to inflate the count. The goal stated to the rep should always be sharper and more diverse questions, never a raw number to hit.

What if the rep gets defensive about being recorded and reviewed? Reframe the recording as the rep's own film room rather than surveillance, and start by watching a top performer's call together so the rep sees the bar before their own gets reviewed. Keep each session focused on one real call and one improvement, never a highlight reel of every mistake.

Do closed questions have any place in discovery coaching? Yes — closed questions are useful for confirming facts and locking explicit commitments, such as "so the plan is you'll loop in your CFO by Friday, is that right?" The problem is a call dominated by closed questions, which usually signals shallow discovery and a weak business case underneath it.

How long before question coaching shows results? Behavior on reviewed calls typically shifts within two to four weeks because it is a highly mechanical, coachable skill. Conversion and win-rate are lagging indicators tied to full deal cycles, so give those a full quarter before judging whether the coaching worked.

Where does this fit inside a broader RevOps enablement motion? Question coaching should be one tracked component of a RevOps team's broader call-quality program, sitting alongside talk-to-listen ratio, next-step confirmation rate, and discovery-to-demo conversion — all pulled from the same call data so the whole enablement function is coaching off one consistent source of truth instead of separate, conflicting metrics.

Sources

• Gong Labs — Discovery call research and question data
• Gong Labs — Talk-to-listen ratio and win rates
• Harvard Business Review — The Surprising Power of Questions
• RAIN Group — Sales Discovery Questions
• Sales Hacker — How to Run a Great Discovery Call
• Performance Consultants — The GROW Model of Coaching
• Winning by Design — Discovery and Sales Frameworks

Related on PULSE

• [Top 10 questions to audit a rep's CRM data accuracy](/knowledge/cg0891)
• [How do you use conversation-intelligence data to coach a team?](/knowledge/cg0124)
• [How do you coach reps to ask better discovery questions?](/knowledge/cg0052)
• [How can I ask a question that helps a rep structure a better follow-up sequence?](/knowledge/cg0854)
• [How do you coach reps to tell better customer stories?](/knowledge/cg0066)
• [How do you coach a rep to write better follow-up emails?](/knowledge/cg0045)