How do you coach a rep to improve conversion at each stage?

The rep who looks like a closing problem and isn't

Picture a mid-market AE named on your board every Monday as "needs help closing." She has eleven deals in Proposal, four of them older than sixty days, and her close rate for the quarter is 18% against a team median of 29%. Every instinct — hers, yours, and the VP's — says coach the close. Run negotiation training. Practice the discount conversation. Get her a pricing objection script.

Then you actually open the funnel and the story inverts. Her Proposal-to-Closed-Won rate is 26%, which is inside a rounding error of the team's 29% and well within the noise band for a rep with only fifteen proposals in the sample. Her Discovery-to-Demo rate is 31% against a team median of 52%. That is the leak. Everything downstream of it looks broken because she is dragging half-qualified deals into Demo and Proposal to keep her pipeline coverage respectable, and those deals die late instead of dying early. The "closing problem" is a discovery problem wearing a costume.

This pattern is so common it should be the default hypothesis. A rep who cannot advance early-stage conversations fills the gap with volume, and the junk they push forward pollutes every rate after it. Coaching the close in that situation is worse than useless — it burns the rep's coaching appetite on a skill that is already at median, and it confirms to her that she is bad at the thing she is actually fine at. Reps have a finite tolerance for being coached. Spend it on the wrong stage twice and they stop bringing you real problems.

The scenario has an adjacent version on the SDR side that looks different and behaves identically. An SDR is booking meetings at target but 40% of them no-show or get disqualified by the AE. The manager coaches "book more meetings" because the booking number is the one on the dashboard. The real leak is connect-to-qualified-meeting: the SDR is booking anyone who says yes, without confirming the buyer has the problem, the budget authority, or a reason to show up. Same diagnostic move, different stage labels. The method transfers cleanly across roles, across motions, and across the RevOps handoff points between them.

There is a third variant that is not a coaching problem at all, and it is the one you must rule out before you open your mouth. Sometimes the leakiest stage is leaking because the system feeds it garbage. Lead routing sends enterprise-shaped leads to an SMB rep. The stage definition in Salesforce says "Demo" but three reps interpret it as "demo scheduled" and two interpret it as "demo delivered," so the transition rate is measuring nothing. A territory has been carved so thin there are 40 accounts in it and 34 are already customers. In every one of those cases the number is real and the rep is not the cause. Role-play will not fix routing. That is a RevOps ticket, not a 1:1 agenda item, and confusing the two is the single fastest way to lose a good rep's respect.

So the framing question before any coaching conversation is not "how do I improve this rep" but "which of four things is producing this number": skill, will, knowledge, or system. Skill means the rep knows what to do and wants to but cannot execute the motion — weak discovery questions, no multi-threading instinct, freezes on price. Will means the motion exists in their repertoire and they avoid it, usually because it risks a no. Knowledge means they do not understand the product, the buyer, or the qualification bar well enough to advance the stage honestly. System means the leak is upstream of the rep entirely. Each one has a different fix, and three of the four fixes are not "more role-play."

How stage diagnosis actually works

The mechanics are unglamorous and that is the point — they should be repeatable by any manager on any Monday without a data analyst.

Step one: pull a clean sample. Export every opportunity the rep touched in the last 60–90 days, including ones still open. Ninety days is usually the right window for a mid-market motion with a 45–60 day cycle; enterprise reps with six-month cycles need two or three quarters, and SDRs can be read on 30 days because their volume is high. The sample size matters more than the window. Below roughly 20 opportunities per transition you are reading noise, and you should widen the window or pool the rep with a peer of similar tenure to establish a baseline.

Step two: compute each transition, not the aggregate. For each stage pair, divide the count of opportunities that reached the later stage by the count that entered the earlier one. Do this for every pair in the funnel, not just the ones you suspect. The whole value of the exercise is that it surprises you.

Step three: compare each rate to a meaningful benchmark. Team median is the best benchmark you have because it controls for your product, your pricing, your market, and your current quarter. Industry benchmarks are useful for sanity-checking the whole team but useless for coaching an individual. If your team is five people, use the median of the three tenured reps, not the mean — one outlier rep with a whale skews a mean badly.

Step four: rank the gaps by revenue impact, not by size. A 20-point gap at the top of the funnel and a 20-point gap at the bottom are not equally valuable. Multiply each gap by the volume flowing through that transition and by the average deal size that survives it. A rep with 200 leads a month and a 5-point lead-to-discovery gap is leaking more absolute revenue than a rep with 8 proposals a month and a 15-point proposal gap. Do the arithmetic; do not eyeball it.

Step five: route the worst gap to a cause before choosing an intervention. This is where most managers skip straight to a fix. Don't. Run the branch logic below.

Two branches on that tree end in a RevOps ticket rather than a coaching plan, and that ratio is about right in practice. Roughly a quarter to a third of "rep conversion problems" that land on a manager's desk are operational: broken routing, ambiguous stage exit criteria, a lead-scoring model that has drifted since the ICP changed, an enrichment gap that means reps are cold-calling wrong numbers. Fixing those is a systems change that lifts every rep at once, which is a far better use of a manager's week than an hour of role-play with one person.

Step six: run the conversation so the rep finds the leak themselves. Use GROW — Goal, Reality, Options, Will — with the numbers on screen. Open on their goal, not yours: "Your Discovery-to-Demo is 31%, team median is 52%. At median that's roughly four more demos a month. Worth chasing this quarter?" Then move to Reality by narrowing to specific events, never the whole pipeline: "Walk me through the last three discovery calls that didn't convert. What did the buyer say right before it ended, and what did you ask them to do next?" Then go quiet. The silence does more work than any question you could add.

The single highest-yield follow-up in the entire method is this one: "What exactly did you say when you asked for the next meeting? Say it to me the way you said it to them." Nine times in ten the rep either did not ask, or asked in a way that pre-apologized — "does it make sense to maybe connect again sometime?" That sentence, said aloud in a 1:1, is the diagnosis. You no longer need to infer anything.

Options should be two or three, not one, and the rep picks: book the next meeting live before hanging up; tie the next step to a specific pain they uncovered; send a recap with a calendar link inside an hour. Then Will locks it to a number and a date: "You'll book live on every discovery call this week. We check Discovery-to-Demo two Fridays from now — what number tells us it worked?"

End with one change. Exactly one. A rep leaving a 1:1 with four things to fix will do zero of them, and you will have no way to attribute any movement in the number to any intervention.

Real numbers, ranges, and benchmarks

Precision here is what separates coaching from encouragement.

Sample thresholds. Twenty-plus opportunities through a transition before you trust the rate. Ten to nineteen, treat as a hypothesis to confirm with call review. Under ten, do not coach off the number at all — go straight to listening to calls, because the qualitative signal is more reliable than a rate built on eight data points.

Gap thresholds that justify intervention. A rep within about 5 percentage points of team median on a transition is inside normal variance for most funnels; coaching there is noise-chasing. Five to fifteen points below is the sweet spot for coaching — big enough to be real, small enough that one skill change plausibly closes it. More than 20 points below median on a single transition is worth a second look for a data or process explanation before you assume it is behavioral, because gaps that large usually have a structural cause.

Cadence and volume. Weekly 1:1 of 45 minutes beats a three-hour quarterly deep dive by a wide margin, because deliberate practice needs frequency more than duration. Two recorded-call reviews per week on the coached stage specifically — not two random calls, two calls at the leaking transition. Most managers who claim to coach weekly are actually running deal reviews weekly; audit your own calendar for how many minutes go to skill versus deal.

Time-to-signal. Behavior change shows up in two weeks. Conversion change shows up in 30 days for short-cycle motions and 60–90 for anything with a long sales cycle, because the deals in flight when you started coaching are still working through the funnel under the old behavior. This lag is the most common reason managers abandon a correct diagnosis — they check at day 14, see the conversion rate flat, and switch stages. Check the *behavior* at day 14 and the *rate* at day 30 or later.

The 30/60/90 spine. Days 1–30: diagnose and attack the single worst transition; weekly 1:1 plus two targeted call reviews; re-measure at day 30. Days 31–60: if the stage has reached median, move to the second-worst; if not, stay on it. Switching early is the most common failure mode and it produces a rep who is slightly better at nothing. Days 61–90: have the rep coach a peer on the stage they fixed. Teaching the motion is what converts a conscious script into an automatic one.

Ramp benchmarks for newer reps. Track weeks-to-first-stage-at-median rather than weeks-to-first-deal, because the stage metric gives you a signal months earlier. A ramping rep who hits median on Discovery-to-Demo in week six is on a very different trajectory than one still at half median in week twelve, and you will know it long before their first close would have told you.

Leading indicators worth instrumenting. The stage-specific conversion rate itself, checked biweekly. Activity into the leaking stage — discovery calls booked, next meetings set live on the call. A behavior-change count, which you verify in recordings rather than take on faith: how many times did the rep actually run the new motion. And the lagging confirmation, win rate on deals that passed through the coached stage after the intervention date. If the leading indicators move and the conversion rate does not, your diagnosis was wrong. Go back to the tree; do not coach harder.

A two-week scorecard, per stage. For a lead-to-meeting leak, track booking rate and time from first contact to meeting held. For a demo-to-proposal leak, track the percentage of demos ending with an explicit commitment question that got a clear answer — if that is under roughly 80% after two weeks, the rep has not internalized the script and needs more reps, not a new script. For a proposal-to-close leak, track objections surfaced on the call versus objections arriving by email afterward; a healthy ratio skews heavily toward the call, because objections raised live are objections you can actually handle.

Share the scorecard with the rep every Friday. The mechanism that makes coaching stick is not the insight in the 1:1, it is the rep watching their own number move and connecting it to something they changed.

Trade-offs: what to coach, what to automate, what to leave alone

Not every conversion problem should be coached, and choosing the wrong instrument is expensive.

Coaching versus enablement. Coaching is one-to-one, diagnostic, and behavioral. Enablement is one-to-many, curricular, and knowledge-based. If four of your six reps have the same leak at the same transition, that is not six coaching plans — that is one enablement problem, and running it as individual coaching costs you six times the hours for a worse result. The threshold is roughly this: one or two reps with a gap, coach it; half the team or more, build the play, ship the content, and change what "good" looks like at that stage for everyone.

Coaching versus process change. Some conversion leaks close faster with a process artifact than with a skill. A stalled proposal stage often improves more from a required mutual action plan field in the CRM than from teaching every rep to build one from scratch. A discovery leak sometimes closes with a mandatory qualification checklist at stage exit. The trade-off is real: process artifacts lift the floor quickly and reliably but cap the ceiling, and over-proceduralized funnels produce reps who fill fields instead of thinking. Use process for the failure modes you never want to see again; use coaching for the skills you want compounding.

Coaching versus tooling. Conversation intelligence tools surface where deals stall and how much reps talk, automatically and at scale. That is a genuine acceleration of step one — they find the stage far faster than manual CRM archaeology. What they do not do is run a GROW conversation, hold a rep accountable to a commitment, or distinguish a will problem from a skill problem. Use the tool to find the leak; use the human to coach the behavior. Managers who invert this end up forwarding dashboards and calling it coaching.

Coaching versus a performance conversation. This is the trade-off managers avoid longest. If a rep has had clear, specific, documented coaching on one stage, has agreed to a behavior, and the recordings show they simply are not doing it — that is not a skill gap. That is a will problem, and the honest response is a performance conversation, not a fourth role-play. Coaching a will problem as though it were a skill problem is a slow way to fail both the rep and the team.

Coaching versus fixing the funnel upstream. The highest-leverage version of this whole exercise often is not a rep intervention at all. If you compute stage conversion for the whole team and one transition is below benchmark for everyone, you have found a RevOps problem: a stage definition nobody agrees on, a lead source that converts at a fifth of the others, a handoff where context evaporates between SDR and AE. Fixing it lifts every rep simultaneously and permanently.

There is also a sequencing trade-off worth naming. When a rep has two real gaps, the instinct is to fix the bigger one first. Often the better move is to fix the *earlier* one first, because upstream leaks distort every downstream rate. Clean up discovery and the proposal number frequently improves on its own — not because the rep got better at proposals, but because the proposals are now going to qualified buyers. Coach in funnel order when the gaps are close in size; coach by revenue impact when one dwarfs the other.

Drills that match the stage, and the pitfalls that kill the whole loop

The drill has to match the leaking transition or you are just doing generic sales training with a personalized label on it.

Lead-to-discovery. Sit side by side for a 15-minute live prospecting block. Script the first ten seconds and the ask. The opener that works names a specific pain and one measurable outcome, then stops: "Most ops directors spend hours a week reconciling data across systems. We cut that substantially. Worth fifteen minutes to see how?" Then silence. Run it until the rep can deliver it without filler — no "just," no "quick," no "I was wondering if maybe."

Discovery-to-demo. Role-play with you as a deliberately vague buyer. The rep has to surface a named pain, a quantified metric, and an economic buyer before they earn the demo. Score on a three-line scorecard: pain named, metric quantified, next step booked live. Three yeses or the demo is not earned.

Demo-to-proposal. Drill the access ask and the commitment question. Have the rep ask for an introduction to the economic buyer three different ways out loud until one sounds like a human said it. Then drill ending every demo with a question that forces the buyer to articulate the business case themselves: "Based on what you saw, what would you tell your CFO the return looks like?" If the buyer cannot answer that, the deal is not proposal-ready, and sending pricing anyway is how you manufacture a stalled proposal stage.

Proposal-to-close. Role-play building a mutual action plan and surfacing objections before they go underground: "Before I send this over, what's the one thing that gives you pause?" Then practice acknowledging, normalizing with a comparable customer situation, and describing what changed after implementation. The mechanism here is simple — an objection raised on a call is coachable, an objection raised in an email after a proposal is usually already a decision.

Record every role-play and make the rep watch it. Self-observation is more persuasive than any feedback you can deliver, and it removes you as the adversary in the exchange.

Now the pitfalls, because the drills are the easy part and the loop is where this dies.

Coaching the deal instead of the skill. Spending a 1:1 rescuing one opportunity feels productive and teaches nothing transferable. The deal closes, the pattern persists, and next quarter you rescue another one. Coach the pattern behind the deal, then let the rep run the deal.

Coaching every stage at once. Attention spread across a whole funnel produces zero measurable lift and, worse, makes attribution impossible. If you change four things and one number moves, you have learned nothing about which change did it.

Rescuing the rep on live calls. Jumping in to close it yourself protects the quarter and destroys the rep. Let them run it, take notes, debrief after. The exception is a genuinely at-risk strategic deal, and you should say out loud that you are taking it and why, so the rep does not read it as a verdict on them.

No follow-through. A brilliant 1:1 with no day-30 re-measure is a conversation, not coaching. Put the re-measure on the calendar during the 1:1, in front of the rep. If it is not scheduled it will not happen.

Coaching everyone identically. A will problem and a skill problem need opposite responses. Skill gets reps and reduction — break the motion into smaller pieces until it is easy. Will gets risk reduction and accountability — make the scary ask lower-stakes, then hold the commitment. A single pep talk fixes neither.

Ignoring the system. If stage definitions in the CRM are ambiguous or routing is broken, no amount of role-play moves the number. This is the pitfall that makes managers look like they are working hard while nothing improves, and it is the one a RevOps partner will spot in an afternoon.

Switching stages too early. The rep's behavior changed at day 14, the rate had not moved, and you moved on. The rate was always going to lag. Hold the line for at least one full sales cycle before you conclude the intervention failed.

Coaching from memory. Walking into a 1:1 with an impression instead of a rate guarantees a conversation about feelings. Open the numbers on screen. The data does the confrontation so you do not have to.

Related questions

Does this method work for SDRs as well as AEs?

Yes, with different stage labels. For SDRs the transitions are lead-to-connect, connect-to-meeting-booked, and meeting-booked-to-meeting-held-and-accepted. The diagnostic sequence is identical: compute each rate, compare to team median, rank by volume impact, coach the single worst one.

What if conversion looks healthy but the rep still misses quota?

Then the constraint is volume or deal size, not conversion. Check top-of-funnel activity, pipeline coverage ratio, and average contract value before touching conversion coaching. Coaching the wrong lever wastes the quarter and costs credibility with the rep.

How do I benchmark when the team is only three or four reps?

Use the rep's own trailing history as the baseline instead of a thin team median, and supplement with call review. With small teams, qualitative evidence from recordings is more reliable than a median built from two comparable people.

Should stage exit criteria be enforced in the CRM?

Usually yes for the transitions where deals stall most. Required fields raise the floor and make conversion rates comparable across reps. The cost is rigidity, so enforce criteria at two or three key gates rather than at every stage.

Who owns fixing a leak that turns out to be systemic?

RevOps, not the frontline manager. Routing rules, lead scoring, stage definitions, and handoff data quality are operational assets. The manager's job is to detect the pattern across reps and escalate it with evidence.

FAQ

How do I know which stage to coach first?

Compute each stage-to-stage conversion for the rep's last 60–90 days and compare every transition to team median. Then rank the gaps by revenue impact — gap size multiplied by the volume flowing through that transition — rather than by raw percentage points. The top-ranked gap is where you start, and you stay there until it reaches median or you prove the diagnosis wrong.

How long before the conversion number should move?

Expect behavior change in about two weeks and rate change in 30 days for short-cycle motions, 60–90 days for longer ones. Deals already in flight when coaching began still carry the old behavior through the funnel, so early rate checks read flat even when the intervention is working. Check behavior at two weeks, the rate at thirty days or later.

What if the rep does the drill in role-play but not on live calls?

That is a will signal, not a skill signal, and it needs a different response. Find out what the rep believes will happen if they run the motion — usually they are protecting a relationship or avoiding a no. Lower the perceived risk of the ask, then hold the specific commitment with a date. If it repeats after clear, documented coaching, it becomes a performance conversation.

Can AI call-coaching tools replace this process?

They replace the slow part, not the human part. Conversation intelligence surfaces where deals stall, how much the rep talked, and which topics never came up — far faster than reading CRM exports. What no tool does is distinguish will from skill, negotiate a commitment, or notice the rep's tone change when a particular objection comes up. The tool finds the stage; you coach the human.

How much of a manager's week should this take?

For a team of six to eight reps, budget roughly 45 minutes of 1:1 per rep per week plus one to two hours of call review across the team, so about eight to ten hours weekly. That sounds heavy until you compare it to the hours the same manager spends in deal reviews that produce no transferable skill. Rebalancing existing time usually costs less than adding new time.

When is coaching the wrong answer entirely?

When the leak is structural — junk lead routing, an ambiguous stage definition, a territory with no addressable accounts, or a product gap the rep keeps losing to. Also when the rep has received clear, specific coaching with documented commitments and no behavior change; that is a performance discussion. Coaching cannot fix a wrong-fit hire, a broken funnel, or a dead patch of market.

Sources

• Harvard Business Review — sales management and coaching research
• Gong Labs — sales conversation and win-rate research
• Salesforce — sales coaching resources
• RAIN Group — sales coaching research and articles
• Winning by Design — SaaS sales method resources
• Sandler — sales coaching and management resources
• MEDDIC Academy — qualification methodology
• HubSpot — sales coaching guides

Related on PULSE

• [What single question can help a rep prioritize which leads to pursue first each week?](/knowledge/cg0850)
• [How much time should a sales manager spend coaching each week?](/knowledge/cg0002)
• [How do you coach reps to use stage exit criteria correctly?](/knowledge/cg0109)
• [What question would you ask a rep who consistently loses deals at the proposal stage to diagnose the real issue?](/knowledge/cg0860)
• [How can I ask a question that helps a rep identify their own pattern of losing deals in the same stage?](/knowledge/cg0841)
• [Top 10 Questions to Diagnose Why a Deal Stalled at the Negotiation Stage](/knowledge/cg0915)