Constructing a design system ROI formula

WHAT THIS TESTS This is a leadership-level question testing whether you can translate engineering value into a finance-credible model with explicit variables and real data sources, not hand-waving.

A GOOD ANSWER COVERS Start with the formula: ROI equals net benefit divided by total cost, where net benefit is total benefit minus total cost. Define costs: the design-system team's fully loaded salaries, tooling and infrastructure, and the migration effort imposed on consuming teams. Define benefits in measurable terms: engineering and design hours saved by reusing a component instead of building it (number of reuses times build-time-avoided times blended hourly rate), reduced duplicate implementation, fewer UI and accessibility defects (defects avoided times average fix cost), faster feature delivery, and faster onboarding for new hires. Gather data technically: component adoption and reuse counts from AST usage analysis, build-time-avoided benchmarked from how long building a similar component takes, defect rates from issue trackers before and after, and velocity from delivery metrics. Be conservative with assumptions and show the formula so finance can stress-test it. Present a range, not a single suspiciously precise number.

COMMON WRONG ANSWERS Claiming the system saves a lot with no variables or units. Counting only benefits and ignoring the team and migration costs. Using unsourced or inflated per-component time savings. A single precise figure with no sensitivity analysis.

LIKELY FOLLOW-UPS How do you defend the time-saved-per-reuse assumption? How do you handle attribution when many factors affect velocity? What is a conservative range?

ONE CONCRETE EXAMPLE Usage analysis shows a component reused two hundred times; each reuse avoids roughly eight build hours at a blended rate, yielding a benefit you offset against a four-engineer team and tooling cost. You present ROI as a range with stated assumptions, letting finance probe the reuse count and hourly figures themselves.