For product owners
Quality for product owners
You do not run the tests, and you decide most of what they will find. These are the four places a product owner actually changes quality — and what to do in each.
Where a product owner changes quality
01
Before the work
The criteria decide what gets built and what gets tested.
02
During the sprint
Whether the green pipeline is evidence or decoration.
03
At the release gate
Whether the decision could have gone the other way.
04
After release
What the process missed, and which gate should have caught it.
Three things worth doing this week
Each takes under an hour, needs nobody’s permission, and tells you something you do not currently know about your own product.
01
Grade a story already in your backlog
Not a fresh one. Take something scheduled for the next sprint and see which questions a tester would still have to ask.
Open the grader →02
Unplug the mouse and try to buy something
Keyboard only, from product page to completed order, on your own site. Whatever stops you is both a lost customer and a legal exposure.
The five checks →03
Write the next release's criteria before the build exists
A threshold written while looking at the results is a threshold that will be met. Write them now, with numbers, and name who decides.
How to write them →
Start from the question you have
Is this story ready to build?
A criterion is ready when a tester who has never seen the feature could tell whether it holds. Most are not, and the gap is always the same handful of missing answers.
Guide
Acceptance criteria a tester can't misread
The eight properties that make a criterion checkable, with one story rewritten from 48 to 100.
Free tool
Acceptance criteria grader
Paste a real story and see what a tester would have to ask before writing a single test.
Guide
Definition of ready: training wheels or anti-pattern?
Before you write one: what happens on your team when an item fails it.
Guide
Templates to start from
The criteria template the grader scores, a definition of done, and a go/no-go scorecard, each with a worked example.
What does “done” mean here?
Acceptance criteria say whether this story does the right thing. A definition of done says whether any work is fit to ship at all, and it is the one a deadline will try to borrow from.
Guide
Acceptance criteria vs definition of done
What belongs in each, who owns them, and how to make the second one a pipeline enforces.
What are we legally required to ship?
Accessibility is the quality attribute that now has a date and a penalty attached. The European Accessibility Act's deadline passed in June 2025, and courts have started issuing orders.
Guide
What the European Accessibility Act requires of your checkout
Who is in scope, what enforcement has looked like, and five checks you can run yourself in fifteen minutes.
Free tool
Accessibility report summary
Paste the axe JSON your team already produces to see which findings the standard requires and which are advice.
Can I trust the green pipeline?
A passing suite is evidence only if the tests could have failed. Two things quietly break that: a test that was rerun until it passed, and a test that was weakened until it stopped complaining.
Guide
Flaky tests hide behind retries
Why a suite that needed three attempts has told you something, and it is not that the build is good.
Guide
AI wrote part of the sprint. Who owns the defects?
What the measured quality gap does and doesn't prove, and the one question to ask at review.
Should we release this?
A release decision is only a decision if it could have gone the other way. That needs thresholds written before the results, one named owner, and three possible answers rather than two.
Guide
Go or no-go: the questions to ask before you release
The evidence to collect, criteria that can actually say no, and what every accepted risk has to carry.
What did we miss?
The support queue is the one quality instrument a product owner already owns, and it is usually read as a report on the support team instead of on the product.
Guide
Reading quality from the support queue
Which columns measure the product, which measure the queue, and how to turn a spike into a defect.
Free tool
Escaped defect calculator
Whether this release really let more through than the last one, or the numbers are too small to tell.
Guide
How to write high-quality bug reports
What a developer needs before they can act, for the defects you end up filing yourself.
Free tool
Bug report grader
Score a report against ten checks before it goes to the team.
What this site is not going to tell you
That a number is a target. Escaped defect rates, test counts, and coverage percentages all improve the moment someone is measured on them, and none of them improves the product on the way. They are useful as triggers for a conversation and misleading as goals.
That a tool decides anything. The graders here run pattern matching in your browser, not AI, and they are checklists rather than verdicts — each one can be fooled by something that says the right words about the wrong feature. That is also why they show you every rule they applied.
If you want the engineering side of any of this, the same material is on the QA engineer roadmap, with a practice task and a way to know you are done at every step.