Reference implementation · v1.4
Run the method yourself.
Set a value for each of the sixteen cells and the five momentum readings. Everything below updates as you go: the score, the band, the weakest stage, the local and systemic root causes, and the capability against momentum read. Nothing is sent anywhere and nothing is stored. Close the tab and it's gone.
This is the arithmetic, not the questionnaire. The real instrument asks 37 questions and derives these values from the answers. Those questions vary by business model and maturity, but the grid, the scales and the arithmetic on this page never do, which is what keeps two companies comparable. It's here so you can confirm the published rules produce the published outputs, before you ask anyone to complete anything.
Capability · 16 cells
Score how true each is in practice today. Use evidence and consistency, not intent.
Momentum · 5 readings
Is the engine producing output now?
The problem lens shown is the documented default mapping. The full rule precedence, its thresholds and the written diagnosis for each cell are part of the licensed layer and aren’t published.
What this page proves
- It’s deterministic. Set the same values twice and you get the same result. There's no model in the scoring path and there never will be.
- It’s reproducible by hand. Every number here comes from the six lines of arithmetic published in the method. Nothing is hidden between the inputs and the output.
- Nothing is weighted by sector or size. The same rules apply to every company, which is the only reason two results can be compared.
- The judgement isn’t in the maths. Load the worked example. Expand scores lowest, yet it isn't what the diagnosis sends you to fix. That precedence is the part you license.