Acceptance Test Procedure Guide: Tie ATPs to Requirements

Contents
- Step 1: Decide what the ATP is for (and how it differs from a qualification test)
- Step 2: Pick the requirements the ATP will verify
- Step 3: Write the ATP sections every reviewer expects
- Step 4: Trace each ATP step to a requirement and a pass/fail criterion
- Step 5: Run the test and store the evidence with the requirement
- Step 6: Re-check the ATP after every design change
- Step 7: Get the ATP through Test Readiness Review and sign-off
- Link the ATP to your requirements before the next design change
An acceptance test procedure (ATP) is the approved, step-by-step instruction for testing each deliverable unit, so you can show it conforms to its qualified design and is fit to ship. It is only as good as its link to the requirements it verifies, and that link breaks quietly the first time the design changes. This guide walks through writing an ATP, tracing every step to a requirement, and re-checking it after a change.
The primary purpose of acceptance testing is to confirm that a specific unit meets its requirements and is acceptable for delivery, and to catch workmanship or material defects before the customer does. It is not a second attempt at proving the design. You will need your requirement set, a verification matrix (or the start of one), and an agreed list of test equipment and interfaces. Everything below applies whether you call the document an ATP, a factory acceptance test (FAT) procedure or a final inspection and test plan.
Step 1: Decide what the ATP is for (and how it differs from a qualification test)
The ATP screens each deliverable unit for workmanship and material defects and verifies required function and performance. Qualification testing does something different. It demonstrates that the design and manufacturing approach meet requirements across mission environments, usually with margin.
The practical differences:
Article: an ATP runs on every deliverable unit. Qualification usually runs on dedicated hardware.
Coverage: an ATP implements a selected subset of the verification program. Qualification is broader and generally more severe.
Levels: an ATP uses acceptance or workmanship-screening levels and avoids consuming flight or service life. Qualification uses specified margins and durations.
Relationship: the acceptance program is established and validated during qualification, and ATP results feed the unit's acceptance data package.
Write one sentence at the top of the ATP stating which of these it is. If a reviewer can read it as a qualification test, someone will eventually ask why you did not run margin levels, or why you did on a unit that has to ship.
Step 2: Pick the requirements the ATP will verify
Start from the verification matrix, not from the test equipment you happen to have. Every requirement is verified by test, analysis, inspection or demonstration. The ATP covers only the ones assigned to test (or demonstration) at the unit level that are worth repeating on every article.
Filter the matrix with three questions. Does the requirement apply to each delivered unit, or was it closed by qualification analysis? Can it be checked at acceptance levels without damaging the unit? Does a workmanship defect show up in this measurement? Requirements that fail all three belong elsewhere.
List the selected requirement IDs in the ATP header. This list is the scope contract. A requirement that is not on it has no step in the procedure, and a step with no requirement behind it is a candidate for deletion. If your requirements live in a spreadsheet and your tests in another, this is where drift begins, so keep both in the same system. Tandem keeps requirements and test case management linked, so the selection is a query against live requirements instead of a copy-paste.
Step 3: Write the ATP sections every reviewer expects
A reviewer will look for the same sections in every ATP. Include:
Identification: hardware, part and revision, serial number, and configuration under test.
Applicable requirements and objectives: the IDs from Step 2.
Test equipment, software and interfaces: with calibration status for each instrument.
Setup: wiring, fixtures, environmental conditions and initial states.
Step-by-step instructions: numbered, one action per step, written so a technician who did not write it can run it.
Operating limits: the bounds the unit must not exceed during test.
Measurements and data sheets: exactly what gets recorded, in what units.
Pass/fail criteria: objective values, not judgment calls.
Anomaly and retest instructions: what to do on a failure, who can authorize a retest, and what must be re-run.
Approvals and result records: signature blocks for test conductor, quality and the customer where required.
The same structure covers a factory acceptance test. A FAT is a documented, usually customer-witnessed check at the manufacturer's site to confirm equipment meets agreed specifications before shipment. Agree the protocol and pass criteria in advance, because a FAT pass confirms only what was tested at the factory. Site-dependent functions may still need verification at a site acceptance test.
Step 4: Trace each ATP step to a requirement and a pass/fail criterion
Every measurement step carries three fields: the requirement ID it verifies, the expected value with tolerance, and the pass/fail rule. If a step has no requirement ID, it is either a setup action or it shouldn't be there.
A minimal trace looks like this:
| Step | Action | Requirement | Criterion |
|---|---|---|---|
| 14 | Apply nominal supply, record current | REQ-PWR-012 | Within the limit stated in the requirement |
| 15 | Command each actuator through full travel | REQ-MECH-031 | Travel and timing within stated bounds |
Two checks make the trace trustworthy. First, run it forward: every selected requirement from Step 2 appears in at least one step. Second, run it backward: every step maps to at least one requirement. Gaps in the first direction are unverified requirements. Gaps in the second are test time spent on nothing anyone asked for.
Write the criterion so it doesn't need the requirement text open next to it, but quote the requirement ID so it can be found. Tandem links test cases to requirements in one system, so the trace is a live relationship you can query instead of a column someone updates by hand. The verification methods guide covers the same idea in more depth.
Step 5: Run the test and store the evidence with the requirement
Run the ATP as written, record each measurement on the data sheet at the moment it is taken, and attach the raw data to the requirement it verifies, not to a project folder.
Three habits matter on the floor:
Execute exactly the revision that was approved. If a step can't be run as written, stop and log a deviation. Don't improvise and fix the document later.
Record failures as carefully as passes. An anomaly entry names the step, the observed value, the suspected cause and the disposition. A retest without a recorded reason is how acceptance packages get rejected.
Store evidence where the requirement lives. When a reviewer opens the requirement, the pass result, data sheet and test run should be one click away.
Tandem stores validation evidence in the same system as requirements and test execution, so a requirement shows its verification status next to the evidence behind it. Where your test or telemetry data comes from a platform like Nominal, Sift or Revel, Tandem works alongside them. That data can feed attributes that verify requirements, and the test platform keeps doing what it does best.
Step 6: Re-check the ATP after every design change
After a design change, treat the ATP as suspect until you have checked which requirements and steps the change touches. This is where most ATPs rot. The part gets revised, the requirement gets a new limit, and the procedure still tests the old number and still passes.
A workable re-check after each change:
Identify the requirements affected by the change.
Find every ATP step traced to those requirements.
Decide for each step whether the criterion, the setup or the measurement itself must change.
Revise the ATP, re-approve it, and note which units were tested to which revision.
Decide whether already-tested units need a delta retest.
Step 1 is the expensive one when requirements, CAD and tests live in different tools. In Tandem, Design Changes shows a before/after 3D CAD diff with a plain-language summary and the affected requirements, so the list of ATP steps to revisit starts from the change itself. Tandem works alongside your CAD through its plugin (SolidWorks, Onshape, CATIA, NX and Fusion), and requirements are verified continuously against live attributes, so a limit that moves is visible instead of discovered at the next run. For a worked example of the same discipline, see re-verifying requirements after a tooling change post-freeze.
Step 7: Get the ATP through Test Readiness Review and sign-off
Bring the ATP to Test Readiness Review (TRR) with the trace complete, the equipment calibrated and the approvals lined up, so the review is about readiness and not about discovering gaps. The reviewers' core question is simple: if we run this, will the result tell us whether each requirement is met?
What to have in the room:
The approved ATP revision and its requirement trace, forward and backward.
Calibration records for every instrument listed.
The unit's configuration and any open nonconformances against it.
The anomaly and retest rules, agreed in advance.
Named witnesses and signers.
After the run, the sign-off package is the executed ATP, data sheets, anomaly log and a statement of acceptance. Because evidence is stored with each requirement, the package assembles from linked records instead of being reconstructed. The Test Readiness Review guide covers the review itself in detail.
Link the ATP to your requirements before the next design change
An ATP that isn't traced to requirements will pass long after it stopped proving anything. Pick one live program, trace its ATP forward and backward this week, and see which steps have no requirement behind them. Then book a demo to see how Tandem keeps requirements, test management and validation evidence linked, and surfaces affected requirements when the design changes.
Visit Tandem
Tandem is the AI platform for hardware engineering. It connects requirements, system architecture, CAD design changes, design reviews, PDM and validation evidence in one system, so design intent survives every review, change and handoff. It works inside SolidWorks, Onshape, CATIA, NX and Fusion and alongside PLM, Jira, Slack and Teams, for hardware teams building complex products from prototype through production.
Get startedSources
Frequently asked questions
What is the primary purpose of acceptance testing?
Acceptance testing confirms that a specific delivered unit meets its requirements and is acceptable to ship, and it screens for workmanship and material defects. It does not re-prove the design. Tandem links test cases and evidence to requirements so each acceptance result stays tied to what it verifies.
What is the factory acceptance test procedure?
A factory acceptance test is a documented, usually customer-witnessed check at the manufacturer's site that equipment meets agreed specifications before shipment. The procedure covers scope, test sequence, pass criteria, calibrated instruments, result records, failure handling and sign-off. A FAT pass covers only what was tested there.
How is a FAT test done?
Approve the protocol and criteria first, prepare calibrated equipment and documents, inspect the unit against drawings, then test components, functions and operating modes. Record every result, log failures with an owner, retest as required, and close out with a signed report of results, open items and the acceptance decision.
How is an ATP different from a qualification test?
An ATP runs on every deliverable unit at acceptance levels to screen for defects and verify function. Qualification demonstrates the design meets requirements across environments with margin, usually on dedicated hardware. Acceptance methods are validated during qualification, and ATP results feed the unit's acceptance data package.
How do I keep an ATP current after a design change?
Identify the requirements the change touches, find every ATP step traced to them, revise criteria or setup, and re-approve the procedure. In Tandem, Design Changes shows the CAD diff with affected requirements, so you know which linked test steps to revisit.
Related reading
Written by
Tandem
Tandem is the AI platform for hardware engineering. It connects requirements, system architecture, CAD design changes, design reviews, PDM and validation evidence in one system, so design intent survives every review, change and handoff. It works inside SolidWorks, Onshape, CATIA, NX and Fusion and alongside PLM, Jira, Slack and Teams, for hardware teams building complex products from prototype through production.