Grading a design system against six dimensions

Nielsen Norman Group retired the design system maturity ladder in July 2026 and replaced it with six dimensions, each scored one to five.

The argument behind the change is that systems do not mature in a straight line. A mature enterprise system can be excellent on components and poor on governance, and a single overall score hides exactly the gap a team needs to see. A merger that introduces a second set of tokens can put an otherwise advanced system back at the foundations on one dimension while the rest stay where they were.

Huei-Hsin Wang, who wrote the framework, names the failure mode directly: teams end up performing maturity rather than practicing it. A single number is what makes that possible, because a single number is reportable.

Scoring the dimensions separately

The six dimensions cover the parts of a system that can advance independently. Foundations and tokens. Components. Documentation. Governance and contribution. Tooling and adoption. Team and resourcing.

Most enterprise systems score high on components and documentation, because those are the visible artifacts and the ones a team is asked to show. The scores that decide whether the system actually holds are governance and adoption, and those are the ones that rarely get measured, because neither produces a screenshot.

A system with excellent components and no contribution process becomes a bottleneck. Product teams needing a variant either wait or build their own, and the second option is what produces the drift the system was created to prevent.

Running the assessment honestly

The framework comes with a self-assessment template, and the discipline it requires is that the scoring is done with evidence rather than with opinion. A governance score of four means there is a written contribution process, someone owns review, and a variant proposed last quarter actually moved through it. If the process exists on a wiki page and nothing has gone through it in six months, the honest score is lower.

Running the assessment with the product teams who consume the system, rather than only with the team who maintains it, changes the result more than any other choice. Maintainers score documentation on whether it exists. Consumers score it on whether it answered the question they had at the moment they had it.

A single score is easy to report upward. Six of them tend to start an argument about which one to fix first, which is the more useful conversation.

BrilliantUX editorial principle

Turning scores into a quarter of work

The output that matters is not the profile. It is the decision about which one dimension to move next, and the framework is useful mostly because it forces that decision to be explicit.

The dimension worth moving is usually the lowest one that is currently blocking the others. A team at four on components and two on tooling will get more from the tooling work, because better components that nobody can install do not reach a user. A team at four on tooling and two on governance will get more from governance, because adoption without a contribution path produces forks.

Re-scoring at the end of the quarter with the same people is what turns this from an audit into a practice. One dimension moving from two to three is a real result. All six inching upward usually means the scoring got generous rather than the system getting better.