One number that predicts whether customers are getting value. Every guide assumes a product org — this is the version for one person and a few hundred users.
A North Star metric is the single number that best predicts whether your customers are getting value. Every guide to picking one assumes a product org with an analytics team. This is the version for one person with a few hundred users.
One metric. Reviewed weekly. Changed at most twice a year.
The point of a North Star is not measurement. It is that a solo founder makes roughly forty small decisions a week, and a single number resolves most of them without deliberation.
Three properties. A metric missing any one of them will quietly send you in the wrong direction.
It reflects customer value, not company value. Revenue is company value. Invoices sent, reports generated, messages delivered — those are customer value. Revenue follows value with a lag, which is why steering by revenue means steering by a delayed signal.
You can move it this week. A metric you cannot influence within a sprint is a scoreboard, not a steering wheel. That usually rules out anything annual or heavily lagging.
It is hard to game without doing real work. If you can inflate the number by sending more emails or loosening a definition, it will drift away from meaning within a quarter.
The best-known examples come from large companies, so their usefulness is mostly as a pattern rather than a template. The pattern is always the same: a unit of customer work, counted over a period.
| Product type | Candidate North Star | Why it works |
|---|---|---|
| Invoicing / billing | Invoices sent per active account per month | Rises only when the customer is running their business through you |
| Analytics / reporting | Reports viewed per account per week | Viewing is the value; building the report is setup |
| Scheduling | Meetings booked per account per week | Direct proxy for the job being done |
| Writing / content | Documents exported or published per week | Export signals the work was good enough to keep |
| Developer tooling | Successful runs per project per week | Counts completion, not attempts |
| Any product | Weekly active users | Too coarse — logging in is not value |
Notice what those north star metric examples for SaaS have in common. Each counts a completed unit of customer work, each is normalised per account, and each has a time window. Drop any of the three and the number stops being comparable week to week.
MRR is the outcome you want. That makes it a poor steering metric for three reasons.
It lags. A change you make this week shows up in MRR over the following months, filtered through trial length, billing cycles and churn. By the time the number moves you have made twenty more decisions.
It is noisy at small scale. At forty customers, two cancellations and one annual upgrade can swamp a real underlying trend entirely.
And it does not tell you what to do. MRR falling is information you already have. Sessions completed per account falling tells you where to look. Track MRR — just do not steer by it. The metrics worth tracking under $10k MRR covers the full short list.
Four steps. None of them need analytics infrastructure you do not have.
1. Write the sentence your customer would say. "I use this to ___." The verb in that blank is your candidate metric. If you cannot fill it in without hedging, that is a positioning problem before it is a metrics problem — see the positioning framework.
2. Count that verb for your ten best customers. Manually if necessary. You are looking for whether the number separates your retained customers from your churned ones.
3. Normalise it. Per account, per week or per month. A raw total goes up simply because you added customers, which hides the thing you actually want to see.
4. Check the doubling test. If this number doubled next month and nothing else changed, would the business genuinely be healthier? If the honest answer is "not necessarily", the metric is measuring activity rather than value.
Instrument one event before you instrument ten. The most common failure here is building a dashboard first — six metrics arrive, none get acted on, and the exercise gets abandoned. One number, visible every Monday, beats a dashboard nobody opens.
A North Star earns its place in the weekly review, not in a quarterly deck. Look at it once a week alongside three other numbers and let it decide what you work on next — the weekly operating cadence covers the format.
When it falls, the diagnostic order is usually: are new customers failing to activate, or are existing customers using it less? Those have different fixes, and the onboarding framework handles the first.
Change the metric rarely. Twice a year is generous. A North Star that changes quarterly is a preference, and the whole benefit comes from comparing the same number over time.
Below roughly twenty customers, the number is noise. Three people having a quiet week moves it more than any change you shipped. At that stage talking to customers beats measuring them.
It also cannot capture everything that matters. Pricing, positioning and churn type are all invisible to a usage metric, and a rising North Star alongside rising churn is a real and common pattern — the churn reduction framework covers reading that combination.
It is the single number that best predicts whether customers are getting value from your product — usually a unit of completed customer work, counted per account over a week or month. It sits between raw usage and revenue, and it exists to resolve day-to-day decisions without deliberation.
Invoices sent per account per month for billing tools, reports viewed per week for analytics, meetings booked per week for scheduling, documents exported for writing tools, successful runs per project for developer tooling. Each counts completed customer work, normalised per account, over a fixed window.
No. MRR is the outcome you want rather than a steering signal. It lags your decisions by months, it is noisy below a hundred customers, and a fall in MRR tells you something is wrong without telling you where to look. Track it, but steer by a usage metric.
Write the sentence your customer would say — 'I use this to ___' — and count that verb for your ten best customers, manually if needed. Then check whether it separates retained customers from churned ones, and normalise it per account.
At most twice a year. The entire value comes from comparing the same number over time, so a metric that changes quarterly is a preference rather than a North Star. Change it when your product's core job changes, not when the number is disappointing.
Usually not. Logging in is not value, so WAU rises with curiosity as easily as with usefulness. A metric counting a completed unit of customer work is almost always more diagnostic at solo scale.
Tell Marcus what your product does and who uses it. You get one candidate metric and the test to confirm it.
Try GhostCoach free →14-day free trial · cancel anytime · 30-day money-back on Lifetime