How do you coach a rep to use CRM data to identify which deals need coaching attention first in 2027

The outcome you should expect

When a rep learns to read their own pipeline as a diagnostic instrument instead of a status report, three things change within about a quarter, and they change in a specific order.

First, the one-on-one gets shorter and sharper. A rep who has not been coached to triage arrives at the pipeline review with a list sorted by close date and dollar value, which is the order the CRM hands them by default. Every deal gets three minutes whether it needs thirty seconds or thirty minutes. A rep who has been coached to triage arrives with four or five deals already named as the ones that matter, and can say why each one is on the list — "this one's been stage 3 for 41 days and I've only ever talked to the champion," "this one slipped twice and the second slip had no reason attached." The manager's job shifts from finding the problems to solving them. In practice this recovers meaningful time from a typical hour-long review, and the recovered time gets spent on the two or three deals where a different action actually changes the outcome.

Second, the forecast tightens from the bottom up. Most forecast error is not managers being optimistic — it is reps carrying deals in a committed stage that have no evidence behind them. When a rep can identify which of their own deals are evidence-thin before the manager does, they either fix the evidence or move the deal, and both outcomes are better than the deal sitting in commit until the last week of the quarter. Teams that run this discipline consistently tend to see slip rates fall and the gap between early-quarter and end-quarter forecast narrow, because the correction happens in week three rather than week twelve.

Third — and this is the part that gets undersold — the rep starts self-coaching. Triage is a repeatable mental motion. Once a rep has run it twenty times with a manager, they run it on their own on a Monday morning without being asked. That is the actual deliverable. You are not trying to produce a rep who can answer "which deals need attention" when you ask; you are trying to produce a rep who has already asked themselves, and shows up with an answer. The measurable proxy is how often the rep raises a deal risk before the manager sees it in the dashboard. Track that ratio and you will see it climb from near zero to a majority within a couple of quarters if the coaching is consistent.

What you should not expect is a jump in win rate in the first thirty days. Triage does not save deals that were already dead; it finds them faster and stops the rep from spending time on them. The first visible effect is usually a *drop* in pipeline dollars, because the rep purges the deals that were never real. Managers who are not warned about this read it as a coaching failure and back off exactly when it is working. Tell the rep and their manager up front: pipeline gets smaller before conversion gets better.

There is an adjacent effect worth naming, because it shows up in the same quarter and often gets misattributed. Reps who triage their own pipeline become dramatically better at CRM hygiene, without anyone running a hygiene campaign. The reason is simple and behavioral — once the fields are the input to a decision the rep cares about, the rep maintains the fields. Next-step dates stop being blank. Close dates stop being end-of-quarter-by-default. Contact roles get filled in because the rep is using contact roles to check their own multithreading. Every RevOps team that has ever begged for data quality has been solving the wrong problem: hygiene is a symptom of whether the data is used, not a discipline to be enforced. Give the rep a reason to read the field and they will start writing it.

What drives that outcome

The mechanism is a small, fixed set of signals the rep learns to read in a specific order, and a rule about which signals earn coaching versus which just earn an action.

The core distinction — teach this first, because everything else depends on it — is between a deal that needs *work* and a deal that needs *coaching*. A deal missing a next step needs work: the rep books the meeting, done, no coaching required. A deal where the rep has been unable to get a second meeting with the economic buyer for three weeks needs coaching, because the rep has already tried and the attempt is failing. The triage list a rep brings to a one-on-one should be the second kind. If a rep brings ten deals that all need work, the coaching conversation is about their calendar, not their deals.

The signals themselves, in the order a rep should scan them:

Stage age versus your own stage benchmark. Not a vendor benchmark — your team's actual median days-in-stage per stage, which RevOps can pull in an afternoon from historical opportunity history. A deal sitting at 2x median in any stage is the single highest-yield flag because it is unambiguous and hard to argue with. Reps rationalize a lot of things; they cannot rationalize a deal that has been in "Evaluation" for 60 days when the team median is 22.

Last meaningful activity. Meaningful, not any. A logged email blast is not activity. The rep should be looking for the last two-way interaction — a reply, a meeting, a document opened by the prospect. Fourteen days without a two-way interaction on a deal in a late stage is a hard flag. In a self-serve or transactional motion that threshold might be five days; in a nine-month enterprise cycle it might be twenty-one. Set the number with the rep rather than for them and they will actually use it.

Contact breadth. Count distinct contacts with a logged interaction in the last 30 days. One contact on a six-figure deal in a late stage is a single point of failure, and it is the most common cause of a deal that "was going great" and then died with no warning. Reps consistently underestimate this risk because their one contact is enthusiastic — enthusiasm from a non-decision-maker reads exactly like progress right up until it does not.

Close-date movement. One push is normal. Two pushes on the same deal, especially two pushes where the new date lands conveniently at the end of the next quarter, is a strong signal the rep does not actually know the buying timeline. The CRM stores the history; most reps have never looked at it. Show them how in the field history and it lands hard, because it is their own behavior reflected back.

Stage-versus-evidence mismatch. The deal is in a stage that asserts something — "Proposal" asserts that a decision-maker has agreed on scope — and the CRM contains no artifact supporting it. No pricing document, no contact with a decision-maker role, no logged conversation about budget. This is the most valuable flag and the hardest to automate, which is exactly why it is worth coaching rather than dashboarding.

The ordering matters more than people expect. Stage age comes first because it is objective and needs no interpretation. Evidence mismatch comes last because it requires judgment, and a rep who has already worked through four objective filters is in a more honest frame of mind when they get to the subjective one. Run it in the reverse order and the rep spends the whole exercise defending their stage assignments.

One more driver: the rep has to run the filter themselves. Not receive a report — run it. A dashboard delivered to a rep produces compliance; a saved view the rep built produces ownership. The practical version is that you sit with them once and they build the list view in the CRM while you watch, with their own column choices and their own thresholds. It takes twenty minutes and it is the difference between a habit and a chore.

Benchmarks and realistic ranges

Numbers here should be treated as starting points to be replaced by your own data within a quarter, because the whole method depends on thresholds derived from your actual sales motion. That said, reps need somewhere to start, and "go figure out your own numbers" is not a usable instruction on day one.

Triage list size. A rep with 20-40 open opportunities should land on a coaching list of roughly 4-8 deals per cycle. Under three and the filters are too loose or the rep is protecting deals. Over twelve and the list is not a triage list, it is the pipeline again, and the rep will not act on it. If a rep consistently produces fifteen flagged deals, the problem is upstream — either qualification is broken or stages are being inflated at entry — and no amount of triage coaching fixes that.

Stage-age multiplier. Start at 2x the team median days-in-stage for that specific stage. Not the overall cycle median — per stage, because early stages and late stages behave completely differently. If a stage has a median of 10 days, 20 days is the flag. If a stage has a median of 45, 90 is the flag. RevOps should publish this table and refresh it quarterly; it drifts as the motion changes.

Activity silence window. Roughly 7-14 days for mid-market, 14-21 for enterprise, 3-5 for transactional or PLG-assisted motions. The correct way to set it is to look at closed-won deals and find the longest silence gap that still preceded a win — that gap plus a small buffer is your threshold. Most teams discover their real number is shorter than they assumed.

Contact breadth. Two engaged contacts is the floor for anything above your average deal size; three or more for anything you would call strategic. The number that matters is *engaged in the last 30 days*, not the total contact count on the account, which is usually inflated by marketing-sourced records that have never spoken to anyone.

Close-date pushes. Zero or one is normal noise. Two is a flag. Three or more and the deal should be re-qualified from scratch rather than coached — at that point you are not fixing a deal, you are deciding whether to keep it.

Time cost. The triage run itself should take a rep 15-25 minutes once the saved view exists. If it takes an hour, the view is wrong. The one-on-one that follows should spend the bulk of its time on two or three deals, not distribute evenly across all flagged ones. A useful discipline: the manager picks two deals from the rep's list to go deep on, and the rep picks one. That split keeps the manager from only ever coaching the deals they personally find interesting.

Ramp. Expect a new rep to need 6-10 supervised triage runs before doing it unprompted, and a tenured rep to need 3-5. Tenured reps are not faster because they are better at it — they are faster because they already know their deals and are mostly learning the filters. Tenured reps also resist harder, because the exercise surfaces deals they have been carrying for reasons of hope.

Data quality floor. This method needs three fields populated with reasonable reliability: stage, close date, and some form of activity logging. If activity logging is below roughly 60-70% coverage, the silence signal is noise and you should coach on stage age and contact breadth only until logging improves. Do not build a triage habit on a field the team does not fill in; the first time the filter flags a healthy deal because an activity was not logged, the rep stops trusting the whole system. Better to run with three signals that work than five where two lie.

An adjacent benchmark that helps calibrate: look at your closed-lost deals from the last two quarters and check how many of them would have been flagged by these filters 30 days before they closed lost. If the answer is under half, your filters are missing the actual failure mode of your business and need adjusting. If the answer is over 90%, your filters are probably flagging everything and providing no signal. Somewhere in the 60-80% range is a set of filters that is doing real discriminating work.

Risks, edge cases, and failure modes

Turning triage into surveillance. The fastest way to kill this is for the manager to run the filters themselves and open the one-on-one with "why has this deal been sitting for 40 days." Same data, opposite effect. The rep learns that the CRM is the instrument used to catch them, and their rational response is to manage the fields rather than the deals — close dates get set to safe distances, stages get held back, activity gets logged theatrically. You will have taught the rep to produce clean-looking data about deals you now know less about than before. The rule is simple: the rep runs the filter and brings the list. The manager never opens with a flag the rep did not name.

Flag fatigue. If every deal is flagged, nothing is. This happens when thresholds are copied from a blog post rather than derived from your data, or when a well-meaning RevOps team adds a seventh and eighth signal because they can. Five signals is about the ceiling for something a human runs in twenty minutes. When you add a signal, retire one.

The healthy quiet deal. Some deals are genuinely dormant for legitimate reasons — procurement, a budget cycle, a champion on leave — and will trip the silence filter every cycle. If the rep has to defend the same deal five weeks running, the exercise becomes theater. The fix is a documented reason and a review date on the deal, after which it is excluded from the silence flag until that date passes. This needs a field, and RevOps should build it rather than letting reps note it in a description somewhere.

Sandbagging by threshold. A rep who does not want a deal on the list learns which field to touch to keep it off. Logging a low-value activity resets the silence clock; a small close-date adjustment that is not technically a push; leaving a deal in an earlier stage so the stage-age benchmark for that stage is more forgiving. This is not usually malicious — it is a rational response to a system that generates unpleasant conversations. The countermeasure is not more filters; it is making the coaching conversation genuinely useful, so that having a deal on the list feels like getting help rather than getting caught. If reps are gaming the filters, the problem is the conversation, not the data model.

Stage-age noise in long cycles. In a nine-to-eighteen-month enterprise cycle, days-in-stage carries far less signal, and a 2x multiplier will flag deals that are progressing normally through a slow procurement process. For those motions, lean harder on contact breadth and evidence mismatch, and consider replacing raw stage age with a comparison against the deal's own historical pace — a deal that was moving every two weeks and has now been static for six is a signal; a deal that has always moved slowly is not.

CRM structure that cannot support the filters. Some orgs have no contact roles, no reliable activity capture, or stages so loosely defined that "Proposal" means five different things across the team. Triage coaching on top of that produces confident conclusions from bad inputs, which is worse than no triage. In that situation the real work is upstream, and it is RevOps work, not coaching work: define stage exit criteria, turn on activity capture, and enforce contact roles at the object level. Six weeks of that before starting the coaching habit will save a quarter of frustration.

Over-indexing on the flagged list. The list finds deals in trouble. It does not find the deal that is about to be won and needs one more push, or the account where the rep should be expanding rather than closing. A rep who only ever looks at their triage list develops a defensive orientation to their own pipeline. Balance it: one healthy deal per cycle gets discussed too, specifically to extract what is working so it can be repeated.

Manager inconsistency. If the habit is coached for six weeks and then dropped because the quarter got busy, it does not survive. This is the most common failure and the least interesting one, and it is worth saying plainly to whoever is sponsoring the rollout. The habit needs about a quarter of unbroken cadence to stick.

A practical rollout plan

Run this over roughly eight weeks. It works for a single rep and it works for a team of thirty; the difference is how many managers you have to train first.

Weeks 1-2 — RevOps builds the floor. Pull days-in-stage medians per stage from opportunity history and publish the table. Audit activity-logging coverage; if it is under 60-70%, fix capture before proceeding. Confirm contact roles exist and are used. Create the "dormant until" field with a date and a reason so legitimately quiet deals have somewhere to live. None of this is coaching yet, and skipping it is the single most common reason the rollout fails at week five.

Week 3 — one rep, one hour, build the view together. Sit with a rep. They share their screen. They build the saved list view — open opportunities, columns for stage, days in stage, last activity date, engaged contact count, close date, and close-date change count. They set the thresholds, using the RevOps table as the starting point but adjusting for their own segment. They save it under their own name. You do not build it for them and you do not send them a template.

Week 4 — first supervised run. The rep runs the view before the one-on-one and brings the flagged list. Work it together. For each flagged deal ask two questions: what does the flag mean here, and have you already tried something that did not work? The second question is what separates coaching from task assignment. End the session with the rep naming one behavioral change, not five deal actions.

Weeks 5-7 — repeat with declining supervision. The rep runs it alone and brings the list. Manager stops naming deals and starts asking why deals are or are not on the list. Around week six, start asking about the deals that did *not* get flagged — "anything you're worried about that the filter missed?" — because that question is what turns the filter from a rule into judgment.

Week 8 — tune and hand off. Review whether the flags predicted anything. Take the deals that closed lost during the pilot and check whether they were flagged in time. Adjust thresholds. Retire any signal that never produced a useful conversation. Then the rep owns it, and the manager's role reverts to showing up to the review and being useful about the two or three deals that matter.

A note on scaling this past one rep. If you are rolling it out to a team, train the managers first and have each of them run the eight weeks with one rep before touching the rest of their team. Managers who have not personally run the exercise will convert it into a dashboard review within two weeks, because that is the path of least resistance and it looks similar from the outside. The tell is whether the manager can name what a flag *means* as opposed to what it *is* — "single-threaded" is a fact; "you have one relationship and it is not with the person who signs" is a coaching frame.

The adjacent use case worth mentioning: the same triage motion applied to renewals and expansion works with two signal swaps. Replace stage age with time-to-renewal, and replace stage-evidence mismatch with product-usage evidence. Everything else — silence windows, contact breadth, date movement — transfers directly. Teams that build the habit on new business usually find CS adopts a version of it within a couple of quarters, because the underlying question is the same: of the things I am responsible for, which ones are quietly going wrong, and which of those can I still affect.

Related questions

How often should a rep run this triage?

Weekly for transactional motions, before each pipeline review for mid-market and enterprise. More often than weekly produces noise, since most flags need days to change state. Tie it to an existing recurring meeting rather than creating a new ritual.

Should the manager see the rep's list before the one-on-one?

Better if not. If the manager pre-reads and comes with their own flags, the rep learns their list does not matter. Let the rep present, then add what they missed at the end of the conversation.

What if the CRM data is too poor to trust?

Coach on the two or three signals whose underlying fields are reliable — usually stage age and close-date movement, since those are system-generated. Fix data capture in parallel. A filter built on an unreliable field destroys trust the first time it lies.

Does this replace formal deal reviews?

No. Triage decides which deals get reviewed. The deal review is what happens to the two or three deals that survive triage as genuinely needing help. They are sequential steps, not alternatives.

Can AI scoring in the CRM do this instead?

It can produce the list faster, but the coaching value comes from the rep making the judgment. Use scoring as a cross-check against the rep's own list — where they disagree is the most interesting conversation in the review.

FAQ

What is the single most useful signal if I can only pick one?

Days-in-stage against your own per-stage median. It is system-generated so it cannot be gamed easily, it needs no interpretation, and it correlates with the most common failure mode in almost every pipeline — a deal that has quietly stopped moving while everyone assumed it was progressing. Start there and add signals only once that one is habitual.

How do I stop this becoming a compliance exercise?

The rep runs the filter and owns the list; the manager never opens with a flag the rep did not name. The moment the manager becomes the one who finds the problems, the rep's incentive flips from surfacing risk to hiding it. Also: make sure a flagged deal reliably gets useful help, not just scrutiny.

Should the triage list be shared across the team?

Aggregate patterns yes, individual lists no. If RevOps notices that 60% of flags across the team are single-threading, that is a team-level enablement finding worth acting on. Publishing whose deals are flagged turns a diagnostic into a leaderboard and reintroduces every incentive problem you were trying to avoid.

What about deals where the rep genuinely cannot get more contacts?

That is a real coaching topic rather than a data problem. Some buyers gatekeep deliberately. The coaching is about how the rep asks for the introduction — tying it to something the champion needs, like a security review or an implementation conversation, rather than asking for access on the rep's own behalf.

How does this fit with an existing sales methodology?

It sits underneath whichever one you use. MEDDIC, Command of the Message, and their relatives all define what evidence a deal should have; triage decides which deals to examine for that evidence this week. If you already have a methodology, the stage-evidence mismatch signal should map directly onto its required fields.

Who owns this — sales management or RevOps?

RevOps owns the inputs: the benchmark table, field availability, activity capture, and the dormant-until mechanism. Sales management owns the habit and the conversation. The rollouts that fail almost always have RevOps trying to own the coaching or management trying to own the data model.

Sources

• https://hbr.org/2015/07/what-salespeople-need-to-know-about-the-new-b2b-landscape
• https://www.gartner.com/en/sales/insights/b2b-buying-journey
• https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights
• https://help.salesforce.com/s/articleView?id=sf.forecasts3_overview.htm&type=5
• https://knowledge.hubspot.com/deals/manage-your-deal-pipelines-and-deal-stages
• https://learn.microsoft.com/en-us/dynamics365/sales/manage-sales-pipeline
• https://hbr.org/2017/03/a-refresher-on-ab-testing
• https://sloanreview.mit.edu/topic/data-data-analytics-and-machine-learning/

Related on PULSE

• How do you build a pipeline review agenda that surfaces risk instead of status
• What CRM fields actually predict whether a deal will close
• How do you set stage exit criteria that reps will actually follow
• How do you coach a rep who is consistently single-threaded
• What is the right cadence for one-on-ones between a manager and a quota-carrying rep
• How does RevOps measure whether sales coaching is working