Why Three Maturity Levels Beat Five

Pick up almost any maturity model and you will find five levels. The pattern goes back to the Capability Maturity Model of the late 1980s and early 1990s, which used five levels from "Initial" to "Optimizing". Since then, five has become something close to a convention. Many models use it without asking whether it suits their purpose.

For formal appraisals carried out by trained assessors, five levels can work well. For self-assessment, where people judge their own organisation with no specialist training, there is a strong case for fewer, clearer levels. Every model on this site uses three. This article explains why, and when five might still be the better choice.

The problem with fine distinctions

A maturity level is only useful if people can tell it apart from its neighbours. With five levels, the descriptions of adjacent levels often differ by a word or two: "documented" versus "standardised", "managed" versus "measured", "consistent" versus "systematic". On paper, the distinctions make sense. In practice, when someone reads them quickly while thinking about their own team, they blur.

The result is predictable. People hesitate between neighbouring levels and settle on whichever feels safest. That is often the middle one.

The middle option trap

When people are unsure, they tend to choose the middle of a scale. It feels moderate and defensible: neither boastful nor self-critical. With five levels, level 3 becomes a comfortable default, and answers cluster there regardless of reality.

This makes results look reasonable while telling you very little. A profile where almost everything is around level 3 suggests an organisation that is broadly fine and consistent, when the truth may be a mix of real strengths and serious gaps that have been averaged into blandness.

Three levels do not remove the middle option, but they make it a genuine choice. When the three descriptions are clearly different, people have to decide which one actually describes them.

Three levels map to three real states

Across most capabilities, organisations tend to be in one of three recognisable states:

1. Informal and person-dependent. Things get done, sometimes well, but it depends on who is involved. There is no consistent way of working, and success is hard to repeat.
2. Defined and consistent. There is an agreed approach, it is followed most of the time, and results are reasonably predictable.
3. Embedded and improving. The approach is part of how the organisation works. People follow it without being told, it is measured, and it gets better over time.

These three states are easy to recognise because most people have experienced all of them. Each is clearly different from the others, both in how work feels and in the results it produces.

Five-level models usually split these states further, for example separating "defined" from "measured", or "managed" from "optimising". Those distinctions are real, but they are much harder to judge from the inside.

Concrete descriptions matter more than the number

The number of levels is only half the story. The other half is how each level is described. A three-level model with vague descriptions is no better than a five-level one.

Good level descriptions are:

• Observable. "A named person reviews the risk register every month" rather than "risk management is mature".
• Specific to the question. Each question has its own three descriptions, written for that topic, rather than a generic scale applied to everything.
• Clearly different. Someone reading all three should immediately see which one describes their situation.

This is why each question on this site has its own three answer descriptions, rather than asking people to rate themselves on a generic scale from 1 to 5. Our methodology sets out the full design principles.

Fewer levels, better conversations

There is also a practical benefit. Three levels make results easier to discuss.

• The next step is obvious. If you are at level 1, the goal is level 2. The description of level 2 tells you what that looks like.
• Disagreements are meaningful. If one person answers level 1 and another answers level 3, that is a real difference of view worth exploring, not a matter of shades.
• Results are easier to explain. Saying "we are mostly at the structured level, with two areas still informal" is something any leadership team can understand.

When five levels make sense

Three levels are not always right. Five can be the better choice when:

• Trained assessors carry out the assessment, with detailed evidence requirements for each level, as in formal certification or appraisal schemes.
• You need to show fine-grained progress over many years in a single area that is already fairly mature.
• You are benchmarking against an external standard that is itself defined in five levels.

For most internal self-assessments, where the aim is to get an honest picture quickly and decide what to do next, three levels give clearer answers.

What about scores between levels?

With three levels per question and several questions per area, area scores naturally fall between whole numbers, for example 1.75 or 2.25. This gives you the finer gradation of a longer scale without asking anyone to make fine judgements on individual questions. A score of 2.25 tells you that an area is mostly at the second level, with some practices already at the third.

Frequently asked questions

Isn't a three-level model too simple for complex capabilities?

The complexity lives in the questions, not the levels. A model with 24 to 28 questions, each with its own concrete descriptions, captures a great deal of detail while keeping each individual judgement simple.

Can I compare a three-level score with a five-level score?

Not directly. The levels describe different things, so converting between them gives a false sense of precision. Compare like with like, and use the descriptions rather than the numbers when talking across models.

Why not use a simple yes or no checklist instead?

Checklists tell you whether a practice exists, but not how well it works. Three levels capture the difference between having something on paper, doing it consistently and doing it well, which is usually where the real improvement lies.