UX Audit · Two weeks
Find out why usersget stuck
People start an important task and don’t finish it. In two weeks we review that one flow alongside your evidence, then show your team what to fix first and what an improved flow could look like.
- The scope
- One flow, one user roleUp to 10 screens or states
- The timing
- Two weeksScope on day 1, walkthrough on day 10
- The agreement
- Fixed scopeAgreed in writing before we start
When it helps
Analytics show where usersdrop off, not why
- New users sign up but never reach the point where the product becomes useful to them.
- People drop off during onboarding, an application, setup or an approval.
- Support keeps answering the same “how do I…?” questions.
- Your team has a long list of ideas and no agreement on what to fix first.
We review one critical flow screen by screen, alongside the analytics, research and support feedback you already have. Customer-facing products and internal tools are both in scope.
What your team walks away with
A ranked fix list anddesign directions
By day 10 you’ll have evidence-led priorities, the likely sources of friction and recommendations to review with engineering. Findings separate observed problems from assumptions that need testing.
- 01
An annotated review
Up to 10 key screens or states of one flow, covering structure, interaction, content clarity and visible accessibility issues.
- 02
A prioritised fix list
Every issue rated for severity, confidence and rough effort, with recommended changes, dependencies and what needs testing first. Ready to plan with engineering.
- 03
Two or three design directions
Example concepts for the biggest problems, so each recommendation shows what the change could look like. A clear direction, not a full redesign.
- 04
A walkthrough and follow-up
A 60-minute session with your team to go through the findings and agree priorities, then one round of follow-up questions answered in writing.
- 05
A way to tell if it worked
On day 1 we agree the measure that matters, such as setup completion or drop-off at a step, and note your current baseline. Each priority in the fix list names the signal to watch, so you can check the result a few weeks after the fixes ship.
What a finding looks like
An example findingfrom an audit
Each finding ties a moment in the journey to its consequence, the evidence behind it and a next step. This excerpt is an illustrative example, not client data.
| Finding | Evidence | Severity | Confidence | Effort | Recommended next step |
|---|---|---|---|---|---|
| Admins can’t see which imported staff records failed, so they re-import everything. | Reproduced in the test account; matches recurring support themes. | Blocker | High | Medium | Show failed rows inline with the reason, and let admins fix and retry them. |
| Role names use internal terms that customers don’t recognise. | Observed in the flow; echoed in “which role do I pick?” tickets. | Serious | Medium | Low | Rename roles in plain language, then check with five admins. |
| Setup progress isn’t saved between sessions. | Reproduced; funnel data shows returning admins restarting at step one. | Serious | High | High, engineering to confirm | Save progress per step and show what’s left on return. |
How it works
Two weeks from scopeto walkthrough
The clock starts once we’ve agreed the scope and you’ve shared access. Working days throughout.
- Day 1
Scope
We agree the flow, the user role and what success looks like. You share access and any analytics, research or support feedback.
- Days 2–8
Review
We go through the flow alongside your evidence, keeping what we can see separate from assumptions that need testing.
- Day 10
Walkthrough
We present the findings, agree what to prioritise with your team, and you get one round of follow-up questions.
Focused on purpose
One workflow,reviewed in depth
One flow, one user role and one device, up to 10 screens or states. A different error, empty or permission state counts as one of the ten.
More than one role, or more than ten states?
Handovers between roles, such as a requester and an approver, and long setup flows, such as import, then permissions, then a first action, are common. We’ll suggest whether to split the flow into two audits or extend one, before you book.
Evidence, clearly labelled
Every finding says whether it’s an observation, backed by your data, or a hypothesis to test. An expert review can find likely problems; it can’t tell you how many users hit them without data.
Not a fit if
You need a full WCAG or code audit (we flag visible accessibility issues, but this isn’t a compliance audit), you want new research with recruited users (ask us before booking), or you already know what’s wrong and need it designed.
Experience behind the review
Led by a designer withengineering experience
Every audit is led by Marian Cozma, who started as a front-end engineer, so effort ratings are realistic and the recommendations are buildable.
Financial services / Complex workflows
Prioritising supervisory alerts for a global investment bank
Research split interface issues from technical and operational ones, so design effort went where design could help.
Read the case studyTelecoms / Platform design
Organising 1,200+ pages into templates for a telecoms provider
An audit of more than 1,200 pages became reusable page families and responsive templates that went into production.
Read the case studyGetting started
Tell us about theflow to review
Tell us about the product, the task people need to complete and what prompted the review. A few lines is enough.
Discuss your flowScope, deliverables and start date agreed in writing before we begin.
What we need from you
- A test environment or demo account with realistic, non-sensitive data.
- The flow and user role to focus on.
- Known constraints and one person who can make decisions.
- Any analytics, research or support feedback you have.
What if we don’t have analytics or research?
We can still review the flow. Findings will clearly separate expert observations from hypotheses that need validating, and we’ll suggest the lightest way to check the important ones.
What format are the findings in?
Annotated screens in Figma and the fix list as a spreadsheet, so it can go straight into Jira, Linear or whichever backlog tool you use. The walkthrough can be recorded for people who can’t attend.
Who does the work?
Marian Cozma leads every audit personally, from scoping to walkthrough. No account managers and no handovers.
Will you sign our NDA?
Yes. We sign your NDA before you share access or materials, work only with test data, and delete what you’ve shared when the engagement ends. See working with us for details.
What happens after the audit?
Your team can act on the fix list straight away. If you’d like the biggest problems designed in detail, that’s Complex Workflow Design. There’s no commitment to further work.