Contact form acceptance criteria you can actually test
Turn 'the contact form works' into testable acceptance criteria for validation, failed submissions, duplicate clicks and delivery. Includes a copyable template.
“The contact form works” is a difficult requirement to sign off. Does it mean the button responds, the visitor sees a confirmation, or the message reaches the right inbox?
Write acceptance criteria for each of those outcomes before testing. A useful criterion names the starting condition, the action and an observable result. The examples below are proposed requirements for a fictional contact form; adapt them to your site’s actual delivery flow.
Agree on what counts as a successful submission
Start with a small field contract. For example: name, email and message are required; company is optional. Decide the permitted lengths, how whitespace is handled and where accepted messages go. Record these decisions in the project specification, where the developer and reviewer can both find them.
Then define the confirmation. If the server has accepted a message for processing, say that. Do not promise that someone has read it or that an email reached its recipient unless the system can support that claim.
For a manual QA pass, use a test inbox and clearly labeled sample data. Agree on who will check the destination. A green notification on the website cannot answer that part of the test for you.
Make validation failures specific
Here is one proposed criterion:
Given a completed message and an invalid email value, when the visitor submits the form, the form identifies the email problem, keeps the other entered values and allows correction without retyping the message.
Specify the error text or its intended meaning. “Something went wrong” does not tell the visitor which input needs fixing. Include an empty required field and a value just beyond the agreed length limit in your test data.
Browser validation helps catch input mistakes, but it can be bypassed. The server must validate submitted data too, as MDN’s form validation guide explains. Treat server validation as a separate developer check; passing the visible form test does not establish it.
Test the moment after Send
The awkward cases happen while a request is in progress. Define the behavior instead of leaving the reviewer to guess:
| Situation | Proposed acceptance criterion | Evidence to collect |
|---|---|---|
| Valid submission | The agreed destination receives one test message with the expected fields. The visitor sees the agreed confirmation. | Test identifier, destination result and screen state |
| Repeated clicks during submission | Repeated activation does not create multiple messages for the same attempt. | Count at the test destination |
| Server rejects the request | No success confirmation appears. The visitor can correct the problem or retry, as appropriate. | Visible error and retained values |
| Connection fails before acceptance | The page explains that submission could not be confirmed and preserves the message for recovery. | Screen state and recovery result |
These are product decisions to implement and verify, not claims about what every form already does. Disabling a button may help with repeated clicks, but it is not proof that the server prevents duplicates.
A timeout deserves particular care: the server may have accepted a request even if the browser never received the response. Ask the developer how retries are handled before approving automatic retry behavior.
Include a keyboard route and a delivery check
Try completing, correcting and submitting the form without a mouse. Record whether each control can be reached and whether the error or confirmation is understandable with the assistive technology in your test scope. Our keyboard QA guide covers a navigation check; forms need their own pass as well.
For delivery, put a unique marker such as QA-CONTACT-01 in the test message. Find that exact marker in the agreed destination. Keep private customer information out of screenshots and reports.
Copy this into the task
Page / language / build:
Required and optional fields:
Length limits and whitespace rules:
Accepted submission means:
Expected confirmation:
Invalid-input behavior:
Pending, failed and repeated-submission behavior:
Test destination and person verifying it:
Keyboard / assistive technology / browser scope:
Test identifier:
Observed result and evidence:
Untested cases or open decisions:
After a fix, rerun the failed case and one valid submission. Use the bug report template for unexpected results, and keep the acceptance criteria with the task so the next reviewer can make the same decision.