01 / BUSINESS DECISION
Start with the decision.
Set a realistic revenue outlook and identify where pipeline assumptions need review, using deal state as it was known on each forecast date.
02 / PRACTICAL IMPLICATION
What the evidence can support.
Forecast ranges could help commercial leaders separate changes in pipeline quality from timing risk and avoid treating weighted pipeline as committed revenue. No forecast accuracy or commercial improvement has been demonstrated yet.
03 / EVIDENCE
A defined question. Work still ahead.
Study design only. No dataset, fitted model, evaluated demonstration or repository has been published for this direction.
Planned evaluation
- Rolling-origin evaluation with point-in-time deal snapshots and untouched final time holdout.
- Compare stage-weighted pipeline and historical stage/cohort conversion with survival or discrete-time hazard models.
- Evaluate aggregate revenue error, conversion calibration, interval coverage and timing bias across forecast horizons.
No evaluated result, client impact or live model is claimed for this planned study.
04 / DATA
Know where the evidence comes from.
A documented synthetic deal-event process is planned. Any real CRM data would require permission, point-in-time snapshots and a clear policy for deleted or revised opportunities.
05 / METHODOLOGY
Match the method to the design.
Start with stage-level conversion and timing baselines. Evaluate separate deal-conversion and time-to-close components, then aggregate a predictive revenue distribution with explicit dependence assumptions. Split by forecast date rather than random deal rows.
06 / LIMITATIONS
Conditions that matter.
- Features must have existed at the forecast cutoff; later stage changes and closing dates are not inputs.
- Open deals are censored, not automatically lost.
- Training, validation and test periods are separated; repeated snapshots of one deal must not create leakage.
Unstable sales processes, correlated deals and reporting changes can defeat historical calibration. This page specifies a future study, not an implemented forecasting system.
07 / CODE & NEXT STEPS
The next useful piece of work.
Implementation has not begun. A repository and source download will be linked when they exist and contain reviewable work.