How do you create a coaching culture on a sales team?

The eight-rep team where coaching only happens after a bad month

Picture a mid-market SaaS team: eight AEs, one manager, average deal size $42K, sales cycle running about 90 days. The manager, promoted eighteen months ago because they were the top rep, holds weekly 1:1s. On paper this looks like coaching. In practice, every 1:1 is a pipeline scrub — "where's Acme, what's the next step, will it close this month?" — and the only time skill comes up is when a rep misses two quarters in a row and HR gets involved.

Watch what that teaches. Reps learn that being coached is a signal you are in trouble. So they stop volunteering the calls that went badly. The rep who fumbled a pricing objection on Tuesday does not raise it Thursday, because raising it looks like admitting a gap in front of someone who decides comp and promotions. Instead they quietly move the deal to next month and hope discovery goes better on the next one. Nobody learns anything, the same mistake recurs across six other reps who never heard about it, and the manager's honest belief is that coaching happens on this team.

Now look at the second-order damage. Because no losing call ever gets dissected in public, there is no shared definition of good. Ask all eight reps "what does a great discovery call look like here?" and you will get five different answers, three of them contradictory. New hires learn by shadowing whoever sits near them, which means ramp quality is a lottery. The manager rescues deals — jumping on the call at the eleventh hour to save Acme — which feels heroic, closes the quarter, and builds exactly zero capability in the rep. Next quarter the same rep needs the same rescue.

That team does not have a coaching problem. It has a *culture* problem, and the distinction matters because the fixes are different. A coaching problem is one manager who does not know how to run a good conversation; you fix that with training. A culture problem is a system where the incentives, the calendar, and the social norms all quietly punish surfacing weakness; you fix that by changing what gets scheduled, what gets modeled from the top, and what gets rewarded.

The tell that you are in the second situation: coaching stops the instant the quarter gets tight. If a forecast call can cancel a coaching block, coaching is not a norm — it is a nice-to-have that survives only in calm weather, and reps read that signal perfectly. In sales, everyone watches what leadership actually protects, not what leadership says.

There is a RevOps dimension people miss. Whether coaching happens is largely determined by systems nobody thinks of as coaching systems: is call recording turned on, are stages defined tightly enough that "stalled at discovery" is a legible pattern, does the CRM capture enough that you can spot the same failure across reps? A manager cannot coach what they cannot see, and visibility is an operations problem before it is a leadership one.

How the coaching loop actually works

Strip away the language and a coaching culture is one loop running often enough that nobody experiences it as a program. Observe a real interaction, diagnose the specific skill gap, coach with questions rather than corrections, practice the new behavior in a low-stakes rep, measure whether the behavior actually changed on live calls, reinforce publicly, and go around again. Every part of that loop fails in a characteristic way, and knowing which part is broken tells you what to fix.

Observe fails when there is nothing objective to look at. If coaching runs on the rep's recollection of a call, you are coaching a story, not a call — and the rep's memory is systematically biased toward the parts that went fine. Recording tools like Gong or Chorus solve this not because the AI is magic but because they end the "that's not how it happened" argument. If you do not have a tool, sit in on live calls and take verbatim notes; the point is a shared artifact, not a specific vendor.

Diagnose fails when managers treat every problem as the same problem. Before coaching, root-cause into one of four buckets. A skill gap means the rep genuinely does not know how to run layered discovery — training and reps fix it. A will gap means they know how and are not doing it — GROW will not fix that, and pretending otherwise wastes months. A knowledge gap means there is no shared standard, so the rep is improvising against an invisible target — a published scorecard fixes it. A system gap means the org has not blocked time, turned on recordings, or set the expectation from above — no amount of manager effort overcomes that.

Coach fails when the manager tells instead of asks. Telling is faster, feels productive, and builds dependence. The GROW model — Goal, Reality, Options, Will — exists to force the rep to do the thinking. In practice: *"What's the one thing you most want to get better at this quarter?"* (Goal — let them name it, buy-in is higher than an assigned focus). *"Let's pull up the Acme call — where do you think it went sideways, and where were you happy with it?"* (Reality — rep self-assesses first, manager stays quiet, and this silence is the hardest discipline in the whole method). *"You said discovery felt rushed. What would you do differently running it tomorrow?"* (Options — they generate the fix). *"Which of those will you try on your next two discovery calls, and how will I know you did?"* (Will — commitment plus a measurable next rep).

When you must correct something genuinely dangerous, wrap it in a rep instead of a lecture: *"Can I show you a different way to handle that pricing pushback? Try it on me right now and tell me how it felt."* That phrase — *try it on me right now* — is the hinge between feedback and practice.

Practice fails when it never happens. Feedback without a rep is trivia. The behavior has to be attempted in a setting where being bad at it costs nothing, which is what role-play is for and why teams that skip role-play see feedback evaporate within a week.

Measure fails when the only measure is quota. Quota moves too slowly and is confounded by territory, pipeline, and luck. You need to verify in the recording that the rep actually did the thing they committed to — not ask them whether they did.

Reinforce is where culture is made or lost. If improvement is only ever discussed privately in a 1:1, coaching stays a manager-rep transaction. If a rep's improved discovery clip gets shared in a team channel and three peers steal the phrasing, coaching becomes something the team does to itself.

The loop is the whole product. A culture is not a value statement on a wall; it is this cycle running weekly for long enough that reps start initiating it without being asked.

The cadence, the ramp, and the numbers that defend it

Coaching is a calendar problem before it is a content problem, so start with the schedule and the ninety-day install.

Weekly rhythm. One 30-minute structured 1:1 per rep built on GROW and one recorded call — not a pipeline scrub, which is a separate meeting with a separate purpose. One 45-minute team call-review where a single call gets dissected against a shared scorecard, hosted by a rotating peer, with the manager speaking last or not at all. One async clip shared in a team channel — a 90-second moment worth stealing. For a team of eight reps, that is roughly four to five hours of manager time weekly, which is the real cost and the reason coaching dies: it is not free, and pretending it is guarantees it loses to the forecast call.

Monthly rhythm. One themed role-play block — objection handling one month, multi-threading the next, pricing the month after — and one manager-of-managers session where leaders coach each other on their coaching. That second one is where second-line leadership either creates the culture or reveals it does not believe in it.

Days 1–30: instrument and model. Turn on call recording. Publish one shared call scorecard — five to seven dimensions maximum, each with a concrete behavioral anchor rather than a vague adjective ("asked at least three questions about business impact before mentioning product" beats "good discovery"). Then do the single highest-leverage thing available to you: share one of your own losing calls first, in front of the whole team, and narrate what you would do differently. Nothing else you do in the first month matters as much. Reps calibrate their safety entirely on whether the highest-status person in the room will be visibly bad at something.

Days 31–60: distribute it. Hand the team call-review to a rotating peer host. Introduce GROW explicitly in 1:1s — name the framework so reps recognize the shape and can run it on each other. Pick exactly one leading indicator per rep to track. Resist expanding the scorecard.

Days 61–90: defend it. Tie a coaching goal into each rep's development plan so it is a stated expectation, not a favor. Push toward peers running two of every three reviews. Report the movement in stage-conversion rates to leadership so the time survives the next pipeline crunch. Every coaching program that gets killed gets killed during a bad quarter, and the only defense is data collected before the bad quarter arrives.

What to measure. Quota is a lagging indicator and a terrible coaching metric — by the time it moves, the quarter is over and you cannot attribute the change. Track these instead:

• Coaching coverage — percentage of reps who received a structured session this week. Target 100%, and treat anything under 80% sustained as a system failure rather than a manager failure.
• Behavior-change rate — of the commitments reps made, how many showed up in the next two recorded calls? Verify in the recording. This is the single most honest number in the whole program and typically the most uncomfortable when first measured.
• Scorecard movement — discovery-quality score trending over a four-week window, not week to week, because single-call variance swamps the signal.
• Stage conversion — discovery-to-demo and demo-to-proposal rates respond to coaching faster than win rate does, usually within one to two sales cycles rather than three.
• Ramp time — weeks to first closed deal for new hires is the cleanest proof a coaching system works, because new hires have no legacy pipeline to confound the result. If your cycle is 90 days, expect to need two or three cohorts before the trend is trustworthy.
• Voluntary participation — how many reps submit a call nobody asked them to submit? This is the actual culture metric. Coverage measures compliance; volunteering measures belief.

Realistic timelines. The instrumentation — recordings, scorecard, calendar blocks — can be live in a week. The cadence takes about 90 days to stop feeling imposed. The psychological safety that makes a rep volunteer their worst call takes six to twelve months and is entirely gated on leadership modeling it repeatedly, not once. Expect a visible dip around week five or six when novelty wears off and the first forecast crunch tests whether you will cancel. That moment is the program.

Trade-offs: manager-led, peer-led, or buy it

There is no single right coaching model, and the honest framing is a set of trade-offs against team size, manager bandwidth, and how tenured the team is.

Manager-led coaching gives the highest quality per session when the manager is genuinely skilled, and it carries authority — what the manager says sticks. It also does not scale past roughly six to eight direct reports, it creates a single point of failure, and it carries the evaluation problem: the person coaching you also decides your comp, which caps how honest reps will be about weakness no matter how safe you claim the room is. Best fit: small teams, junior reps, early install phase where you are setting the standard.

Peer-led coaching scales without adding manager headcount, produces better psychological safety because nobody in the room controls your compensation, and has a strong secondary effect — teaching a skill is how reps consolidate it, so the coach improves as much as the coached. The risks are real: without a scorecard, peer review degrades into either mutual reassurance or the loudest rep imposing their idiosyncratic style. Peers also cannot address will problems or performance issues, and should never be asked to. Best fit: teams past eight reps with a published standard already in place.

Buying it — external sales training, a coaching platform, or a fractional coach — delivers a proven framework quickly and brings outside credibility that a homegrown standard sometimes lacks. It costs real budget, and the classic failure is the two-day workshop that produces enthusiasm and no behavior change because nothing in the weekly calendar changed afterward. External training works as an accelerant on top of an existing cadence and fails as a substitute for one. Best fit: teams that need a framework fast, or where internal managers genuinely lack the skill and you need a floor immediately.

AI-assisted review is the newest option: automated call summaries and scoring that flag talk-ratio, monologue length, question count, and competitor mentions without anyone listening to the full recording. This removes the largest practical excuse — "I didn't have time to listen" — and makes triage cheap: a manager can scan twenty calls for signal in the time it used to take to review two. What it cannot do is judge whether a question was the *right* question at that moment, read the buyer's hesitation, or hold a rep to a commitment. Treat it as a triage layer that tells you which calls to actually watch, never as the coaching itself. Teams that mistake a dashboard of talk-ratio scores for a coaching culture end up with reps optimizing the metric and no improvement in outcomes.

Most teams land on a blend that shifts over time: manager-led during install to set the standard, peer-led once the standard is shared, AI-assisted throughout for triage, external training when a specific skill gap is too wide to close internally.

One trade-off deserves naming directly: coaching time is not free, and any honest cadence takes four to six hours a week from a frontline manager. That time comes from somewhere. The usual honest answer is killing a low-value recurring meeting or shrinking the pipeline scrub, not asking managers to absorb it on top of everything else — which is how coaching programs quietly become optional by week seven.

The failure modes that kill it, and the countermoves

Only coaching the strugglers. This is the single biggest culture-killer. The moment coaching is reserved for people missing quota, it becomes a scarlet letter, and every rep spends energy avoiding it. Countermove: coach A-players visibly and on harder material — deal strategy, multi-threading into the economic buyer, mentoring a newer rep. When the top rep is openly being coached on the next level, coaching reads as an investment rather than a warning.

Coaching the deal instead of the skill. Deal review asks "what's the next step on Acme?" Skill coaching asks "what pattern keeps stalling your deals at this stage?" Both are necessary; conflating them means the skill conversation always loses, because a live deal is always more urgent. Countermove: separate meetings, separate names, and never let the pipeline scrub eat the coaching block.

Rescuing. The manager jumps on the call, handles the objection, closes the deal. The quarter is saved and the rep learned nothing except that the manager will handle hard moments. Countermove: if you must join, brief the rep beforehand on the one thing they will handle, and debrief on that specific moment afterward. Rescue with a rep embedded in it, or the rescue is pure cost.

No follow-through. Feedback that is never checked is feedback that never happened. If a rep commits to a behavior and nobody verifies it on the next two calls, you have taught them that commitments in coaching are theater. Countermove: open every session by checking the previous commitment in the recording. Thirty seconds, every time, no exceptions.

Coaching everyone identically. Your top rep needs a stretch, your new hire needs scaffolding and repetition, your stalled mid-tenure rep needs a diagnosis before anything else. Running the same template on all three wastes the first, overwhelms the second, and misses the third entirely.

Confusing a will problem with a skill problem. No amount of GROW fixes a rep who has checked out. Countermove: run the diagnostic honestly, and when the answer is will, have the direct performance conversation instead of scheduling a fourth role-play. Coaching a will problem is a kindness that helps nobody, and it corrodes the program's credibility with everyone watching.

Tying coaching to evaluation. The instant coaching notes feed performance reviews or comp decisions, psychological safety evaporates and reps bring you only their competent calls. Countermove: keep coaching artifacts explicitly out of the review process, say so out loud, and make sure your behavior confirms it — one instance of a coaching admission surfacing in a review kills the program permanently.

Leaving it optional. Anything not on the calendar with a recurring invite does not happen during a busy quarter. Countermove: recurring blocks, a named owner, and a rule that coaching is the last thing cancelled rather than the first.

Skipping the scorecard. Without a shared, behaviorally anchored standard, every reviewer coaches toward their personal style and reps receive contradictory guidance from three directions. Countermove: publish it, keep it to five to seven dimensions, and revise it quarterly rather than continuously.

Related questions

Does a coaching culture require call recording software?

No, but it removes the biggest friction. Recordings end debates about what was said and let reps watch their own improvement. Without a tool you can still coach live-call ride-alongs and role-plays — you lose objectivity, async review, and the ability to coach across time zones.

How many reps can one manager coach well?

Roughly six to eight before quality degrades. Past that, a manager can maintain coverage or maintain depth, not both. The standard answer is shifting weight to peer-led review with the manager auditing quality and owning performance conversations.

Should coaching notes go in the CRM?

Keep skill-coaching notes out of systems tied to performance review. Deal-specific next steps belong in the CRM; a rep's skill gaps and commitments belong somewhere the rep trusts. Mixing them is how psychological safety quietly dies.

What's the first thing to do if we have nothing today?

Turn on recording and share one of your own losing calls with the team. Instrumentation plus modeled vulnerability is week one. Frameworks, scorecards, and cadence all land better once reps have seen a leader be publicly imperfect.

How do you coach a fully remote sales team?

Lean harder on recordings and async clip-sharing, since you cannot overhear calls. Run the team review on video, rotate a peer host, and protect the block absolutely. Remote teams need more deliberate cadence than co-located ones, not less.

FAQ

How is a coaching culture different from just doing 1:1s?

A 1:1 is a meeting; a culture is a norm. In a coaching culture, reps coach each other in peer reviews, share losing calls without being asked, and request feedback unprompted. 1:1s are one input — the culture is what happens when the manager is not in the room. You can run perfect 1:1s for a year and still have no culture if nothing propagates between reps.

How long does it take to build one?

About 90 days to install the cadence and six to twelve months for it to feel automatic. Recording and a shared scorecard can be live in week one. The slow part is psychological safety, which depends almost entirely on leadership repeatedly modeling being visibly imperfect — not once at kickoff, but consistently enough that reps believe it is safe.

What if managers say they don't have time?

Then it is a priority problem, not a time problem — but respond to the substance rather than dismissing it. An honest cadence costs four to six hours weekly per manager. Find that time by killing a low-value recurring meeting or shortening the pipeline scrub, and report conversion-rate movement so the block defends itself when the quarter gets tight.

Should top performers get coached too?

Yes, and it is the fastest way to make coaching feel like investment rather than punishment. Coach A-players on harder material: deal strategy, multi-threading, mentoring newer reps, expanding accounts. When the best rep on the team is openly being coached, nobody reads a coaching invitation as a warning sign.

Can AI call summaries replace manager coaching?

No. They excel at triage — flagging talk ratio, monologue length, question count, competitor mentions — so a manager can find the calls worth watching in minutes instead of hours. They cannot judge whether a question was right for that moment, read buyer hesitation, or hold a rep accountable to a commitment. Use them to decide what to review, never as the review itself.

Where does RevOps fit in building this?

RevOps owns the visibility layer that makes coaching possible: recording coverage, clean stage definitions so "stalled at discovery" is a legible pattern, and the reporting that proves coaching moved stage conversion. Without that instrumentation, coaching runs on anecdote and gets cut first when budgets tighten because nobody can defend it with data.

Sources

• Harvard Business Review — Companies Are Finally Doing Something About Bad Sales Coaching
• Performance Consultants — The GROW Model
• Gong — Sales Coaching Resources
• RAIN Group — Sales Coaching Techniques
• Sandler — The Importance of Coaching in Sales
• Salesforce Blog — Sales Coaching
• Winning by Design — Resources
• Harvard Business Review — The Leader as Coach

Related on PULSE

• [How do you coach a rep who struggles to create urgency?](/knowledge/cg0798)
• [How do you create coaching playbooks for common rep gaps?](/knowledge/cg0197)
• [How do you coach a rep to create urgency without fake deadlines?](/knowledge/cg0085)
• [How do you coach a sales team through a major change like a new product?](/knowledge/cg0199)
• [How do you scale sales coaching as your team grows?](/knowledge/cg0192)
• [Top 10 discovery questions every sales manager should ask their team](/knowledge/cg0859)