What product metrics actually matter?
The product metrics that matter are the ones you can tie to a decision and an outcome. Activation, cohort retention, feature adoption weighted by the revenue behind it, and time-to-value pass that test. If a number goes up and nothing about your roadmap changes, it is decoration, not a metric.
Most dashboards fail this test. They are full of numbers that feel like progress — signups this week, total registered users, pageviews — and short on numbers that would actually redirect a sprint. This is a field guide to the difference: which metrics earn a place on the wall, which ones to quietly retire, and why the hardest part is usually not choosing the metric but connecting it to the customer and the money behind it.
The one test every metric has to pass
Before you argue about definitions, ask one question of any metric: if this moves, does something I do change?
A good metric is a decision waiting to happen. Retention drops in the week-two cohort, so you look at onboarding. Adoption of the feature you shipped last quarter is flat among your paying accounts, so you either fix it or kill it. Time-to-value creeps up, so you know activation is breaking somewhere in the first session.
A vanity metric is one that moves and leaves you with nothing to do. Total signups went up — fine, but you can’t tell whether those people activated, stuck around, or paid, so the number can’t point at a next action. It flatters the deck and starves the roadmap.
That is the whole framework. Everything below is an application of it.
Vanity metrics, and what to track instead
Here is the swap most teams need to make. The left column is the number that feels good in a status update; the middle is the number that answers an actual product question; the right is the question it answers.
| Drop this vanity metric | Track this instead | Because it answers |
|---|---|---|
| Total registered users | Activation rate (first value reached) | Are new users getting to the point? |
| Weekly signups | Signup → activation → retained funnel | Where do we lose people? |
| Pageviews / sessions | Retention by cohort | Does the product keep its promise over time? |
| Total feature usage | Feature adoption by paying account | Did what we shipped move the accounts that pay us? |
| Raw NPS score | Revenue-weighted NPS + linked feedback | Which unhappy customers actually put money at risk? |
| App-store rating | Time-to-value (first session) | How fast does a new user get the point? |
None of the left-column metrics are lies. They are just answers to questions you were not asking. “How many people signed up” is a marketing question wearing a product costume.
The four that earn their place
Activation. The single most predictive early metric. Define one honest “aha” event — the first time a user gets the thing they came for — and measure the rate at which new users reach it. Almost everything downstream (retention, expansion, referral) is gated by this one number.
Retention by cohort. Not “monthly active users,” which blends your strong week-one crowd with people who churned six months ago and hides the trend. Cohort retention shows whether the product still delivers in week four, week twelve, week fifty-two. A flat or smiling curve is the closest thing to product-market fit you can put on a chart.
Feature adoption weighted by revenue. Total clicks tell you a feature gets used. Adoption among the accounts that pay you tells you whether the last quarter of engineering moved the business. These are different numbers, and the second one is the one that should decide what you build next.
Time-to-value. How long from first touch to first real outcome. It is the leading indicator sitting behind activation, and the number most directly in your control — a faster path to value lifts nearly everything else.
Instrument for the decision, not the dashboard
A practical corollary: when you add tracking, start from the decision and work backward, not from “what can we log.” Every event should map to a question someone will act on. If you can’t name the decision an event feeds, you are collecting exhaust, and exhaust is where vanity metrics come from. Fewer, sharper events beat a firehose you never query — and they keep your analytics legible six months later, when the person who added the tracking has moved on.
Why the metric’s location matters as much as its definition
Here is the part most “top metrics” lists skip. Choosing the right metric is the easy half. The hard half is that most product metrics live in one tool, revenue lives in a second, and customer feedback lives in a third — so tying a number to a decision means exporting three CSVs and hoping the IDs line up.
That gap is expensive. Context-switching costs an estimated $450B a year, and the average employee loses about 40% of productive time to it (Gallup/TheTab). For product teams, a real slice of that is the tax of reconciling metrics that were never on the same record.
A metric only matters if you can act on it — and acting on it is easy when the usage number, the revenue behind it, and the customer who generated it sit on one record, and painful when they live in separate silos. This is the wedge AIOProductOS is built on: first-party product and web analytics, revenue-weighted funnels and retention, on the same spine that joins usage, revenue, and feedback per customer. Every shipped feature carries a verdict — adoption, MRR adopted, retention lift — on its own task card, so the metric and the decision are never two tools apart. If you are still assembling that picture by hand, our rundown of the best product analytics tools walks the options honestly.
When a “vanity” metric is actually worth watching
The honest counter-case, because the framework has real edges.
Total signups during a launch. For the week around a Product Hunt post or a big announcement, raw signup count is a legitimate pulse — it measures reach, and reach is exactly what a launch is buying. Just don’t let it stay on the dashboard after the spike fades.
Pageviews for a content- or SEO-led surface. If organic traffic is your acquisition engine, pageviews and their trend are a real input metric, not vanity. The mistake is importing that logic into the product, where a pageview tells you almost nothing about value delivered.
Raw user count for a fundraising narrative. Investors read total users as market pull, and that framing is fair on its own terms. It is a story metric, not a build metric — useful in the deck, useless in sprint planning. Keep the two audiences separate and you can serve both without fooling yourself.
The rule underneath all three: a metric is vanity when it is used to answer a question it can’t. The same number can be a real signal in one context and pure decoration in another. Label the context, and you will know which one you are holding.
The short version
Track fewer numbers, and make each one earn its place by pointing at a decision. Activation, cohort retention, revenue-weighted adoption, and time-to-value will carry most product teams further than any dashboard full of totals. And remember the metric is only half the job — the other half is having the revenue and the customer sitting right next to it, so the number can actually change what you do next.
See how the product analytics module ties every metric to the customer and the revenue behind it →