How do you coach a rep whose forecast is always wrong?

The outcome you should expect

The realistic destination for a rep whose forecast is always wrong is not perfection — it is predictability. A rep who is reliably 15% high is worth more to a RevOps team than a rep who swings between +40% and −25%, because a consistent bias can be corrected with a coefficient while variance cannot be corrected at all. So set the target as *tightening the band*, not *hitting the number*. In practice that means moving a rep from an unusable ±35% swing down to a ±10% band over roughly one full quarter of disciplined weekly inspection, with the first visible change showing up in behavior — evidence attached to Commit deals — two to three weeks before it shows up in accuracy.

Expect the accuracy curve to get worse before it gets better, and tell the rep that in advance so the dip doesn't feel like failure. When you install real evidence gates, deals that used to sit in Commit on vibes get pushed down to Best Case, and the rep's committed number drops — sometimes 20-30% in the first cycle. That looks like a collapse on the board and it is actually the fix working. The old Commit total was fiction; the new smaller one is bankable. Prepare your own boss for that dip too, because if leadership panics at the smaller number and pressures you to "get the pipeline back up," you will re-teach the rep that inflation is safer than honesty and undo the whole program in one meeting.

The second outcome, and the one that matters more over a year, is that the rep gets better at *selling*, not just at reporting. Forecast accuracy is a lagging measure of qualification discipline. A rep who cannot tell you the economic buyer, the quantified cost of inaction, and the buyer-agreed close date is not a bad forecaster — they are running deals they do not control. When you force the evidence, they start gathering it earlier in the cycle because they know Friday is coming, and gathering it earlier changes the deal itself. Champions get tested sooner. Ghost deals get killed in week three instead of week eleven. Cycle time compresses because the rep stops nursing corpses. Managers who run this program consistently report the same secondary effect: pipeline shrinks, close rate rises, and the total closed number stays flat or climbs even though the pipeline looks thinner on paper.

There is a third outcome worth naming because it changes how you sell the program to the rep. Reps you can predict get treated better. When a rep's Commit is trustworthy, you fight harder for them at the deal desk, you approve the discount faster, you give them the inbound lead, and you defend their quarter at the QBR without hedging. Say that out loud in the first conversation. Accuracy framed as a compliance chore gets grudging compliance; accuracy framed as the thing that makes you their advocate gets actual buy-in. Reps who are always wrong on forecast are usually not resistant to accuracy — they have simply never been shown what it buys them.

Finally, expect the fix to be partly yours. If your stage definitions are vague, if "Commit" has never been defined in writing, if the CRM has no field for the buyer-agreed close date, then a meaningful slice of the rep's error is a system you handed them. Audit that before the first 1:1. It is deeply unfair — and coaching-destroying — to hold someone accountable to a standard that exists only in your head, and reps can smell that mismatch instantly.

What drives that outcome

Forecast error is a downstream symptom, and the four upstream drivers are skill, will, knowledge, and system. Sort every miss into one of those buckets before you plan a single conversation, because the four require genuinely different interventions and applying the wrong one makes things worse.

Skill is the rep who cannot read buying signals. They heard "let me take this to the team" and logged it as a green light, when it actually meant no decision maker was in the room. This rep is coachable with call review — pull the recording of a Commit deal that slipped and listen together for the exact moment the buyer hedged. Skill gaps respond to reps and repetition, not to rules.

Will is the rep who knows the deal is soft and calls it Commit anyway, because a big Commit number keeps management off their back this week and next week is a problem for next-week them. Or the mirror case: the sandbagger who buries a signed-in-spirit deal in Pipeline so they can pull out a hero quarter. Will problems are incentive problems. Look at what your comp plan, your leaderboard, and your own reaction pattern actually reward. If you visibly get annoyed when someone downgrades a deal, you have built a sandbagging machine and the rep is behaving rationally inside it.

Knowledge is the rep who genuinely does not know what Commit means at your company. This is the cheapest fix and it is astonishingly common. Ask three reps to define Commit in writing without conferring and you will often get three different answers. If that happens, you do not have a rep problem, you have a definitions problem.

System is stage exit criteria that do not exist, a CRM with no field to hold the evidence, an opportunity record where close date defaults to end-of-quarter automatically, or a forecast category that auto-maps from stage so the rep never makes a judgment call at all. Fix the system before you coach, or you will coach against a headwind you created.

The diagnostic move that separates these fast is a four-quarter variance pull: committed dollars versus closed dollars, per quarter, per rep. Consistent direction means calibration — the most coachable case. Wild swings mean no repeatable qualification method. And if the whole team misses in the same direction by roughly the same margin, stop coaching the individual entirely; that is a RevOps definitions problem and no amount of 1:1 time will fix it.

One adjacent driver worth checking: territory and segment mix. A rep who moved from mid-market to enterprise six months ago will forecast badly for two or three quarters regardless of skill, because their internal model of "this feels like a done deal" was trained on 45-day cycles and now they are running 160-day cycles with procurement, security review, and a signature chain they have never navigated. That is not a coaching failure, it is a recalibration period. Treat it as one and set expectations accordingly.

Benchmarks and realistic ranges

Anchoring the conversation in numbers keeps it out of the realm of opinion, and opinion is where these conversations go to die. Use ranges rather than absolutes, and be honest that the right target depends heavily on deal size, cycle length, and segment.

Forecast accuracy band. For a rep running transactional or mid-market deals with 30-60 day cycles, ±10% of committed is a reasonable steady-state target and a good rep gets there. For enterprise reps carrying six or seven deals a quarter, ±10% is arithmetically brutal — losing one deal out of six blows the band regardless of judgment quality. In low-deal-count territories, judge the rep on *deal-level* call accuracy (did each individual Commit land in the period they said?) rather than aggregate dollar accuracy, because with small n, dollar accuracy is mostly noise.

Commit-to-close conversion. Roughly 80-90% of Commit dollars should land in the forecasted period. Below 70% and the category has lost its meaning entirely. Above 95% sustained is not excellence, it is sandbagging — the rep is only calling deals Commit once they are already effectively signed, which makes the number accurate and useless, because you learn about the revenue too late to do anything with the information.

Best Case conversion. Expect something in the 25-50% range. If Best Case converts at 80%, your rep is hiding real Commits there. If it converts at 5%, Best Case has become a graveyard for wishful pipeline and it is not informing anyone's planning.

Slip rate. The share of deals that move out of their forecasted period is one of the cleanest coaching signals available, because unlike a lost deal it isolates *timing judgment* specifically. A rep with a 40% slip rate does not have a losing problem, they have a close-date problem — usually because the close date was picked by the rep to fit the quarter rather than agreed with the buyer against a real business event.

Time to improvement. Evidence completeness — the share of Commit deals carrying full documented proof — should move within two weeks, because it is a behavior fully under the rep's control. Accuracy follows two to four weeks behind that, because deals already in flight were qualified under the old regime and have to work through the system. If evidence completeness is at 90% for six weeks and accuracy has not moved, you have misdiagnosed: the problem is not forecast discipline, it is the qualification or the deal skills underneath, and you pivot the coaching there.

A benchmark to ignore. Do not benchmark a rep against team-average accuracy without adjusting for territory. A rep in a mature, high-repeat-purchase patch will forecast better than a rep opening a greenfield vertical, and hammering the second rep for it teaches them to sell only safe deals — exactly the opposite of what you want from a territory you are trying to open.

Risks, edge cases, and failure modes

Punishing honesty. This is the program-killer. A rep moves a deal from Commit to Best Case with clean reasoning, and the manager sighs, or asks "so what happened," or says "we really needed that one." Within one cycle the whole team learns that downgrades cost you something and inflation costs you nothing until quarter-end. Reward the accurate downgrade visibly and immediately — name it in team meeting as good judgment — or accept that you will never get honest data.

Rescuing the rep. Quietly haircutting their number yourself so the roll-up works keeps your forecast clean and guarantees the rep never learns. Worse, it makes your own number a private guess that nobody can inspect, which is the same disease one level up. If you must adjust, adjust openly and tell the rep exactly what you adjusted and why.

Coaching the deal instead of the rule. Spending the 1:1 re-categorizing this week's six Commit deals fixes this week and nothing else. You will do the identical exercise next Friday, forever. Coach the decision rule the rep uses, then test the rule against deals — the deals are the exercise, not the objective.

Over-instrumenting. There is a real failure mode where the manager responds to bad forecasting by adding six required fields, a mandatory MEDDIC scorecard, and a deal-review deck. The rep now spends four hours a week on forecast hygiene and less time selling, and the data quality goes *down* because they start filling fields to pass the check rather than to record reality. Keep the evidence requirement to three or four things that genuinely predict close: economic buyer engaged, pain quantified, buyer-agreed close date, paper process known.

Tooling as a substitute for judgment. Engagement-signal platforms give you objective inputs the rep may be ignoring — whether the economic buyer has actually opened anything, whether multithreading is real or one lonely champion. Used as a mirror in the 1:1 they are excellent. Used as an oracle that overrides the rep's read, they teach the rep to stop thinking, and then you have automated the guessing instead of fixing it. The signal tells you *where to ask a question*, not what the answer is.

The blameless-system edge case. Sometimes the rep is right and the forecast is wrong for reasons above them: a pricing change mid-quarter, a product gap discovered in security review, a procurement freeze at a major account. Check for correlated misses across the team in the same period before you make it a person problem. A whole team missing the same way in the same quarter is almost never a coaching issue.

The fit conversation. After a clean 90 days — clear definitions, weekly inspection, honest feedback, no moving goalposts — if the rep still cannot qualify, this stops being a coaching problem. Say so plainly and move to a performance conversation. Dragging forecast coaching into month six is unkind to the rep and corrosive to the team, who can all see it. The tell is usually not the accuracy number; it is whether the rep can *explain* their own misses. A rep who can articulate why they were wrong is learning. A rep who is surprised every quarter is not.

Remote and async risk. If your inspection happens over Slack rather than live, evidence review degenerates into checkbox reporting fast. Keep at least one live weekly touch where the rep has to defend a deal out loud, because verbal defense exposes thin qualification in a way a form field never will.

A practical rollout plan

Run it in three 30-day arcs layered over a weekly loop. The weekly loop is where the coaching actually lives; the 30-day arcs just describe how much you are in it.

Days 1-30 — definitions and evidence. Before the rep is involved, write the category definitions yourself and pressure-test them against last quarter's closed deals: would this rule have classified them correctly? Then bring it to the 1:1 and co-author the final version. Use the GROW frame — open with the goal ("I want to make you the most trusted forecaster on this team, because reps I can predict are the ones I can fight for"), establish reality with the four-quarter variance pull rather than your opinion, generate the options together, and close on an explicit commitment with a scheduled first inspection before they leave the room. The rule itself should be short enough to say from memory: a deal is Commit only if the economic buyer is engaged, the pain is quantified in dollars, and there is a mutual close plan with a date the buyer agreed to. Everything else is Best Case. During this arc you inspect every single Commit weekly.

Days 31-60 — calibration. Monday the rep submits. Friday you score last week against reality together, deal by deal, and discuss only the misses. Track their personal accuracy as the headline metric and show them the trend line — the trend is more motivating than any individual week. Layer in two drills. The Commit defense drill: the rep presents three Commit deals and you play a skeptical CRO demanding evidence; anything they cannot defend gets re-categorized on the spot, no penalty. The reverse forecast: have them forecast a *peer's* pipeline cold from the CRM, which exposes how thin qualification evidence looks from the outside and transfers back to their own deals almost immediately. Add call review for the specific deals that slipped — listening to the moment a buyer hedged, and hearing themselves take it as a yes, does more than any amount of explanation.

Days 61-90 — independence. Step from inspecting every Commit to spot-checking two or three, and shift the rep to self-reporting accuracy at the start of the 1:1 before you say anything. That reversal matters: a rep who opens with "I was 12% high last week, here's why" has internalized the loop. The exit criterion is a rep forecasting within their agreed band, unprompted, for three consecutive cycles.

One rollout warning: do not run this on your whole team at once as a policy rollout. It reads as a crackdown, your accurate forecasters resent the new overhead, and the rep you were actually trying to help gets to hide inside a group initiative. Run it with the individual, let the results show up in the roll-up, and let other reps ask you for it.

Related questions

Should forecast accuracy be part of a rep's comp plan?

Rarely, and never as the primary lever. Paying for accuracy invites sandbagging — the safest way to be accurate is to commit only what is already signed. If you tie anything to it, tie it to Commit-to-close conversion within a band, and keep the weight small relative to quota attainment.

How does this change for a rep in their first two quarters?

New reps should not be forecasting Commit independently at all. Have them submit a recommendation with evidence, you make the call, and you explain your reasoning each week. Hand over the call once they can predict your decision correctly three cycles running.

What if the whole team's forecast is always wrong, not just one rep?

Stop individual coaching immediately. Uniform team error means the definitions, stage criteria, or CRM defaults are broken — a RevOps fix, not a coaching one. Rewrite the category definitions, test them against last quarter's closed-won data, then re-launch with the team.

Does forecast coaching work the same for renewals and expansion?

The structure holds but the evidence changes. For renewals, the predictive signals are product usage trend, executive sponsor continuity, and support escalation history rather than MEDDIC-style buying committee mapping. Same weekly inspection loop, different proof.

How much time should this take per rep per week?

Roughly 30 minutes of your prep plus a 30-45 minute 1:1 during the first 30 days, dropping to about 20 minutes total by day 90. If it consistently exceeds an hour, you are re-qualifying deals for the rep rather than coaching them.

FAQ

What if the rep is always overly optimistic?

The chronic optimist labels deals Commit based on belief rather than evidence, and they are usually sincere about it — that is what makes it hard. Do not argue about any single deal. Instead require documented proof before a Commit is allowed: the economic buyer named and met, the business pain quantified in dollars, and a close date the buyer agreed to out loud. Then inspect one deal live in the 1:1 and let their inability to answer four qualification questions make the point for you. The pattern does the confronting, not you.

How do I handle a rep who consistently sandbags their forecast?

Sandbagging is almost always a rational response to how the environment treats misses. Start by asking, honestly, what happens on your team when someone commits and loses — if the answer is "public grilling," you built this. Then set a symmetric rule: any deal with a verbal commitment or a contract in signature moves to Commit within 24 hours, and accurate downgrades are treated as good judgment rather than failure. Make the cost of hiding upside visible by showing what a surprise quarter does to hiring plans and territory decisions downstream.

What if the rep has no real qualification process at all?

Give them a single lightweight checklist and require them to walk it out loud for every Commit deal. MEDDIC works well; so does a stripped-down four-item version. The specific framework matters far less than using the same one every week, because consistency is what converts it into judgment. Expect the first three weeks to feel mechanical and slow. That is the point — you are building a habit, and habits feel artificial before they feel automatic.

Can revenue tooling actually fix a rep's forecasting?

Engagement and conversation-intelligence platforms help by supplying objective inputs the rep may be discounting — whether the economic buyer has engaged at all, whether the deal is genuinely multithreaded, how the language in the last call compares to deals that historically closed. Use them as a mirror inside the coaching conversation, not as an override. A tool that silently re-scores the rep's deals teaches them their judgment does not matter, which is the opposite of coaching. The signal should prompt a question, not supply a verdict.

How long before a rep's forecast accuracy actually improves?

Behavior changes within about two weeks; accuracy follows two to four weeks after that, because deals already in the pipeline were qualified under the old standard and have to clear the system. A skill or system gap typically resolves inside 4-8 weeks of consistent weekly inspection. A will or trust issue takes longer because you are rebuilding the environment, not teaching a technique, and it may require changing something about your own reaction pattern first.

What if nothing improves after a full 90 days?

If definitions were clear, inspection was weekly, feedback was honest, and the goalposts never moved, then the constraint is not forecasting. Usually the underlying issue is qualification or deal control — the rep cannot forecast deals they do not actually influence. Pivot the coaching to discovery and multithreading for one more cycle. If that also produces nothing, treat it as a fit conversation and be direct about it, because a rep who is surprised by their own results every quarter is not learning from them.

Sources

• Harvard Business Review: Improving the Accuracy of Your Sales Forecast
• MEDDIC Academy: What is MEDDIC?
• Gong: Sales forecasting research and guides
• Clari: Sales forecasting best practices
• Salesforce: Sales forecasting methods
• RAIN Group: Sales coaching resources
• HubSpot: The sales manager's guide to sales forecasting
• McKinsey: Insights on sales and channel management

Related on PULSE

• [How do you coach a rep who sandbags their forecast?](/knowledge/cg0110)
• [How do you coach a rep whose pipeline keeps drying up?](/knowledge/cg0024)
• [How do you coach a rep whose attitude is hurting the team?](/knowledge/cg0148)
• [How do you coach a rep whose demos are technically strong but boring?](/knowledge/cg0065)
• [How do you coach a rep whose emails are too long and salesy?](/knowledge/cg0043)
• [How do you coach a rep whose tone kills their cold calls?](/knowledge/cg0040)