Managing the number
The number that reaches the board is the output of a negotiation.
A rep decides what to say. A manager decides how much to believe. Somebody decides what to report. Each layer adjusts the last, nothing records what was adjusted, and by the third week of the quarter nobody can reconstruct who was right.
Keeping the three numbers separate is most of the fix.
Three numbers
Judgement, sum, and evidence — none of them averaged into the others.
What the reps submitted, added up. What their manager will actually commit to. And what the recorded calls suggest.
They are three separate figures on one screen. A product that blends them into one has quietly removed the only conversation a forecast review is for.
- What the team submittedthe mechanical sum of six reps£1.84m
- What the manager commitstheir judgement, submitted and kept£1.62m
- What the calls suggest5 of 6 reps' deals have call evidence£1.41m
- Categories roll up cumulatively —
Closed within commit, commit within best case, best case within pipeline — one ladder to read rather than four numbers to reconcile. And the ordering is guaranteed at the point of storage, so an impossible roll-up cannot be saved at all.
- A submission is a new record, not an edit —
Every revision through the quarter is kept, each pointing at the one it supersedes, with exactly one current. The history of what somebody said in week two is the coaching material.
- The submission freezes the working that justified it —
The full breakdown as it stood at that moment is stored with it — so editing a deal next week cannot retroactively reshape what was submitted.
- A deal lands in the period it actually closed —
By its real close date when it has one, and its expected date when it does not.
Evidence
Each deal weighted by what its calls sounded like, and the coverage disclosed.
A deal whose objections went unanswered, whose sentiment fell across the last three calls, or whose competitor was weighed twice, is worth less than its probability says. So the projection weights it down.
A deal with no calls is weighted exactly as it stands. It is not marked down for a silence that might only mean the rep sells by email. And the projection reports how many deals had evidence and how many did not, so nobody reads more into the number than is in it.
Where the rep's own commit exceeds the projection by a wide margin, a note says so. Where it falls below, nothing is said — under-commitment is a self-correcting problem and flagging it teaches people to inflate.
- Currency conversion refuses to guess —
A deal whose currency pair has no rate you have recorded is excluded from the roll-up and counted, with the count shown and a link to add the rate. Most products silently add mixed currencies at one to one.
- The roll-up walks your reporting lines —
With every level carrying its own subtotal, so a manager reads a subtree's number without re-adding it.
- Nightly snapshots make a waterfall possible —
What was added, won, lost, expanded, contracted — and, separately, pushed out of the period. That last one is the number most tools fold into losses, and it is a completely different conversation.
›Does the forecast follow our fiscal year?
Periods follow the calendar quarter today. If your financial year is offset, the roll-up will not line up with your board pack — worth knowing before it is relied on for reporting.
›What if a rep never submits?
Their live roll-up still reads, but the snapshot history that feeds the waterfall only accumulates for people who submit. The waterfall will be empty for them.
›Are there forecast reminders?
No. Nothing in this area notifies anybody — not a submission reminder, not an attainment alert. It is a pull surface by design, and that is a real limit if your process depends on nudges.
›Can I forecast by product or by territory?
By person and by reporting line. Not by product line, and territory is not a roll-up axis.