Project Forecast Engine

Stop reporting one number you can't defend

"We forecast £46M" invites an argument. "We are 80% confident of coming in under £49M" ends one. The Forecast Engine produces P50 and P80 outturn and completion dates from the progress you have already earned. In Excel, with no macros and nothing to install.

Home › Forecasting

Six forecasts, one data entry

Enter progress once. Every forecast below recalculates from the same dataset, so your cost forecast and your date forecast can never disagree with each other.

1

Cost at completion

Three EAC methods side by side, because clients and auditors disagree about which one applies. Seeing all three is how you defend the one you pick.

EAC₁ = BAC / CPI
EAC₂ = AC + (BAC − EV) / (CPI × SPI)
EAC₃ = AC + (BAC − EV)
2

Completion date

Earned Schedule, not the naive SPI method. SPI always drifts back to 1.0 as a project finishes, so a late project reports as on time exactly when it matters most.

ES = t + (EV − PV_t) / (PV_t+1 − PV_t)
IEAC(t) = PD / SPI(t)
3

Cash flow

Monthly draw curve and cumulative S-curve, giving the peak funding requirement. The number that determines whether a profitable project still sinks you.

Peak funding = max(cumulative AC − cumulative receipts)
4

Resource forecast

Histogram by period with peak demand and over-allocation flags, so you find out you need 40 fitters in March while you can still recruit them.

Demand_t = Σ (remaining hours × allocation %)
5

Risk-adjusted range

Monte Carlo over 1,000 iterations producing P50, P80 and P90. This is the difference between "we forecast £46M" and "we are 80% confident of coming in under £49M".

P80 = PERCENTILE.INC(iterations, 0.8)
6

Productivity trend

Units per man-hour tracked over time and extrapolated forward, so the forecast reflects the rate you are actually achieving rather than the one you assumed.

Forecast hours = remaining units / trended rate

Why P80 changes the conversation

The same project, forecast two ways

A £42.0M construction project at month 9, running CPI 0.91 with three trades behind on productivity and an unagreed variation package. These are the figures the Forecast Engine produces from the sample loaded in the workbook.

CPI
0.91
Over budget
SPI
0.94
Behind plan
EAC
£46.2M
BAC ÷ CPI
VAC
−£4.2M
Against £42.0M BAC
TCPI
1.08
Needed to still hit BAC
Complete
42%
EV ÷ BAC
£0M£10M£20M£30M£40M£50M048121620 data date P50 £46.8M P80 £48.9M BAC £42.0M period
Planned value Earned value Actual cost Forecast range, P50 to P80
EAC£46.25M P50£46.83M P80£48.87M P90£50.02M £2.04M contingency gap
Single-point EAC
£46.2M
P50
£46.8M
P80
£48.9M
Contingency gap
£2.0M
The £46.2M EAC is not wrong, just incomplete: the middle of a distribution with no width. The £2.0M between P50 and P80 is the contingency conversation, and one number cannot have it.

Eight industry editions

The engine is identical in every pack. The pack supplies the WBS, cost codes, risk register and milestone set for your sector. So it is useful the first time you open it, not after a fortnight of setup.

NuclearMulti-decade, regulator-gated ConstructionNEC4 valuation cycles Data CentresCost per MW, long-leads AirportsNight possessions, live ops OffshoreVessel rates, weather windows RailPossessions, GRIP/PACE stages Power T&DOutages, connections, consents Energy GenerationEscalation, grid dependency

Forecast Engine pricing

Engine only

£299
One-time purchase
  • All six forecast outputs
  • Monte Carlo P50 / P80 / P90
  • Generic WBS and cost codes
  • Free updates

All eight editions

£999
Best value
  • Everything in Engine only
  • All eight industry packs
  • Sector WBS, rates and risks
  • Regulatory milestone sets
  • Priority updates

Single industry edition

£349
Complete, standalone
  • One industry of your choice
  • Sector WBS & cost codes
  • Preloaded risk register
  • Regulatory milestones

Forecasting questions

Because the forecasting maths does not change between a reactor and a runway. EAC is BAC/CPI everywhere. What changes is the WBS, the cost codes, the risks and the milestones, and that is content, not engineering. One engine means one set of formulas to trust and to fix.
No. It is a plain .xlsx using native formulas only. Macro-enabled workbooks are stripped by most corporate mail filters and blocked by default in Excel, which would make the file undeliverable.
Each uncertain input is sampled by inverse-transform from a triangular distribution across 1,000 pre-built iteration rows, then PERCENTILE.INC reads the P50, P80 and P90 off the results.
No, the Forecast Engine is standalone. They complement each other. The main tool tracks where you are, the engine forecasts where you will end up.
An industry edition is a complete, standalone Forecast Engine with that sector's WBS, cost codes, risk register and regulatory milestones already loaded, plus a driver sheet modelling what actually governs cost in that sector. It is not an add-on to the generic Engine and does not require it. That is why an edition costs more than the generic version, not less.
The packs ship with placeholder ranges and a clear instruction to replace them with your own historic data. Your rates always beat a published average, and the tool is built to make swapping them easy.

Forecast with a range, not a guess

Six forecasts, eight industries, one workbook.