A UX audit checklist for SaaS products
Yes or no checks for one task in a SaaS product, from the first invitation to the failed save, with what a no usually means and what to fix first.
This is a checklist for auditing one task in a SaaS product, with the product open beside it. Each line is a statement you can check and mark yes or no. The groups follow the order a person meets the product, from getting in to recovering from a failure, and end with access and a test with real users.
Our guide to running a UX audit covers the method around it: choosing the task, gathering evidence, writing findings and ranking them. Mark a line not applicable when it does not fit your task, and not tested when you cannot reproduce it, because a check you could not run has not passed.
1. Before you start
2. Entry and setup
Start here, because a person who cannot get into the account never reaches the task you are auditing.
A no here usually means the product was tested by people who arrive the way the team does, with a fresh invitation and one account. Fix the invitation states first, since every other check depends on people getting in.
4. Data and consequential actions
This group covers the points where someone approves, exports or changes many records at once.
A no here usually means the interface reports what was asked for and stays quiet about what happened next. Fix the completion and partial failure states first, because they are where someone walks away believing a payment or an export went through.
5. Roles and handovers
Work in a SaaS product often passes between people. Run this group in the restricted and administrator accounts, with two test users for the last check.
A no here usually means the product was designed and tested in an administrator account. Start with the disabled controls, which need a sentence each. Give the server check to engineering as its own item, because the interface can look right while the rule is wrong.
6. Failure and recovery
This group needs a controlled test environment and an engineer nearby, because some checks interrupt a save on purpose.
A no here usually means the failure paths were never designed as states of their own. Fix anything that can lose work or repeat an action first, even if it happens rarely, and agree the cause with engineering before designing the message. A reassuring spinner does not make an operation safe to repeat.
7. Access and testing with people
The access checks are preliminary. W3C’s Easy Checks describe the same kind of first review and say it is not a complete accessibility evaluation.
A no on keyboard, zoom or contrast is usually a problem in a shared component, so fix it there and it is fixed on every screen that uses it. A no on the user test means you have not run one yet. GOV.UK’s guidance on moderated usability testing explains how to set tasks that do not give away the answer. Where users fail at a step your walkthrough passed, record it as a no and give it more weight than your own review.
8. After the last check
Write each no up as a single finding: the step, what happened, who it affects and how sure you are of the cause. Our guide to running a UX audit shows a finding written out in full and how to rank the list with your engineers.
When the first fixes ship, run this checklist again on the same task, in the same accounts, failures included.
If you would like to try working with us before committing to more, we run a UX audit of one journey as a one-week trial for $2,500, credited in full against work that starts within 60 days. An audit of a whole product, or anything larger, is scoped and priced before it starts.
9. Questions
A SaaS UX audit checklist should follow one task through six groups: entry and setup, navigation and account context, data and consequential actions, roles and handovers, failure and recovery, and access with a test on real users. Each line should be one thing you can mark yes or no, tried in new, established, restricted, administrator and interrupted accounts.
Yes. A product team can work through a UX audit checklist on one task, as long as someone tries it in accounts they do not use every day and records evidence for each result. The hard part internally is keeping what you saw apart from why you think it happened, because the team already knows how the product is meant to work.
No. A heuristic evaluation checks an interface against general usability principles, such as Jakob Nielsen’s ten heuristics. A UX audit checklist follows one task in your product from start to finish, in accounts with different roles and data, and includes the failure paths. Several of its checks test the same principles, such as visibility of system status and recovery from errors.
Partly. A UX audit checklist can include preliminary access checks: keyboard use, visible focus, zoom, contrast, headings, labels and error messages. W3C’s Easy Checks describe this kind of first review and say it is not a complete accessibility evaluation. If you need to show that a product meets an accessibility standard, commission an assessment against that standard.
It depends on how much of the product is in scope. For people who want to try working with us first, we audit one journey as a one-week trial for $2,500, credited in full against work that starts within 60 days. An audit of a whole product takes longer and is scoped and priced before it starts. Our UX audit cost guide covers what moves the price.
Written on by