Every founder who has run a startup for more than a few months has a metrics dashboard. They know their weekly active users, their month-over-month revenue growth, their customer acquisition cost. These metrics are tracked because they are important, because investors ask for them, and because they go up often enough to sustain motivation. What is less frequently examined is the list of things that are not on the dashboard — the measurements that would answer the most uncomfortable questions about whether the product is actually working. The information a founder chooses not to collect is more diagnostic than the information they do collect, because selective measurement is almost always a way of avoiding the answer to a question they already suspect they will not like.
Startup metrics that go unmeasured are not usually unmeasured by accident. A founder who has not tracked retention by cohort since launch is not unaware that cohort retention is informative. A founder who has not surveyed churned customers is not unaware that churned customers have useful information. A founder who has not calculated revenue per customer conversation is not unaware that this metric exists. The absence of these measurements is a decision, and like most decisions made to avoid discomfort, it produces costs that accumulate in proportion to how long the avoidance continues.
Why founders measure selectively
Selective measurement is a form of motivated reasoning made structural. When a metric is tracked, it produces an output that has to be responded to. A cohort retention analysis that shows 60% churn in the first ninety days requires a response: the team needs to understand what is happening, why, and what can be done about it. That response is difficult, implies the current product is failing at something important, and requires effort to address. Not running the analysis avoids all of these costs. The situation does not improve, but the evidence of the situation does not have to be confronted.
The social dimension of metrics compounds this. A startup’s metrics are not only information — they are a narrative. When a founder shares metrics with investors, co-founders, or the team, the metrics create expectations and frame the story of how the company is doing. A founder who tracks only the metrics that look good is curating a narrative as much as they are monitoring performance. Adding a metric that looks bad adds a complication to the narrative that requires explanation, context, and a credible plan. Leaving it off the dashboard leaves the narrative cleaner, at the cost of operating without the information.
A third dynamic is that founders often do not know how to act on information they suspect will be bad. A founder who suspects that churned customers left because the product fundamentally does not do what was promised faces a product problem, not a retention optimization problem. The response to that information is not a better onboarding email sequence — it is a reassessment of whether the product is doing the right job for the right customer. That reassessment is more difficult than any tactical metric response, which makes the information that would require it easier to leave uncollected.
What the unmeasured metrics usually reveal
The pattern in what founders choose not to measure is not random. It clusters around the questions that would most directly challenge the current direction: whether customers are finding ongoing value, whether the price is right relative to the value delivered, whether the growth is coming from the customer segment that will sustain the business, and whether the problem the product solves is the problem the customer actually has.
Retention by cohort is the most commonly avoided metric in early-stage SaaS. It is avoided because it reveals whether the product is generating ongoing value or whether growth is masking churn. A product with 15% month-over-month growth and 40% monthly churn is not growing — it is on a treadmill that is getting harder to run as the churn compounds. The growth metric looks positive. The retention metric, if tracked, reveals the treadmill. Not tracking retention by cohort is a decision to look at the growth without looking at what is sustaining it.
Revenue quality metrics — revenue per customer type, lifetime value by acquisition channel, net revenue retention broken out by customer segment — reveal whether the business is growing in the segments that will sustain it or in segments that will not. A founder who is growing overall but has not broken down growth by segment does not know whether the growing segment is the profitable one. This information is available. Not analyzing it is a choice that protects the aggregate growth narrative from the segment-level data that would complicate it.
How to audit what you are choosing not to measure
An honest measurement audit begins with a question: what is the thing most likely to be wrong about our current direction, and have we measured it? If the answer to the second part is no, that is the first measurement to add.
-
List every assumption your current product strategy depends on and check whether each is measured. Every strategy assumption has a metric that would validate or invalidate it. “Our customers are finding ongoing value” has retention as its metric. “Our pricing is appropriate for the value we deliver” has willingness to pay and net promoter score broken down by price tier. “Our target customer segment is the one adopting the product” has cohort analysis by acquisition source and customer profile. List the assumptions. Check the metrics. The gaps between the assumptions and the measurements are where avoidance is happening.
-
Add the metric you most dread to your next reporting cycle. Not because it is the most important metric in general — because the dread is the signal that it is the most important metric for you, now. A founder who is dreading the cohort retention analysis is a founder who suspects the retention number is bad. The dread is correct. Running the analysis gives you the information. Not running it gives you the dread without the information.
-
Contact every churned customer from the last ninety days. Not with a survey — with a direct message asking for fifteen minutes. The response rate will be low. The responses you get will be disproportionately informative. A churned customer who agrees to talk is a customer who left for a specific reason they are willing to name. That reason is the product failure that the metrics dashboard was not capturing. Five such conversations will reveal more about the product’s core weaknesses than any volume of in-app analytics.
-
Calculate the metric you have been calling “not yet relevant.” Founders frequently defer metrics as premature: “we don’t have enough customers to do cohort analysis yet,” “our LTV is hard to calculate at this stage,” “we don’t have enough churn to see a pattern.” These deferrals are sometimes correct. They are more often a way of avoiding a metric that would be informative even at small sample sizes. Calculate it anyway. A cohort retention analysis on twenty customers is imprecise. It is not uninformative.
-
Show the full dashboard — including the metrics that look bad — to one trusted advisor every month. The discipline of full disclosure to an external observer removes the incentive to curate the dashboard for narrative management. An advisor who sees both the metrics that look good and the ones that do not can give useful advice. An advisor who only sees the metrics that look good is validating a partial picture and producing feedback calibrated to a version of the company’s situation that is incomplete.
What changes when founders measure honestly
A founder who tracks the metrics they have been avoiding does not get better news. They get more accurate news, earlier. The retention problem that has been compounding for six months is not worse for being discovered — it was already that bad. The discovery just changes when the response happens. Early discovery means the response happens with more runway, more options, and a smaller amount of work built on the wrong assumption. Late discovery means the response happens with less of each.
The habit of selective measurement is not neutral. It produces organizations that are operationally active and strategically blind — generating more data about the things they already understand and less about the things that matter most. The gap in the dashboard is not a gap in reporting. It is a gap in the founder’s model of reality, maintained deliberately to avoid the discomfort of updating that model. Every week the gap persists, the cost of the eventual update grows. The signal was available. The measurement was the only thing that would have made it visible.
