Define · Core
Choose success metrics and guardrails
Decide how you will know the work succeeded, and what must not get worse along the way. Clear metrics make every later decision easier.
About 1-2 daysEvergreenLast reviewed 5 November 2026
Who this applies to
- Skills it builds
- Data and measurement
- PM types
- All
- Product types
- All
- Starting points
- All
- Ways of working
- All
- Experience
- All
- Size of organisation
- All
Steps, by experience
Beginner (under 1 year)
- Start from the outcome in your problem statement. Ask: what would we see if this worked?
- Pick one main metric that shows the outcome, such as task completion, time saved, adoption or retention.
- Add one or two supporting metrics that show progress earlier, such as how many people try the new step.
- Add one or two guardrail metrics: things that must not get worse, such as support calls, errors, satisfaction or cost.
- For each metric, write how it is calculated, where the data comes from and today's baseline.
- Set a target and a date, and agree them with your team and sponsor.
Intermediate (1 to 4 years)
- Define one primary outcome metric with a clear formula and data source.
- Add leading indicators and guardrails.
- Record baselines and set targets with dates.
- Confirm the data can be collected; if not, add tracking to the plan.
Experienced (5+ years)
- Link the primary metric to the product's north star and to company goals.
- Choose guardrails that protect users, quality and the business, and set thresholds that would stop a launch.
- Check that metrics cannot be easily gamed and that they reflect long-term value, not short-term spikes.
- Agree how and when results will be reviewed and who decides what happens next.
Who can do this
Owner: Product manager
Can be delegated to
- Data Analyst or Performance Analyst or Data Scientist / ML Engineer: Define formulas, find baselines and check the data exists
- Tech Lead: Confirm what can be tracked and what new tracking is needed
Reviewed by: Data Analyst, UX Designer, Policy / Legal / Compliance
Tool examples
- Confluence
- Add a metrics table to the problem statement page: name, formula, source, baseline, target, date.
- Amplitude
- Save each metric as a chart now, so you can watch the baseline before anything changes.
- Google Analytics 4
- Check which events already exist for the journey and list the ones you will need to add.
- Looker
- Define the metrics once in a shared model so every report uses the same definition.
- Claude
- Describe your outcome and ask for candidate metrics and guardrails, then ask how each could be gamed.
Definition of done
- One primary metric, supporting metrics and guardrails agreed
- Formula, data source and baseline recorded for each
- Targets and review dates agreed
Common pitfalls
- Measuring activity, such as features shipped, instead of outcomes
- No guardrails, so a win in one place hides harm in another
- Choosing metrics you cannot measure yet without planning the tracking
If you lead the team
What good looks like: Each piece of work has a few clear success metrics and guardrails, set before building starts.
Ask your PM
- What number should move, by how much, and by when?
- What must not get worse while we do this?
Warning signs
- Success is measured as 'shipped on time'.
- Metrics are chosen after launch to fit the results.