Sales forecasting is treated as an art and is largely arithmetic applied to two measurements that most organisations track badly.
The two numbers
Conversion rate between stages, and time spent in each stage.
Together these determine how much pipeline is needed to produce a given revenue in a given period.
Which is calculable, and the calculation is frequently not done because the underlying measurements are unreliable.
Why stage definitions matter
Stages must be defined by observable buyer behaviour rather than by seller activity.
A stage defined as demo completed is objective. A stage defined as interested is not.
Which means subjective stages produce inconsistent data across a team, and the conversion rates calculated from them are meaningless.
Defining each stage by a specific verifiable event is the fix, and it is unglamorous work that improves everything downstream.
The stagnation problem
Opportunities remain in pipeline long after they are dead, because nobody wants to close them as lost.
Which inflates apparent pipeline and destroys forecast accuracy.
Age-in-stage reporting identifies these, and a policy of closing opportunities that have not moved within a defined period keeps the data honest.
The resistance to this is emotional rather than analytical, and it is universal.
Cycle time
Measured properly, from first qualified contact to closure, and it is generally longer than sellers believe.
Which matters because it determines how far ahead pipeline must be built.
A business with a six-month cycle building pipeline for this quarter is already too late, and this is the most common structural cause of a bad quarter.
Deal size and cycle length
Correlated, generally, since larger purchases involve more people and more scrutiny.
Which means a business moving upmarket experiences lengthening cycles and should expect a period where revenue appears to stall while the longer pipeline builds.
Failing to anticipate that has caused a number of companies to abandon a correct strategy prematurely.
Buying committees
Business purchases above a modest size involve multiple people, and research on this consistently finds committee sizes larger than sellers assume.
Which means a deal with one enthusiastic contact and no other relationships is fragile, and the most common cause of a stalled deal is an unidentified stakeholder.
Mapping who must agree, and what each cares about, is basic and frequently skipped.
Loss reasons
Recorded loss reasons are systematically unreliable, since sellers attribute losses to price far more often than buyers do.
Which means loss data collected from sellers describes seller beliefs rather than buyer decisions.
Win-loss interviews conducted by someone other than the seller produce substantially different and more useful findings, and they are among the highest-value activities available to a sales organisation.
No decision
A large share of losses are to inaction rather than to a competitor.
Which is a different problem requiring a different response — the case for change was not made, rather than the case for this solution.
Treating no-decision losses as competitive losses leads to competing on features against an alternative that was never in play.
What actually improves forecasting
Objective stage definitions. Honest pipeline hygiene. Measuring cycle time properly. And forecasting from historical conversion rates rather than from seller confidence, which is consistently optimistic in every organisation that has measured it.
Qualification
Deciding which opportunities deserve effort, which determines the productivity of everything downstream.
Frameworks exist and their common element is establishing that a real problem exists, that budget is available, that the person can decide or influence, and that there is a reason to act now.
The last is the one most often missing, and opportunities without it stall indefinitely regardless of interest.
Disqualifying early is uncomfortable and it is what creates capacity for the opportunities that can close.
Compensation design
Incentive structures shape behaviour more than any training.
Commission on revenue drives volume. Commission on margin drives discipline about discounting. Accelerators above target drive year-end concentration.
Which means unintended behaviours are generally traceable to the plan rather than to the people.
Plans that change frequently destroy trust, and plans that never change stop reflecting what the business needs.
Handover
The transition from sales to delivery is where expectations set during selling meet reality.
Documented commitments, and involvement of delivery people before signature, reduce the mismatch that produces early churn.
Forecasting categories
Most organisations classify opportunities by confidence — commit, best case, pipeline — which relies on judgement.
Weighting by historical stage conversion rates produces a mechanical forecast that can be compared against the judgement-based one.
Where they diverge consistently in one direction, the judgement is biased and the size of the bias is measurable and correctable.
Territory and coverage
How accounts are allocated determines whether the addressable market is actually being worked.
Which is an arithmetic exercise — accounts per seller, touches per account per year, hours available — and it frequently reveals that coverage is impossible as designed.