StackLift
User Acceptance Testing: A Practical Guide for Software Buyers
Client VisibilityClient Visibility
Client Visibility / 5 min read

User Acceptance Testing: A Practical Guide for Software Buyers

A practical guide to user acceptance testing for software buyers: who should test, how to write criteria, how to run the cycle, and when to sign off.

How this article was made: drafted with AI assistance, then reviewed and approved by the StackLift team before publishing.

User acceptance testing is often the first time a business really uses the software it paid for, and it is frequently squeezed into the last days before launch. Done well, it catches the problems only real users can see, such as a workflow that passes every test and still does not match how the team works. This guide is for the buyer's side of a project: how to prepare, how to run the cycle, and how to make a sign-off that means something.

What this article makes clear

  • UAT confirms that software fits the business, while QA confirms that it works as specified.
  • Use real end users with protected time, realistic data, and end-to-end scenarios that include exceptions.
  • Agree on acceptance criteria before the build and exit criteria before testing starts.

What UAT is for, and what it is not for

The delivery team's QA asks whether the software works as specified. UAT asks a different question: does it let our people do their jobs? It checks the fit between the system and the business, including the awkward cases that only show up in real work, such as the customer with two accounts, or the approval that needs to happen when the manager is away.

UAT is not a substitute for QA. If your testers are finding basic crashes and broken pages, the build was not ready, and their time is being spent on the wrong problems. Agree in advance what state the software must be in before UAT begins, and send it back if it falls short.

Choose testers who do the work every day

The best testers are the people who will use the system, including at least one who is skeptical about it. Project sponsors and managers often know how a process is supposed to run, while the people doing the job know how it actually runs. It is the gap between those two that UAT is there to expose.

Protect their time. Testing on top of a full workload tends to become a rushed click-through at the end of the day. Get their managers to agree to dedicated hours, and name a UAT lead who collects feedback, chases answers, and keeps the process moving.

Write acceptance criteria before the build begins

Acceptance criteria turn opinions into checks. Instead of "invoicing should be easy", write "an accounts user can create an invoice from an approved order, apply tax by region, and send it to the customer". If this is agreed before development starts, the team builds toward it and the conversation at the end has a shared reference point.

Base your test scenarios on real workflows from start to finish, not on individual screens. Include exceptions, such as a rejected payment, a cancelled order, or a user without permission. Those paths are where problems tend to hide, and where a demo will not take you.

How to run a UAT cycle

Use a stable environment that matches production as closely as possible, with realistic data. Anonymize anything sensitive, and do not test against live customer records. Freeze new features during the cycle, so testers are not chasing a moving target. A short briefing helps people understand the scope and how to report what they find.

Run the scripted scenarios first, then leave time for people to explore and try things the way they naturally would. Report every issue in one place, with steps to reproduce it and a screenshot. Then triage together using simple severity levels: blocks go-live, must fix soon after, or nice to have. Requests for new features are valid, but they go through change control and not the defect list.

Allow for at least one retest round. Fixes can break other things, and planning a single pass assumes nothing will go wrong, which is the one assumption UAT exists to challenge.

Making a sign-off that means something

Before testing starts, agree on the exit criteria: for example, all scenarios run, no open blocking defects, and an agreed plan for anything left over. A named business owner then makes the decision. Sign-off should state what was tested, what is still open, and who accepted that risk, because accepting a known issue is a legitimate choice as long as it is made on purpose.

Check what sign-off means commercially as well. It can trigger a payment or begin a warranty period, so read the contract before signing. The decision should also cover readiness beyond the software: is training done, is support ready, is there a rollback plan?

Why shared visibility makes UAT easier

UAT goes badly when it is a surprise, and when testers see the product for the first time at the end. Regular demos through the build mean the big misunderstandings surface early, while they are still cheap to correct, and leave UAT to confirm rather than discover.

It also helps when the buyer and the delivery team work from the same record. In the StackLift AI client portal, clients see live project state, approvals, invoices, and releases, so what has been delivered and what has been accepted is clear to everyone involved.

Frequently asked questions

Common questions on this topic.

Who is responsible for user acceptance testing?

The buyer or business owns UAT, since only they can confirm the software suits their work. The delivery team supports it by providing the environment and test data, and by fixing defects. A named UAT lead on the business side should coordinate testers and bring decisions to the sign-off owner.

How long should UAT take?

It depends on the number of workflows and how available your testers are, so there is no standard length. Plan enough time for scripted scenarios, free exploration, fixes, and at least one retest round. If your plan leaves no room for fixes, it is not really a testing phase.

What is the difference between UAT and QA?

QA is run by the delivery team to check that software meets the specification and has no defects. UAT is run by the people who will use it, to check that it meets the business need. A product can pass QA and still fail UAT if the specification missed something.

Can we go live with open UAT defects?

Yes, if none of them blocks the business and a named owner accepts the risk. Record each open defect with a workaround and a date for its fix. Going live this way is often sensible. Going live without knowing what is still open is the mistake to avoid.

Apply this to your project

Turn your idea into an architect-reviewed delivery plan.

Use the StackLift estimate flow to convert your requirements into scope, story points, risks, timeline, and a practical delivery baseline.

Start an estimate