The method

1. Definitions

Each term below has one meaning in this specification and is used no other way. Where a word is in common use with a looser meaning, the definition here is the one that applies.

Terms deliberately not used. This is not described as a rating, a ranking, a credit score, a benchmark position or a prediction. Each implies something the method does not do. See Limits.

4. The arithmetic

Reproducible by hand. Nothing is weighted by sector, company size or anything else, and no generative model touches the scoring path. The same answers return the same score today and in five years.

// every step, in order
cell        = mean(its 2 questions)                  // 1 to 5
stage       = mean(the 4 cells in that row)          // 1 to 5
rootCause   = mean(that cause across all 4 stages)   // 1 to 5
capability  = mean(all 16 cells) / 5 * 100          // 0 to 100
momentum    = mean(the 5 momentum answers) / 5 * 100
score       = round((capability + momentum) / 2)

Bands

Band	Score	What it means
Forming	0 to 39	The commercial engine is not yet built. Pick one stage and install the basics before widening.
Developing	40 to 59	Parts of it work. What has been built and what is done daily are not yet pulling together.
Established	60 to 79	A working commercial engine. The next step is making it hold at greater volume.
Durable	80 to 100	Works without the founder in the room and survives more volume. Attention shifts to keeping customers.

A 58 and a 61 are functionally the same engine. A few points either side of a band boundary is noise, and the method says so rather than letting the band imply precision it doesn’t have.

6. Capability against momentum

Most assessments collapse these into one number and lose the finding. The gap between them is a diagnosis in itself.

Capability	Momentum	What it usually means
Low	Low	The engine isn’t built and not producing. Pick one stage and install the basics.
Low	High	Results are coming from founder effort. Capacity-limited: when the hustle stops, so does the output.
High	Low	The engine is built and nobody is running it. Usually about who inspects the pipeline and how often, not about capability.
High	High	Built and producing. Focus shifts to expansion and durability.

8. Motions and maturity

Startups are not all software companies, and a measure that assumes they are will produce confident nonsense for most of them. A direct-to-consumer brand and a deeptech hardware business do not have the same commercial engine, and scoring them against identical questions is a category error.

So the instrument is motion aware. The construct never changes. The evidence used to populate it does.

Fixed for every company

• The four stages and four root causes
• The sixteen cell structure, two questions per cell
• Both answer scales, the arithmetic and the bands
• The diagnosis logic

Varies by motion

• The wording of the questions and the indicator each one asks about
• The stage labels a respondent sees
• The worked examples in the report

A subscription software founder is asked about qualification and pipeline progression. A DTC founder is asked about storefront conversion, contribution margin and repeat rate. Both produce a comparable position on the same scale, because they are answering the same question about their own engine in the language of their own business.

What each stage means, by motion

Motions

A motion earns its own question set only when its commercial engine is genuinely different. Adding one for the sake of coverage would fragment the data for no gain.

Maturity

Each motion runs at three levels, because the same stage asks a different question depending on how far along the company is.

What this costs us, stated plainly. Peer comparison is only fair within a comparable group, and a comparable group is now a motion at a maturity level. That multiplies the number of segments and pushes any published peer benchmark much further out. We would rather say so than publish a percentile drawn from a group that is not really a group. See the version history for the threshold.

9. Machine-readable artefacts

The specification is published as data, not only prose, so a third party can implement against it.

Artefact	Contains	Status
/method/v1.4/scales.json	Both answer scales, values, labels and meanings	Published
/method/v1.4/bands.json	Band thresholds and definitions	Published
/method/v1.4/model.json	Stages, root causes, the cell map, the arithmetic	Published
/method/v1.4/instrument.json	The full question set	release pending
/method/v1.4/vectors.json	Conformance test vectors	release pending

// scales.json, abridged
{
  "version": "v1.4",
  "capability": {
    "instruction": "Score how true this is in practice today. Use evidence and consistency, not intent.",
    "options": [
      { "score": 1, "label": "Not in place" },
      { "score": 2, "label": "Inconsistent" },
      { "score": 3, "label": "Partly in place" },
      { "score": 4, "label": "Consistent" },
      { "score": 5, "label": "Proven and improving" }
    ]
  }
}