Designing forecast governance and pipeline coverage that trigger action
Moving the forecast out of a spreadsheet and making coverage something a manager acts on, not admires.
The case in brief
Current reality
The forecast is rebuilt every Monday in a spreadsheet, overrides are invisible, and coverage is reported without anyone being asked to generate pipeline.
What must become true
A forecast built in the system from stage semantics, with recorded judgement, a named owner of the final number, weekly snapshots and measured accuracy—and a coverage contract that starts pipeline generation below its threshold.
Design question
Whose number is the forecast, how is judgement recorded—and what does low coverage oblige someone to do?
01The reality · current state
How the business is steered today.
Sellers keep opportunities in the CRM, but managers export them every Monday, adjust amounts and dates in a spreadsheet and send a regional total upwards. Probability is typed by hand. Coverage is reported on a dashboard, weighted in one region and unweighted in another, against a target nobody can name. Forecast accuracy is discussed only when a quarter is missed.
- The forecast lives in spreadsheets; the CRM is updated afterwards, if at all
- Manager overrides are invisible and cannot be explained later
- Probability is typed by sellers and varies by optimism
- Coverage is weighted in one region and unweighted in another
- Low coverage produces a red number but no pipeline-generation action
- Stale opportunities stay in commit with close dates in the past
02Operating loop
Each step with what happens, who owns it and how often.
CRMData platformPlanning
- 01Pipeline
Opportunities with stage, close date and amount; stale ones drop out of commit automatically.
- Owner
- Opportunity owners
- Cadence
- Daily
- 02Seller call
Category derived from the stage; the seller can move a deal between best case and commit, with a reason.
- Owner
- Opportunity owner
- Cadence
- Weekly
- 03Manager judgement
Adjustments at the forecast level, never on the record—who, what, from, to, why.
- Owner
- Sales manager
- Cadence
- Weekly
- 04Final number
Submitted by the unit leader from the same view; snapshot taken.
- Owner
- Sales leader
- Cadence
- Weekly
- 05Accuracy
Submitted forecast against actual at a fixed lag, by unit and motion.
- Owner
- Sales operations
- Cadence
- Quarterly
The outcome of the last step changes the first one next period.
03Key dimension · Forecast ownership & coverage contract
Whose number, at which layer, recorded how
Each layer adds judgement without erasing the layer below it. Categories are defined once and derived from stage semantics.
- PipelineOpportunity owner
Stage, close date and amount kept current; stale deals leave commit
Recorded asField history - Seller callOpportunity owner
Category derived from stage; moves between best case and commit need a reason
Recorded asWeekly snapshot - Manager adjustmentSales manager
Overrides at the forecast level, never on the opportunity
Recorded asWho, when, from, to, why - Submitted forecastSales leader
One number per unit and period, from the same view
Recorded asWeekly snapshot - AccuracySales operations
Compared with actuals at a fixed lag, by unit and motion
Recorded asQuarterly review
Forecast categories, defined once
| Category | Meaning | Evidence |
|---|---|---|
| Pipeline | Qualified, not yet likely this period | Qualification exit criteria met |
| Best case | Could close this period if things go well | Solution fit confirmed; proposal in progress |
| Commit | Will close this period; the owner stakes their call on it | Negotiation; terms and date agreed in principle |
| Closed | Won in the period | Order or contract reference |
The forecast improves when judgement is recorded, not when it is removed. What cannot be explained afterwards cannot be learned from.
04KPI contracts
Every measure with an owner, a cadence and an action.
| Metric | Definition | Owner | Cadence | Action |
|---|---|---|---|---|
| Pipeline coverage | Eligible open pipeline (qualified, in horizon, unweighted) ÷ remaining signed target | Sales management | Daily; reviewed weekly | Pipeline generation below 3× |
| Forecast accuracy | |Submitted − actual| ÷ actual, at a four-week lag, by unit and motion | Sales operations | Quarterly | Review categories and overrides where accuracy drifts |
| Commit slippage | Commit value whose close date moved out of the period | Sales managers | Weekly | Requalify, or move out of commit with a reason |
| Stale share of commit | Commit value past its ageing benchmark or with a past close date | Sales managers | Weekly | Owner updates or removes the deal |
05Accountability
Who owns what, decides what—and escalates to whom.
| Role | Owns | Decides | Escalates to |
|---|---|---|---|
| Opportunity owner | The deal’s stage, date, amount and category call | Best case versus commit, with a reason | Sales manager |
| Sales manager | The team forecast and its adjustments | Overrides, recorded at forecast level | Sales leader |
| Sales leader | The submitted number for the unit | What is submitted each week | Sales leadership |
| Sales operations | Definitions, snapshots and accuracy measurement | Changes to categories and coverage contract, via governance | Governance forum |
Design decision
Derive forecast categories from stages, record judgement as overrides at the forecast level, and govern coverage as a metric contract with one definition, one threshold and one required action.
06The trade-offs
Credible options, judged against these premises.
Spreadsheet forecast rebuilt weekly
Very small teams
Cost: Invisible judgement; the CRM becomes optionalFully automated forecast from probabilities
High-volume, homogeneous deals with long history
Cost: Removes accountability; hides what managers knowStage-derived categories with recorded overrides and snapshots
Mixed motions and managers who add real information
Cost: Snapshot storage and review discipline07The second layer
Questions that change the design.
Forecast
- Stage-based or probability-based?
- Who owns the final number?
- How are changes logged?
Coverage
- Which opportunities are eligible?
- Weighted or unweighted?
- Which target, over which horizon?
Action
- What must happen when coverage is low?
- What happens to stale opportunities?
- How is accuracy used—and by whom?
08Decisions & outputs
What the work produces.
- 01Forecast category model
- 02Override and snapshot rules
- 03Forecast ownership matrix
- 04Coverage metric contract
- 05Pipeline-generation trigger
- 06Accuracy review