At a sprint review, a developer says the “reset password” story is done. The tester says it is not tested on mobile. The product owner says the email wording was never approved. The security lead asks whether the reset link expires. Everyone is telling the truth about their own idea of “done”. Agile teams use three separate agreements to avoid exactly this, and they work best when each is kept to its own job.
Three questions, three agreements
| Agreement | Question it answers | Scope | Who mainly owns it |
|---|---|---|---|
| Definition of Ready (DoR) | Is this item clear enough to start? | Every backlog item, before sprint planning | The whole team, often driven by the product owner |
| Acceptance criteria | What must this particular item do? | One story | Product owner, refined with developers and testers |
| Definition of Done (DoD) | What quality steps has every finished item gone through? | Every item and the increment as a whole | The team, sometimes set at organisation level |
The distinction between the last two causes most confusion. Acceptance criteria are specific to one story: “a reset link expires after 30 minutes”. The Definition of Done is generic and applies to all stories: “code reviewed, unit tests added, automated regression passing, no open critical defects”. A story is finished only when it meets both.
It is worth knowing their status in Scrum. The Scrum Guide (2020 edition) makes the Definition of Done a formal commitment attached to the Increment: work that does not meet it cannot be released or even presented at the sprint review as complete. The Definition of Ready is not in the Scrum Guide at all. It is a widely used team practice, which is why its content varies far more between teams.
Acceptance criteria, worked through for a login feature
Good acceptance criteria are testable statements of behaviour, written from the user’s or the business’s point of view, without dictating the implementation. Take a story: “As a registered customer, I want to log in with my mobile number and password so that I can see my orders.” A useful set of criteria might read:
- A registered customer who enters a correct mobile number and password is taken to the My Orders page.
- If the mobile number or password is wrong, the message “Mobile number or password is incorrect” is shown, without revealing which one was wrong.
- After 5 consecutive failed attempts, the account is locked for 15 minutes and the customer is told how to reset the password.
- The mobile number field accepts only 10 digits; a leading +91 or 0 is stripped before validation.
- The password field masks input, with an option to show it.
- A successful login session stays active for 30 minutes of inactivity, after which the customer must log in again.
Look at what each one does for a tester. Criterion 2 contains a security decision (do not reveal which field was wrong) that a developer might otherwise miss. Criterion 3 contains two boundaries, the 5th attempt and the 15-minute window, that point directly to boundary value tests. Criterion 4 settles an argument about formats before it happens. None of them mentions database tables, frameworks or button colours.
Some teams write criteria in the Given/When/Then style, such as: given a customer with 4 failed attempts, when they enter a wrong password again, then the account is locked for 15 minutes. That form maps neatly into automated BDD scenarios. Plain bullet statements work just as well if they are specific.
Weak criteria share a few signs: words like “fast”, “user-friendly” or “secure” without a measure; restating the story (“user can log in”); or a single criterion covering only the happy path.
What goes into a Definition of Ready
A DoR is a short checklist the team uses during refinement. A typical one asks whether the story has a clear user and purpose, acceptance criteria the team agrees are testable, known dependencies identified (an API from another team, a design from UX, test data from the client), an estimate, and a size small enough to finish within one sprint.
Testers contribute here more than anywhere else. Asking “how would I test this?” during refinement exposes missing criteria, unavailable environments and test data needs early. If a story depends on a payment gateway sandbox the team does not yet have access to, that is a readiness problem, not a testing problem to discover on day eight.
The risk with a DoR is turning it into a gate that blocks all flexibility. Agile coaches have warned that a rigid DoR can drift back towards waterfall-style sign-offs. A practical approach is to treat it as a conversation guide: if an item is not ready, the team discusses what is missing and decides, rather than refusing it automatically.
What goes into a Definition of Done
A DoD describes the quality bar for every item. A realistic example for a web product team:
- Code peer reviewed and merged to the main branch.
- Unit tests written for new logic; build and all automated tests pass in CI.
- All acceptance criteria verified, with automated checks added at the appropriate layer.
- No open defects of critical or high severity against the story.
- Tested on the agreed browser and device list, including accessibility checks for new screens.
- Deployed to the staging environment; user-facing documentation or release notes updated.
Testers help shape this list, and its contents should rise over time. A team that cannot yet automate regression might start with “regression checklist run manually” and tighten it as automation matures. Lowering the DoD quietly near a deadline is the classic warning sign: the work still gets called done, but the gap reappears later as technical debt and production defects.
Keeping the three in their lanes
Two habits keep these agreements useful. First, never copy DoD items into each story’s acceptance criteria; it bloats stories and hides the criteria that actually matter. Second, when a defect escapes, ask which agreement should have caught it. A missing business rule points to acceptance criteria. A missing dependency points to readiness. Skipped regression points to the Definition of Done. That one question turns each escaped bug into a specific improvement instead of a general complaint.
Written by the RP Strativa Team. The login rules and limits in the examples are illustrative.
Leave a Reply