success criteria
Also called: success criterion
The checks that say whether a project worked, written before it starts. Useful ones have two parts: what must be delivered by when, and what must change for the users or the business afterwards.
Success criteria are the checks, written before a project starts, that say whether it worked. Written afterwards, they tend to describe whatever happened.
Useful criteria have two parts. Delivery says what must exist by when and within what money: the date, the budget, the scope. Result says what must change for users or the business afterwards, measured on a set date. Delivery alone is a trap: a team can hit every delivery criterion on time and on budget while nothing changes for anyone. The result part keeps the goal in view.
Each criterion needs a number, a date, a way to measure it and someone who checks it. "Families like physics" is a hope; "120 packages bought by 28 February, checked on 1 March" is a criterion. A result criterion is often measured after the build is finished, so the check date belongs in the brief from the start, or nobody looks.
Success criteria judge the whole project. Acceptance criteria judge one deliverable, and the goal explains why the criteria matter. When a result falls short, the criteria are the start of a conversation about what to change, not a verdict on the people who did the work.
Example
Klasna's physics launch had two-part criteria in its brief:
- Delivery: open on 1 February 2027 with at least 8 tutors who passed a demo lesson, within the 600,000 UAH ceiling.
- Result: 120 physics packages bought by 28 February 2027.
The check on 1 March:
Every delivery line is green, and the result is under half the target. Without the result line the launch would have been reported as a success. The plan's own funnel shows where to look: it assumed 2,500 visits × 40% starting the diagnostic test × 75% finishing it × 16% buying = 120. Traffic came in at 2,620 and 1,060 started the test (40.46%), but only 472 finished it (472 ÷ 1,060 = 44.53%, not 75%), and 58 of the 472 bought (12.29%, not 16%).
Common mistakes
- Only delivery criteria. On time and on budget says nothing about whether anything changed.
- A result with no number or date. "Learners are happy" can't be checked; say what, how many and by when.
- No check date in the brief. If nobody books the look-back in advance, the result is never measured.
- Writing or moving the criteria afterwards. Criteria set after the facts describe the facts.
- Reading an early, small number as final. A first-month figure from a few dozen people is a signal to follow, not a verdict.
Learn it in the course
- Goal or outputs: what success means · Why finished outputs aren't success, and how to write two-part success criteria: delivery and result.
- Case: the brief for the physics launch · Turn Leo's list of outputs into the one-page brief he will sign: a goal that names the change, two-part success criteria, an out list and assumptions you can test.
- Changes without a committee, and a decision log · Route every change through one door, write a short impact note in days and money, let a named person decide, and record each decision in a few lines.
- The benefits check and a blameless retrospective · Check the result weeks after launch against the success criteria, run a retrospective on facts and decisions, turn lessons into owned actions and close with a one-page report.
- The one-page brief · Goal, success, scope in and out, limits, testable assumptions, names and a change rule: one page a sponsor can sign.