wardley-maps.sgit.ai / Method / The custom-axis verdict
The custom-axis verdict
Original contribution 28 July 2026 W7 · glossary
the evolution scale doesn't necessarily need to be Genesis, custom-built, product and commodity; you can go from air gap to file to API to event-driven.
Relabelling the evolution axis is one of the most common things people try in their second month of mapping, and one of the fastest ways to produce something that looks like a Wardley map and reasons like a spreadsheet. This is the sharpest answer in the corpus to when you may do it, and it fits in one sentence.
Relabel the axis when the thing genuinely evolves; use a maturity model when the thing merely improves.
What the rule is actually testing
Wardley's four stages are not four adjectives for "how good is it". They name a process: something appears, gets built badly by people who need it, gets productised by people who see a market, and finally becomes so ubiquitous and so well-understood that it is bought without thought. The axis measures ubiquity against certainty. It is not a time axis and not an adoption curve.
So the test the rule proposes is: do your labels name stages of that same process, or do they name rungs on a ladder of quality?
| Proposed axis | Verdict | Why |
|---|---|---|
air gap → file → API → event-driven |
A relabelled evolution axis | Each stage is a genuinely different kind of thing, with its own economics and its own failure modes. Moving right is not "the same integration, better" — it is a different integration that became possible because the one to its left became ubiquitous. That is evolution. |
no tests → some tests → good coverage → full coverage |
A maturity model | Same activity, more of it. Nothing new becomes possible at the right-hand end that was inconceivable at the left. Draw it as a maturity model and stop calling it a map. |
not-visible → fully-visible (the visibility overlay) |
Open | Proposed in the corpus as an alternative axis. It is a property of a component relative to a user rather than a stage of its evolution — which is, notably, what the other axis of a Wardley map already measures. Unresolved: see below. |
risk: not-understood → fully-accepted |
Open | Same question. There is a real argument that risk understanding does evolve — from nobody having seen this failure to it being a line item in a standard control set — and an equally real argument that this is a maturity model wearing a map's clothes. |
The six axes that were proposed and never drawn
The corpus proposes alternative axes in two places — "by treating the evolution axis as a configurable dimension rather than a fixed one, and by storing axis metadata in the graph alongside everything else" (7 February 2026), and again as visibility and risk overlays (19 June 2026). Six were named: openness, automation, documentation, test coverage, graph connectivity, plus the visibility and risk overlays.
air gap → file → API → event-driven — was ever drawn. Six proposed
axes, one explicit ruling in their favour, and no artefact. Applying the rule above to the six
proposals is uncomfortable, too: test coverage and documentation look like
maturity models by the verdict's own test. That is the verdict working correctly, and it is
worth saying rather than quietly not mentioning.What changes if you accept it
Two things, and the second is the useful one.
You stop arguing about whether custom axes are allowed and start arguing about whether a specific one passes the test — which is a much better argument, because it is about the domain rather than about the method.
You get a reason to reject most of them. The rule is not permissive. Applied honestly it rejects the majority of relabelled axes people propose, including several in the corpus that proposed it. A rule that only ever says yes is not doing any work.
What would falsify it
- A relabelled axis that passes the test and still produces nonsense. If genuine evolution is not sufficient for the map to behave — if, say, the climatic patterns stop holding on a relabelled axis even when the stages are real — the rule is necessary but not sufficient and needs a second clause.
- A maturity model that behaves like a map anyway. If you can take
no tests → full coverage, place components on it, and find that anticipation and gameplay still work, then the distinction is aesthetic and the rule should be dropped. - Wardley's own position. This has not been put to him, and the question of when the axis may be relabelled is old enough that there may be a published answer this corpus has not found. If there is, it wins on priority.
Left open
Q2: does the rule hold for the risk and visibility overlays? The corpus proposes both and the verdict does not adjudicate them. This page has argued above that visibility is the weaker of the two, since a Wardley map already has a visibility axis — but that is an argument, not a resolution.
Related
- You cannot map a gap — the same brief, 28 July 2026, and the map that provoked both.
- De-commoditisation — pushing against what the stages mean rather than what they are called.
- Matt Edgar's critique — if the axis encodes consensus rather than fact, the question of what you may relabel it to gets considerably more awkward.