Enterprise software · 2025
Nova
Ten years of enterprise complexity reduced to one legible interface.
In short
Nova had accumulated a decade of features and four acquired products into an interface that required training to operate. Sparza audited every screen, rebuilt the product around the five jobs that mattered, and let a 62-component design system emerge from that work — a library the product teams now own and extend themselves.
The challenge
Nova sells operations software to mid-market manufacturers. Ten years of feature requests and four acquisitions had produced an application where the same action could be performed three ways, each inherited from a different product, each looking and behaving differently.
The commercial symptom was specific: customers bought confidently on the demo and then churned during onboarding. Sales could show the product working; customers could not reproduce it unaided. Every team also held a different idea of what a button looked like, so each new feature widened the gap.
What we found
We audited 240 screens, shadowed eleven customers through real tasks, and interviewed every product team. Four findings set the direction:
- Five jobs accounted for most of the value. Despite the surface area, customer time concentrated heavily in five workflows. Everything else was long-tail.
- Duplication came from acquisitions, not decisions. The three ways to do one thing mapped exactly onto three acquired codebases nobody had reconciled.
- Onboarding failure was navigational. Customers understood the concepts; they could not find where to act on them.
- Previous system attempts failed by starting abstract. Two earlier component libraries had been built up front, adopted by nobody, and abandoned.
The strategy
We rebuilt around the five jobs rather than around the feature list, giving each a clear primary path. Long-tail functionality was kept but demoted, so it stopped competing with the work customers actually do daily.
Given two failed library attempts, we inverted the usual order: the design system came out of the redesign rather than preceding it. A component earned its place only after appearing three times in real screens, which meant every entry had proven demand behind it.
Three decisions that shaped it
Five jobs, five primary paths
Each of the five core workflows got one obvious route through the product, with the alternatives inherited from acquisitions retired rather than hidden. Retiring them was contentious internally and was the single change customers noticed most in testing.
The system earned its way in at three uses
Nothing entered the component library until it had appeared in three real screens. Two previous attempts had failed by defining components ahead of demand; the rule meant the resulting 62 components all had genuine usage behind them and adoption did not need to be mandated.
Long-tail features demoted, not deleted
Enterprise customers depend on obscure functionality in ways no audit fully reveals, so nothing was removed. Rarely-used features moved out of primary navigation into a consistent, searchable secondary layer — which let the primary interface get dramatically simpler without breaking anyone's workflow.
How it was built
We worked screen-by-screen alongside the product teams rather than delivering a finished redesign. Each of the five jobs was designed, built, and released before the next began, so the system was validated continuously instead of at the end.
The design technologist worked inside the client's codebase from week three, which meant components were specified in the medium they would ship in. Accessibility was handled per component as it entered the library — keyboard paths, focus order, and contrast checked once, at source.
How the work ran
Each job shipped to a cohort of customers before the next started, so fourteen weeks produced five validated releases rather than one large migration with no feedback until the end.
Audit and discovery
240 screens catalogued, eleven customers shadowed, product teams interviewed.
Direction
Five jobs defined, primary paths mapped, two territories tested with customers.
Design and build
One job at a time, designed and shipped in sequence with components emerging from real work.
System consolidation
Library documented, accessibility verified, ownership transferred to the product teams.
The outcome
Four products became one interface, and the component library that came out of it is maintained by Nova's own product teams — the first of three attempts to survive past handover, because it was assembled from components they had already shipped rather than handed to them as a finished abstraction.
4 → 1
Products unified into one interface
62
Components, each proven in three real screens
5
Validated releases in fourteen weeks
What we handed over
- Audit of 240 screens with consolidation plan
- Five primary workflow designs
- 62-component design system, documented
- Accessibility specification per component
- In-codebase component implementation
- Ownership handover to product teams
Work with us on something similar
Nova drew on the services below. Tell us what you’re building, or see the rest of our work.