Different stages need different numbers. Measuring everything is the same as measuring nothing.
12 min read · Level 2
A dashboard with forty numbers changes no behaviour, and the ones that survive are usually the ones that go up regardless of whether the business is working. The test for any metric is brutally simple: if this doubled, would I do anything differently? If not, it is decoration that costs attention. Cumulative totals are the worst offenders — total signups ever rises steadily while a company dies.
What matters changes with stage.
Early (pre-PMF): users, activation, retention, and the number of customer conversations you had. Growth (post-PMF): MRR, CAC, LTV, churn, NRR, gross margin, burn, runway.
Everything else is context, not a dial you steer by.
Because a dashboard with forty numbers on it changes no behaviour. Nobody looks at it after the second week, and the ones that do get looked at are usually the ones that go up reliably regardless of whether the business is working.
The test for any metric: if this number doubled, would I do anything differently? If not, stop reporting it.
A pre-PMF company measuring badly: cumulative signups, page views, social followers, app downloads, press mentions. All rising. All meaningless — every one can be increased without a single person getting value.
The same company measuring well:
These numbers are worse-looking and far more useful. The 8% activation says exactly what to work on, and the 4 conversations says the founder is not talking to enough people.
One of these dashboards makes you feel good. The other tells you what to do on Monday.
First-time founders often track cumulative totals — total signups, total users ever — which only go up and therefore contain no information. Use active, current and rate-based numbers instead.
The second mistake: adding metrics without removing any. A metric added is a metric someone must maintain and read. Five that drive decisions beat forty that decorate a screen.
The third: measuring before there is anything to measure. With twelve users, talk to all twelve.
If this number doubled, would I act differently? If no, remove it.
Weekly active, not cumulative signups. Retention rate, not total registered.
Five numbers you check weekly and can recite from memory. More than that and none of them get real attention.
Pre-PMF: activation, retention, conversations. Post-PMF: the growth and money numbers.
'If activation falls below 20%, we stop building features and fix onboarding.' A metric with no attached decision is decoration.
From your first users, reviewed whenever your stage changes.
Do not build analytics infrastructure with under about thirty users. A spreadsheet updated weekly is genuinely enough and far faster.
Registered users, page views, social followers and press mentions share two properties: they only go up, and they can be increased without anyone getting value. They persist in reporting because they are the numbers that make a bad month feel survivable.
Replace every cumulative total with an active or rate-based equivalent. The discomfort of that change is the point of making it.
The practice that separates a useful dashboard from a decorative one is writing, next to each number, what you would do if it moved. "If activation falls below 20% we stop building features and fix onboarding" turns a metric into a trigger.
Five metrics, each with a written decision attached. Any number without one comes off the list.
Create a free account to read the rest and apply it to your own startup.