QA FundamentalsLesson 13 of 13
Bahasa IndonesiaThe tester in a sprint
What you do in each ceremony, how to write acceptance criteria that are testable, and how to avoid the mini-waterfall trap.
11 minQA Fundamentals
The trap first
Most teams that say "we're Agile" run a mini-waterfall: developers build for eight days, throw it over the wall on day nine, and QA has a day and a half to test everything. Sprint after sprint, testing is the thing that gets squeezed, and quality becomes a negotiation with the calendar.
The fix isn't heroics on day nine. It's being involved from day zero.
Your job, ceremony by ceremony
Backlog refinement — the highest-leverage hour of your sprint. You read the stories before anyone estimates them and ask the questions that turn assumptions into decisions:
- What should happen when it fails? (network, payment, permission)
- What are the boundaries of that number?
- What happens to existing data — is there a migration?
- Who can't do this? Which roles are excluded?
- How will we know it worked in production?
Every one of these is cheaper here than anywhere later.
Sprint planning — make testing visible. Testing effort is part of the estimate, not a tax paid afterwards. If a story can't be tested in the sprint, that's a planning fact, not a QA problem.
Daily stand-up — what you're testing, what's blocking, what's at risk of not being testable before the end.
During the sprint — test each story as it becomes ready, not all of them at the end. Pair with the developer before they call it done; five minutes of "what happens if I do this?" at their desk beats a defect report tomorrow.
Sprint review / demo — you often know the feature best. Demo the edge cases too, not just the happy path.
Retrospective — bring evidence, not vibes: escaped defects, where the time actually went, which stories arrived untestable.
Testable acceptance criteria
The single most useful artifact you can influence. Given/When/Then works because it forces concrete data:
Given a cart containing 1 × SKU-1042 and stock of 0 When the customer clicks Checkout Then the cart page shows "Kaos Polos is out of stock — remove it to continue" and no order is created
Compare with: "Checkout should handle out-of-stock items properly." The second one cannot fail a test, which means it cannot pass one either.
Push for criteria that name the data, the trigger, and the observable result. If you get that into refinement, half of your test design is already done — and so is the developer's.
Definition of Done
The team's shared bar for "finished". A DoD that includes testing looks like:
- Acceptance criteria covered by tests, and those tests pass
- No open critical or high defects on the story
- Regression around the touched area executed
- Automation added or explicitly deferred with a reason
- Docs/release notes updated
The value isn't the list; it's that "done" stops being an opinion.
The agile testing quadrants, briefly
A map for "what kind of testing are we even talking about":
| Business-facing | Technology-facing | |
|---|---|---|
| Support the team | Q2: functional tests, story tests, examples | Q1: unit & component tests |
| Critique the product | Q3: exploratory, usability, UAT | Q4: performance, security, reliability |
Most teams do Q1 and Q2 and forget Q3 and Q4 until something burns. Knowing the map is enough to ask the question in planning.
Risk-based testing in one paragraph
You never have time for everything, so rank by impact × likelihood. Payment and permissions are high impact; recently-changed, complex, or historically buggy areas are high likelihood. Test the top-right corner deeply, sample the rest, and write down what you didn't cover. That last part is what makes it a strategy rather than an excuse — and it's the professional version of principle 2.
Where this track ends
You now have the vocabulary and the four core design techniques, and you can write cases and defect reports that hold up in review. That is genuinely the bar for a junior QA role.
Next comes doing it under real conditions: planning, exploratory testing, APIs, SQL, and reporting to people who don't read test cases — the Manual QA Professional track. From there, automation.
Check your understanding
- Name three questions you'd ask about a story in refinement.
- Rewrite "Login should be secure" as a testable acceptance criterion.
- Your sprint is a mini-waterfall. What is the first thing you'd change, and which ceremony does it happen in?
Check your understanding
3 questions. No account needed, nothing is sent anywhere but the grader.
1. Your team builds for eight days and hands everything to QA on day nine. Where is the highest-leverage place to intervene?
2. Which acceptance criterion can actually fail a test?
3. Which belong in a Definition of Done that includes testing?(choose all that apply)