Improve a journey when the underlying product is sound. Redesign the interface when inconsistency affects several tasks. Investigate a rebuild when the architecture prevents safe, maintainable improvements.
01 / Define the problem before naming the project
“The app feels old” describes a reaction, not a diagnosis. Ask what happens in practice: customers fail to finish registration, staff cannot find a record, a release breaks an existing flow or the product cannot support a required integration. These problems can coexist, but they do not have the same remedy.
Combine support questions, observed sessions, interviews, analytics and an engineering review. Check whether instrumentation represents the real journey before interpreting a funnel. If data is limited, start with task walkthroughs and documented technical constraints rather than a confident percentage forecast.
02 / Compare the three interventions
A focused UX improvement changes a defined task: clearer navigation, fewer unnecessary inputs or better feedback. A broader interface redesign establishes consistent components, information hierarchy and interaction patterns across the product. A technical rebuild replaces substantial foundations, often including data access, infrastructure or the application architecture.
Use the comparisons below as a discussion framework, not an automatic scoring system. A technical investigation may reveal that an incremental migration is more practical than either a cosmetic refresh or a complete replacement.
| Intervention | Best starting signal | Main risk |
|---|---|---|
| Improve a journey | One task causes repeated confusion; the implementation can change safely. | Fixing a symptom without checking the wider journey. |
| Redesign the interface | Different areas use inconsistent navigation, components or terminology. | Replacing familiar behaviour without testing it. |
| Rebuild or migrate | Critical dependencies, reliability or architecture block necessary change. | Underestimating parity, migration and operational continuity. |
Choose the smallest intervention that can solve the verified problem and remain maintainable.
03 / Make accumulated UX problems visible
Audit the important journeys from entry to completion. Record the issue, who encounters it, the consequence and the evidence. Group recurring causes: conflicting terminology, duplicated controls, missing states or a structure that reflects internal departments rather than customer tasks.
Nielsen Norman Group describes UX debt as experience problems that accumulate through shortcuts and unresolved decisions. A useful response is a prioritised backlog with ownership, rather than a visual refresh with no explanation of what it improves. Separate urgent blockers from inconsistencies that can be resolved gradually.
04 / Ask engineering what can change safely
Review dependency support, build reliability, test coverage around key journeys, data ownership and integration boundaries. Ask the team to demonstrate a representative change. A slow release process may improve through targeted refactoring and tests; a complete replacement is not the only route to a healthier product.
If a rebuild is proposed, request a list of constraints it resolves and a migration plan. Include existing customers, permissions, data quality, old links and compatibility with connected systems. Name what will run in parallel, how success will be checked and how the team could roll back a problematic release.
A staged migration
- PilotMove a small group first.
- Run in parallelKeep the old and new systems available.
- ValidateCheck records and critical tasks.
- Switch overMove the rest once the checks pass.
05 / Test the decision on one valuable journey
Choose a task with a clear outcome and representative users. For a service CRM, that might be finding an appointment and recording the next action. Prototype the proposed experience and check completion, hesitation and comprehension. Use the findings to change the design before funding the whole rollout.
Matrix CRM is a relevant interface reference in our portfolio: it brings operational tasks into a mobile workspace. Its screens can support a conversation about hierarchy and task structure. They cannot tell you whether your own codebase requires replacement; that needs investigation of your product.
06 / Agree what will count as improvement
Define measures before release: task completion, time on a specific task, activation or support requests tied to a particular problem. State the audience, observation period and comparison conditions. A change in traffic source or customer mix can affect results even when the interface is unchanged.
Release in a controlled way where possible. Watch for regressions as well as the intended outcome, and leave time for corrections. If the team cannot yet explain the baseline, begin by improving measurement. An honest decision with uncertainty stated is more useful than a confident rebuild based on a screenshot.
What to take away
Separate experience debt from technical constraints, test one meaningful journey and account for migration. Choose the intervention that solves the evidence-backed problem with manageable delivery risk.
Bring the product you already have.
Share the task that needs to work better, the constraints and any existing research. We can discuss a useful first investigation.
Discuss a redesign