Cockpit › Scenario lab

What would have to be true?

Move the drivers and watch the quarter-end cash move with them. Forecasting isn’t a single number — it’s a number per set of assumptions.

Drivers

restart_alt Reset to base

Pipeline win rate last 90 days: 38%

38%

Average project size trailing: €88K

€88K

Retainer churn annualised: 6%

Collection lag current: 48 days

48d

New hires this quarter cash out before they bill

Cash at quarter-end · 30 Sep

base

2.58M €

P10–P90 · 2.04M – 3.05M

1.4M3.6M

target

Chance of hitting plan

plan target · €2.42M by 30 Sep

72%

The base case clears plan with room. The single biggest swing factor is win rate — eight points either way moves quarter-end cash by roughly €145K.

Saved scenarios

click one to load its drivers

trending_down Conservative

2.16M €

18% chance of hitting plan

Win rate 29%
Avg project €80K
Churn 11%
Collection lag 56d
New hires 1

check_circle Base

2.58M €

72% chance of hitting plan

Win rate 38%
Avg project €88K
Churn 6%
Collection lag 48d
New hires 0

rocket_launch Stretch

2.90M €

94% chance of hitting plan

Win rate 47%
Avg project €98K
Churn 4%
Collection lag 42d
New hires 2

Backtest — was the model right?

Dec 2025 – May 2026 · revenue

Dec ’25

−3.2%

Jan ’26

−4.0%

Feb ’26

+11.0%

Mar ’26

−2.8%

Apr ’26

−4.9%

May ’26

−2.9%

Forecast P10–P90 Actual, in band Actual, out of band

MAPE

4.8%

Inside the band

5 / 6

Bias

−0.9%

February’s actual broke through the top of the band — a campaign closed three weeks earlier than the pipeline implied. The model ran slightly conservative (negative bias): when it’s wrong, it tends to under-promise. We’ll take that direction of error over the alternative.

A forecast that’s never scored is just confidence with a chart attached. This panel is the part most tools quietly omit.

The board doesn’t need one number it can’t question. It needs to see which assumption the number is standing on — and what happens to the cash when that assumption moves.