North star metrics and the proxy trap
Teams adopt a north star metric to focus the company. Then they optimize a proxy — signups, sessions, feature usage — that rises while customer value flatlines. The proxy trap is choosing what is easy to measure over what actually matters.
The north star went up fourteen quarters in a row. Weekly active users. The board celebrated. Support tickets about churn reasons mentioned the product never delivered the outcome customers bought it for. The metric moved. Value did not. The team had fallen into the north star metric proxy trap: optimizing a convenient indicator that correlated with success once, until the organization learned to game it.
A north star metric should represent customer value delivered at scale — the single measure that, if it grows sustainably, implies the business model works. Proxies are easier to instrument: page views, signups, time-in-app, features clicked. Proxies become dangerous when mistaken for the star. Goodhart's Law applies: when a measure becomes a target, it ceases to be a good measure.
North star vs proxy: the distinction that prevents dysfunction
| Type | Definition | Example (B2B SaaS) | Risk |
|---|---|---|---|
| North star | Customer value × reach at sustainable unit economics | "Projects completed per active team per month" | Harder to define; slower feedback |
| Input metric | Leading indicator of north star | "Teams that complete onboarding checklist" | Useful if validated against star |
| Proxy | Correlated metric chosen for ease | "Weekly active users" | Optimized independently of value |
| Vanity | Looks good in decks | "Total registered accounts" | Actively misleading |
The north star answers: if this number grows healthily, are customers better off and is the business stronger? A proxy answers: can we ship a dashboard update this sprint? Both questions matter. Conflating them causes teams to ship engagement hacks that inflate WAU while core job-to-be-done stalls.
A rising proxy with flat retention is a warning, not a win.
Hybrid PLG replaced pure PLG when signup volume stopped predicting revenue — the proxy (signups) diverged from value (activated teams paying). North star discipline catches that divergence earlier if retention and outcome metrics sit beside the headline number.
How teams fall into the proxy trap
Instrument what is easy. Analytics tools count events. Value delivery often requires qualitative follow-up or delayed outcomes — "did this integration save the team hours?" Hard to auto-track; easy to track "integration clicked."
Investor and board pressure for simple charts. One number on a slide. WAU fits. "Successful workflows per customer" needs definition work.
Local optimization. Growth optimizes signups. Product optimizes feature adoption. Neither owns downstream retention. Each proxy rises; north star flatlines.
Lagging north star frustration. Value metrics move slowly. Teams substitute proxies for quarterly goals because proxies respond faster — then forget to reconcile.
Survivorship in correlation. Early cohorts that signed up during a launch campaign also had high intent. Signup count correlated with success by selection, not causation.
Choosing a north star that resists proxy drift
A defensible north star shares properties:
1. Customer-centric outcome
Describes value received, not company activity. "Messages sent" is activity. "Conversations resolved" is closer to outcome — if resolution means customer success.
2. Measurable with agreed definition
Ambiguity invites proxy substitution. Write the definition: what counts, what excludes, over what window.
3. Leading enough to steer
Pure lagging revenue is a scoreboard, not a steering wheel. Balance with input metrics validated against the star.
4. Hard to game without delivering value
If the metric rises when users are nagged into empty actions, it is a proxy. Minimum viable proof for product bets should tie experiments to north star movement, not proxy spikes.
5. Stable across product evolution
Features change; job-to-be-done should not. Refine measurement, not the underlying value thesis, unless strategy pivots.
Example reframes:
| Weak proxy | Stronger north star candidate |
|---|---|
| Daily active users | Tasks completed successfully per active user per week |
| Signups | Activated accounts (completed core workflow once) |
| Time in app | Time to outcome for primary use case |
| NPS alone | NPS + retention cohort on outcome metric |
None is universal. The exercise is forcing the question: what would customers pay for again?
Input metrics and guardrails complete the system
One number is insufficient. Healthy measurement stacks three layers:
North star (value × reach)
↑ validated by
Input metrics (2–4 leading indicators)
↑ bounded by
Guardrail metrics (things that must not break)
Guardrails prevent proxy gaming: support ticket volume, churn rate, error rates, gross margin per active customer. If north star rises but guardrails break, pause and diagnose.
Input metrics require periodic validation — quarterly review: do input metric gains still predict north star gains? When correlation breaks, demote the input metric before it becomes a rogue proxy.
What makes a north star metric actually useful?
These questions separate stars from slide decoration.
How is a north star different from a KPI?
KPIs are operational health checks — many per function. North star is the one outcome metric the whole company aligns on. KPIs support the star; they do not replace it.
When should a team change its north star?
When strategy pivots — new customer segment, new business model, proof the current star no longer predicts revenue or retention. Not when the star is inconveniently flat while proxies look good.
Can a company have more than one north star?
Usually one per product line or business unit. Multiple "north stars" at the same level reintroduce proxy wars between teams optimizing different headline numbers.
A common argument runs the other way
The opposing view holds that north star metrics oversimplify complex businesses — that functional teams need their own targets and a single star creates perverse incentives at the edges.
Functional KPIs are necessary. The north star coordinates them toward a shared definition of customer value. Without it, each function optimizes local proxies that sum to dysfunction. Simplification is the point — not simplicity of measurement, but simplicity of alignment.
Oversimplification is the proxy trap, not the north star concept itself. The fix is better stars and guardrails, not more headline metrics.
Key takeaways
- North star metrics represent customer value at scale; proxies represent what is easy to count.
- Goodhart's Law: optimizing a proxy decouples the metric from value.
- Validate input metrics quarterly — correlation breaks silently.
- Guardrails (churn, support load, margin) catch gaming before it compounds.
- Define the north star precisely; ambiguity invites proxy substitution.
- A rising proxy with flat retention is a diagnostic, not success.
Conclusion
The proxy trap is not dishonesty — it is measurement gravity pulling teams toward what dashboards already show. Escaping it requires naming the star explicitly, writing its definition, pairing it with guardrails, and rehearsing the uncomfortable review when proxies green and outcomes flatline.
The next planning cycle should include one question: if our north star doubled, would customers unquestionably be better off? If the honest answer is unclear, the metric on the slide is probably a proxy.