Sales Forecasts: The One Fix

In most quarterly business reviews, the sales forecast misses. It misses by enough that the CFO notices, the CEO notices, and the board notices. The conversation about why it missed produces, with depressing regularity, the same set of remedies.

New BI. New CRM hygiene. New MEDDIC discipline. New deal-review cadence. A consultant. A tool. A workshop on multi-threading. Sometimes a leadership change at the front-line manager level. Occasionally a complete re-architecture of the pipeline-stage definitions.

Each of these helps, marginally. We have watched several of them implemented carefully and with goodwill. The forecast keeps missing. It misses less wildly, perhaps, but the directional miss — the structural overshoot at the start of the quarter that resolves into the structural shortfall in the last two weeks — persists across every tool change and every methodology rollout we have observed.

The reason the persistent fixes don't work is that they treat the wrong layer of the problem. Tooling and methodology improve the inputs. They do not correct for the systematic bias in the way the inputs are produced. There is one fix that does, it is brutally simple, and it is almost never done.

The fix

The fix is to keep each rep's historical forecast accuracy visible, by individual, and let it discount the next forecast automatically.

The rep who has missed by 30% on the downside for four quarters running is not a 100% commit. They are, on the math, closer to a 70% commit. The rep who has consistently come in 10% above their commit is not a 100% commit either; they are roughly 110%. The CRO should not be running their territory model on the assumption that every rep's stated forecast is equally credible, because the empirical record says they aren't, and the record is sitting in the CRM right now waiting to be queried.

The mechanics are not complicated. For each rep, in each forecast category (commit, best case, pipeline), compute the rolling four- or eight-quarter accuracy ratio: what fraction of their stated commit, on average, has actually closed. Multiply this quarter's stated commit by that ratio. That is the adjusted commit. Roll it up.

This is a five-line change to the forecasting spreadsheet. It does not require new software. It does not require new process. It requires, mostly, the willingness to put numbers next to people's names and let the numbers do what they are going to do.

The improvement in forecast accuracy is, in our experience, larger than any tooling or methodology change we have seen. Companies that install this typically see their consolidated commit-to-close ratio tighten from the 70s into the 90s within two or three quarters — not because the reps got more accurate, but because the system stopped treating different reps' forecasts as if they carried equal information.

Why this is rarely done

The math is not the problem. Any FP&A analyst could build this in an afternoon. The CRM contains the historical data. The territory model is already a spreadsheet. The reason it doesn't exist is political, and the politics are worth being honest about, because the install will fail if you don't address them directly.

The first political reason is optics. A rep-level accuracy column reads, to most sales organizations, like a surveillance instrument. The CRO who introduces it without preparation will be accused, accurately, of grading individuals by a number they did not previously know they were being graded by. Some of the reps with the worst track records are also the loudest, longest-tenured, and most politically connected. The conversation about their numbers will be a conversation about them.

The second political reason is fear of the actual data. In every organization where we have run this exercise, the distribution of rep-level forecast accuracy is wider than anyone expected. The best forecasters and the worst forecasters are usually not who the sales leadership would have predicted. The longest-tenured reps are not always the most accurate. The top-quota reps are sometimes the worst forecasters, because the same swagger that closes deals also produces aggressive commits that don't pencil out. The data, once visible, disrupts the existing hierarchy of credibility. Some of that disruption is welcome. Some of it is not, and the organization will resist it.

The third political reason is the CRO's own number. A consolidated forecast that has been honestly discounted against rep-level accuracy is, almost always, a lower forecast than the one the CRO has been carrying. The first quarter the CRO presents the honest number to the CEO, the conversation is uncomfortable. The number is below the previous commit. The CRO is, in effect, telling the CEO that the forecast they have been receiving for the last two years has been systematically optimistic, and that the new, honest number is what they should plan against. Many CROs would rather keep the old number and absorb the quarterly miss than have the conversation that produces the new one.

The reason most companies don't do this is not that they can't. It is that the first month of doing it requires the CRO to tell the CEO, in writing, that the numbers they have been planning against were wrong. After that month, the company is better off. Before that month, it is unstable. The unstable month is what blocks the install.

How to do it without the team revolting

The political problem is real, but it is solvable. The pattern that works, in our experience, has three elements.

Frame the metric as the rep's tool, not the manager's weapon. The accuracy ratio belongs, first and most visibly, to the rep. They see their own number weekly. They are encouraged to use it to adjust their own commit before it rolls up. The metric becomes, for the rep, an instrument for getting their own forecast right, not a verdict on whether they are a good salesperson. The same rep can be a strong closer and a poor forecaster — these are different skills, and most sales organizations conflate them.

Calibrate against forecast accuracy separately from quota attainment. Quota attainment measures whether the rep closed the number. Forecast accuracy measures whether the rep predicted what they would close. The two are different, and the second is a more useful signal for the CRO's roll-up. A rep at 120% of quota who consistently forecasts at 80% of what they close is a strong closer and a strong forecaster — but they are sandbagging, and the CRO needs to know that. A rep at 95% of quota who consistently forecasts at 130% is missing their number and over-committing — a worse problem than the simple quota miss suggests.

Install the discount as a system feature, not a manager judgement. This is the key. If individual managers are applying their own discounts to their reps' commits — informally, in their heads, based on what they think of each person — the system already runs on rep-level accuracy adjustments, just badly and politically. Surfacing the adjustment as a formal, mechanical, transparent calculation removes the manager-judgement layer and replaces it with arithmetic. The reps experience this, after the initial discomfort, as fairer than the previous system, because the previous system was a black box and the new one is not.

A practical addition: pair the install with the five-line forecast format for the consolidated forecast that reaches the CFO. The five-line format makes the historical track record a required field, which means the rep-level accuracy data has a natural home in the artifact the CFO actually sees. The two changes reinforce each other.

The CRO conversation

The conversation the CRO needs to have with the CEO, before any of this is installed, is the conversation that the install rises or falls on. We have helped a number of CROs prepare for it, and the version that works runs roughly as follows.

The historical forecast has been missing by X%. The miss is structural, not episodic. It has persisted across multiple tooling and methodology improvements. The root cause is that the consolidated forecast treats every rep's commit as equally credible, and the empirical record says they are not. We are going to install an accuracy-weighted forecast as the primary number we report internally. The unweighted forecast will continue to be tracked for comparison. In quarter one, the weighted number will be lower than the unweighted number, by roughly Y%. In quarter two onward, the weighted number should be meaningfully closer to actual closings than the unweighted number has historically been.

That conversation, run honestly, is a one-time political cost in exchange for a permanent improvement in the company's relationship with its own pipeline. Most CROs, once they have had it, wonder why they didn't have it years earlier.

We help CROs and CFOs build the discount model, design the rep-facing version, and run the first quarter's roll-up so that the political weight does not land entirely on the sales organization. If your last four sales forecasts have missed in the same direction by similar magnitudes, the conversation is worth having before the next quarterly cycle begins. The math is easy. The choice to use it is the part that matters.