Acceptance test

Definition

An acceptance test sets out objective criteria that a deliverable must meet for the client to accept it. In IT and engineering contracts the parties agree the test scenarios, success thresholds and a window for re-testing after defects are fixed. Passing the test typically triggers acceptance, payment milestones and the start of warranty periods, so the criteria are negotiated with care.

Example

The software is deemed accepted once it runs the agreed test scenarios for ten consecutive business days without critical errors.

Why this is a business risk

Vague or absent acceptance-test criteria are a leading cause of IT project disputes. Without agreed pass/fail thresholds, a supplier can claim the deliverable is "good enough" while the client insists it is not, and neither side has an objective anchor. This delays both payment and the start of warranty protection.

How to manage it

• Define test scenarios, input data and expected outcomes in an annex to the contract before work begins, so there is no room for later disagreement on what "passing" means.
• Set a clear re-test window with a maximum number of attempts so the acceptance process does not drag indefinitely.
• Specify what happens if the client fails to test within the agreed window -- most contracts deem non-response as acceptance after a set period.
• Track the test schedule as a milestone and send reminders when the test window opens and closes.
• Record test results formally in writing, dated and signed by both parties, to close out the acceptance step.