Forecasting Without the Fiction

A forecast is a promise the business makes to itself. When it is wrong repeatedly, trust decays in both directions: executives stop believing the number, and sellers stop believing the exercise matters. The fix is not a better tool. It is a different set of inputs.

Why commit/best-case/worst-case is broken

The traditional forecast is a three-bucket system. Reps pick a commit number, a best-case number, and a worst-case number for each deal, and the aggregate rolls up to a team forecast. In theory this captures probability distribution. In practice, it captures three cognitive biases.

The first bias is anchoring. Whatever a rep said last week becomes the psychological anchor for this week, even if the deal has changed materially. The second is loss aversion. Moving a deal out of commit is harder than keeping it in, so deals linger in commit long after they should have moved. The third is optimism on new pipeline. Deals enter the forecast at inflated probabilities because sellers see the upside more clearly than the downside.

You can try to fix this with coaching. It does not work. The biases are built into how humans estimate probability, especially when their compensation is at stake. What works is changing the inputs.

Stage exit criteria

The first input change is stage exit criteria. Every stage in the pipeline should have a small, unambiguous list of things that must be true, verifiable, and documented before the deal advances. 'Champion engaged' is not a stage exit criterion. 'Champion has introduced us to the economic buyer within the last fourteen days, with the meeting logged and a confirmed agenda' is a stage exit criterion.

We typically recommend four criteria per stage. More than that is hard to enforce. Fewer than that leaves too much to interpretation. Each criterion should pass four tests:

1. Is it verifiable by someone other than the owning rep?
2. Does the criterion correlate with downstream closed-won in your own historical data?
3. Can the criterion be captured as a structured field in the CRM?
4. Is the criterion owned by a specific role at a specific moment in the deal?

Criteria that fail these tests become performance theater. Criteria that pass them become the single most reliable upgrade you can make to your forecasting process. The first time a deal cannot advance because an exit criterion is missing, the culture shifts.

Deal health scoring

On top of stage exit criteria, we layer deal health scoring. A health score aggregates eleven factors about the deal that, together, predict close-won materially better than commit/best-case ever did. The factors are:

• Champion engagement: number of meetings with the champion in the last thirty days.
• Multi-threading depth: number of distinct contacts at the buyer who have been meaningfully engaged.
• Economic buyer confirmation: has the EB been met, or at least confirmed by the champion in writing.
• Mutual action plan progress: percentage of MAP milestones on or ahead of schedule.
• Competitive position: known primary competitor and current standing against them.
• Business case strength: quantified, signed-off value proposition specific to this buyer.
• Technical validation status: proof of concept, security review, and architecture sign-off.
• Procurement status: engaged or not, known timeline, known process.
• Legal and MSA status: any red-flag clauses surfaced, any known delays.
• Stage age relative to historical average: is the deal moving or stalled.
• Last meaningful activity: days since the last outbound or inbound deal-advancing activity.

Each factor is scored on a three-point scale (healthy, warning, risk). The aggregate is a composite health score. The weighting is calibrated quarterly against your actual close-won data. Do not borrow weights from a benchmark. Your business is specific enough that the weights matter.

The inspection cadence

Inputs are not enough. You also need a cadence that uses them. The three inspection rituals we recommend are:

Weekly deal review

Run by frontline managers with each rep. Focus is the top ten deals by amount or close date proximity. Every deal gets inspected against its current stage exit criteria and its health score. Any deal that fails either is assigned a recovery action with a named owner and a due date before the next review.

Monthly forecast review

Run by the head of sales with frontline managers. Focus is the aggregate: are we ahead, on, or behind plan; which segments are driving the variance; and which deals are swing factors for the quarter. The outputs are management actions, not deal actions.

Quarterly calibration

Run by RevOps in partnership with sales leadership. The calibration looks at the past quarter's actual versus forecast at multiple time horizons (90, 60, 30 days out). Where forecast missed reality, we look for patterns: specific segments, specific stages, specific managers, specific times of quarter. The output is adjustments to stage exit criteria, health score weights, or process.

Rollup and commitment

The commit number that goes to the CFO is not an average of rep commits. It is a modeled number based on the health score distribution of the pipeline, stage conversion rates, and close-date confidence. Reps still commit individual deals, but the commit is an input to the model, not the output.

In practice this usually means a modest discount on commits and best-case, based on the health score of the underlying deals. A deal that is in commit but has a poor health score gets discounted. A deal in best-case with a strong health score gets weighted up. The model tightens the forecast against reality, and the variance between forecast and actual drops materially within two quarters.

Tooling notes

You can build most of this in your existing CRM with disciplined field design and a BI layer. You do not need a dedicated forecasting tool in most cases. Where a dedicated tool helps is in automating health score computation and in letting managers drill in visually without building dashboards from scratch. The tools are fine. The process discipline is the hard part, and no tool will produce the discipline for you.