NPS 1.7 to 8.23.
A data-backed audit of 500+ screens, then a two-year unification programme across five brands. Drop-offs fell from 87% to 34.7%, and NPS climbed from 1.7 to 8.23.
Yubi grew fast, partly by acquisition, and every product team made sensible choices in isolation. The result for users moving between products was an irrational whole: different navigation, different tables, different language for the same concepts.
The NPS of 1.7 was not mysterious. Drop-offs sat at 87%. Users were confused, stakeholders were frustrated, and engineering was fixing the same UI bug in six products at once.
Project Sherlock was a data-backed audit before a single fix: heatmaps, analytics, session recordings, and automated screenshot mapping across every product. 500+ screens catalogued. Every navigation pattern, filter placement, table layout, and card style named and compared.
Precision was the point. A vague sense that things are inconsistent moves nobody. A named, numbered catalogue of exactly what diverges, and what it costs, moves everyone.
UNIFY was deliberately not a central takeover. I ran it through the design leads of each brand, with a shared review cadence and a governance gate: nothing shipped without being checked against UNIFY standards.
The design language we standardised was written to feel like an evolution of each product’s identity, not an occupation. That framing is why five teams adopted it rather than resisted it. A unified accessibility framework brought WCAG AA compliance to every core flow along the way.
Drop-offs fell from 87% to 34.7%. NPS climbed from 1.7 to 8.23. 92% of the product experience now runs on shared patterns. The number that mattered most was quieter: users moving between Yubi products stopped feeling like they had left the building.
Nobody at Yubi wanted an inconsistent experience. It happened anyway, one reasonable local decision at a time. Fixing it took a programme, a rhythm, and a gate, which is to say, it took leadership, not just design.