Your design system doesn’t need more components. It needs better evidence.

TAP Mindset had eight card variations, six chips and four inputs. Most looked related but behaved differently. Consolidating them sounded like the obvious answer.

The deeper problem was maintenance. Knowledge lived in individual files and documenting a component took 45 to 60 minutes. Every new feature made the library harder to understand.

I mapped the critical flows and inspected more than 28,000 component instances. I was not looking for the most popular rectangle. I was looking for repeated product decisions.

1. Load-bearing patterns used across critical journeys.
2. Accidental variants representing the same concept.
3. Local exceptions created for a constraint that had never been evaluated systemically.

For each cluster I recorded its job, states, usage, behavioural differences and cost of change. That changed the review from ‘which version looks cleaner?’ to ‘does the product need these concepts to behave differently?’

The useful output was not a giant spreadsheet. It was a smaller set of decisions with enough evidence for Design and Engineering to challenge together.

Start with one important journey, capture concepts rather than screenshots and record uncertainty explicitly. Do not promote a pattern into the system just because it appears often: repetition can be evidence of debt too.

The goal is not perfect consistency. It is a system where a shared decision is easier to use than another local exception.