User acceptance testing checklist: before you sign off
About to sign off a new website or app? A user acceptance testing checklist for clients: who tests, what to test, how to log a defect, and what accepted means.
EK Ra Sunya team
EK Ra Sunya Inc.
User acceptance testing is the last check before you accept delivered software: the people who will actually use it run through real tasks, on real devices, with realistic data, and compare what happens against what you agreed. A useful user acceptance testing checklist covers five things. Who tests, which scenarios, on what devices, how defects are reported, and what "accepted" commits you to.
Treat it as a quick click-around the day before go-live and you risk signing off an invoice screen that rounds wrongly, or a booking form that fails on an older Android phone. Once you've accepted it, that's your problem.
What is user acceptance testing, exactly?
The ISTQB glossary defines it as a type of acceptance testing "performed to determine if intended users accept the system". Acceptance testing in general is the test level focused on deciding whether to accept the system at all.
Two words in that definition matter. Intended users means the people who will use it every day, not the person who signed the contract. Accept means it's a decision, not a demo. Your developer has run their own tests already. UAT asks a different question: does this do the job we need, the way we agreed?
Who on your side should run UAT?
Pick testers by role, not by seniority. For each type of user in the system, find one person who does that job now.
- A front-line user for each role. The receptionist who will take bookings, the accountant who will raise invoices, the field officer who will use the mobile app on site.
- One owner of the requirements. Someone who knows what was agreed and can say "that's in scope" or "that isn't". Usually the person who wrote or approved the brief.
- One decision-maker. The person who signs acceptance. They don't need to test every screen, but they need to read the defect list before signing.
Give the testers time in their calendar. UAT squeezed between normal work turns into a few clicks and a thumbs-up.
How do you write test scenarios?
Start from your requirements document and your Definition of Done. If you wrote these before the quote (our guide on writing software requirements before you ask for a quote covers how), most of the work is already done. Every job in the requirements becomes one or more scenarios, and every line in the Definition of Done becomes something a tester checks.
A good scenario names the user, the starting point, the steps and the expected result:
- Scenario: Branch manager approves a leave request.
- Start: Logged in as a branch manager with one pending request.
- Steps: Open the request, approve it, add a note.
- Expected: Request shows as approved, the employee gets a notification, the leave balance drops by the requested days.
Then add the awkward cases. What happens with a request that overlaps a holiday, a name with an apostrophe, an amount with decimals, a form submitted twice, a session that times out halfway through? Real users will find these eventually. Better that your testers find them first.
Which devices and data should you test with?
Test on the devices your users actually have, not the developer's laptop. The GOV.UK Service Manual tells teams to use analytics to understand how users reach a service and to test in a defined list of browsers. If you have an existing website or system, its analytics will tell you which phones and browsers matter. If you don't, test at minimum on one recent Android phone, one older or cheaper Android phone, one iPhone, and Chrome and Safari on a computer.
For mobile apps, testers can install pre-release builds on their own phones before anything goes to the public store:
| Platform | Pre-release route | What the official docs say |
|---|---|---|
| iOS | TestFlight | Up to 100 internal testers and up to 10,000 external testers. The first build for external testers needs App Review. Testers can send screenshots marked up with feedback. |
| Android | Google Play internal testing | Up to 100 testers per app, who join through an opt-in link. New builds reach testers within minutes. |
Data matters as much as devices. Ask for a test environment loaded with realistic data: your real product names, real price ranges, a customer list of a believable size. Don't use live customer data unless you've agreed how it's protected. A system that works with five test records can crawl with five thousand, and a report that looks fine with "Test Product 1" can break on a 60-character product name.
If the site or app must work for people with disabilities, run an accessibility pass during UAT as well. Our website accessibility checklist gives you a short list to work through.
How do you log a defect so it actually gets fixed?
A defect report that says "the invoice page doesn't work" will come back with a question, and a day is lost. A report a developer can act on straight away has six parts:
- What you did. The exact steps, starting from where you were.
- What you expected. Quote the requirement or scenario if you can.
- What happened instead. Including any error message, word for word.
- Where. Device, browser or app version, and which test account.
- Evidence. A screenshot or short screen recording. On iOS, TestFlight's built-in feedback does this for you.
- Severity. How much it stops the work (see below).
Keep every defect in one shared list, a spreadsheet or the developer's issue tracker, with one row per defect and a status column. WhatsApp messages and phone calls get lost. A list lets both sides see what's open, what's fixed, and what has been retested.
How serious is each defect?
You don't need a formal standard. A three-level scale agreed with your developer before UAT starts is enough:
- Blocker. A core task can't be done, data is wrong or lost, or there's a security problem. Not accepted until fixed.
- Major. A task can be done, but only with a workaround, or it fails for some users or devices. Usually fixed before acceptance, or accepted with a written fix date.
- Minor. Cosmetic or wording issues that don't stop anyone working. Can be listed and fixed after go-live if both sides agree.
Agree the scale in writing first. Arguing about whether something is "major" in the middle of UAT wastes everyone's time.
Is it a bug or a change?
Some of what your testers report won't be a defect at all. "Can the report also show last year's figures?" is a good idea, but if the requirements never asked for it, it's a change, not a bug. The test is simple: does the software do what was agreed? If it doesn't, it's a defect. If it does, but you now want something different, it's a change.
Log changes in a separate list so they don't hold up acceptance. Our post on change requests in fixed-price projects covers the grey cases and how to handle them. At EK Ra Sunya a change request follows a set order: a written request, then a time and cost estimate, then your approval before anything is built.
What should "accepted" mean?
Acceptance should be a written statement, not a silence. Before you sign, check that:
- every scenario has been run and marked pass or fail
- no blocker is open, and every open major has a written fix date
- open minors are listed by name, so "accepted" can't later be read as "accepted with no known issues"
- changes found during UAT are on their own list, outside the acceptance decision
- you know what support or maintenance terms apply after this date, as set out in your contract
Read your contract for what acceptance triggers. It often releases a payment, starts a support period, or both. Don't sign because a launch date is close. Sign because the list says you can.
Frequently asked questions
How long should UAT take?
It depends on how many scenarios you have and how many defects turn up. Agree a UAT window in the project plan, with time for at least one round of fixes and retesting, rather than a single day before launch.
Can the developer do UAT for us?
No. The developer should test before handing over, but UAT is your check that the system does your job. Only your own users can tell you that a screen works on paper but doesn't fit how the work is actually done.
What if we find a defect after signing off?
Check your contract and support terms, because they decide how post-acceptance defects are handled. This is why open minors should be listed in writing at sign-off.
Do we need special testing software?
No. A spreadsheet with scenarios and a defect list is enough for most projects. If your developer already uses an issue tracker, ask for access and log defects there.
Planning a project with UAT built in
Our web development team builds websites and web applications, and UAT is easier when scenarios are agreed at the start rather than written the week before handover. If you're planning a project, tell us what you need.
Have a project in mind?
Let us help you build something scalable, fast, and built to last.
Start a conversationKeep reading
Website accessibility checklist for a business website
A website accessibility checklist you can run yourself in an hour: keyboard, zoom, contrast, alt text, forms, headings, captions, focus and target size.
Web DevelopmentProgressive web app vs native app: when a PWA is enough
Progressive web app vs native app for a Nepali business: what a PWA can do with no store account, where iOS limits bite, and when you still need a store app.