Counter metrics: The guardrails I’d put next to a north star
Our reviewers evaluate writing services independently. Learn how we stay transparent, our methodology, and tell us about anything we missed.
North star metrics create focus. Humans also love making the glowing number go up—even when it quietly breaks something else.
That's why I care about counter metrics: measurements that catch over-optimization before it becomes a postmortem.
What a counter metric is
A counter metric is what you watch to ensure gains on the north star didn't come at the expense of customers or the business.
If the NSM is "messages sent," counters might be spam reports or mute rates. If the NSM is "trial starts," a counter might be activation quality or day-7 retention. If the NSM is "checkout conversion," watch refunds, chargebacks, or support volume.
Why you need one
Without counters:
- Teams ship dark patterns that inflate short-term conversion
- Load increases without quality
- One surface wins while another collapses
- Leadership celebrates a spike that predicts churn
With counters, you can say: "we moved the NSM and stayed inside the guardrails."
How I'd choose a counter metric
- List failure modes of gaming the NSM
- Pick the failure mode that would hurt most
- Find a measurable proxy you can track on the same cadence
- Prefer leading indicators when you can (support tickets lag; error rates or completion quality may lead)
- Keep the set small—one or two counters beat a second dashboard of noise
Good counters are as discussable as the NSM. If nobody can explain them, they won't protect you.
Setting thresholds
A counter without a threshold is a vibe. I'd set:
- A red line (stop / roll back)
- A yellow band (investigate, slow experiments)
- Context for seasonality and launches
Example: "refund rate may not rise more than X relative while we push conversion experiments."
Thresholds should be boring and pre-agreed—not negotiated after a "winning" test.
For experiment guardrails, I spell out the primary and counter measures in an A/B testing for product managers plan. When the tradeoff touches margin or payback, I also pressure-test the bet with unit economics for product managers. If the guardrail is retention, I also look at churn prevention for product managers; pricing bets deserve the same discipline with pricing strategy for product managers.
Pairing NSM + counter in rituals
| Ritual | What I'd do |
|---|---|
| Experiment design | State expected NSM lift + counter constraints |
| Launch review | Report both, not the vanity one |
| Weekly metrics | Same definitions, same window |
| Strategy refresh | Revisit whether the pair still matches the bet |
My take
A north star without a counter is how good teams ship bad incentives. Pick the guardrail as carefully as the goal.
For more PM analytics craft, the Product Manager Certification is a solid path—and the newsletter keeps these notes practical.
