The Numbers That Tell You the Truth
Every product team tracks metrics. Most track too many. Dashboards fill with charts that nobody acts on. Numbers go up and down with no clear connection to whether the product is working. The team feels analytical without being analytical, and meaningful decisions get made on instinct anyway.
Good metrics are different. They tell the team whether the product is succeeding, in language the team can act on. They are few enough to remember. They connect to user behaviour and business outcomes. They get reviewed regularly. They shape decisions, not just dashboards.
This article is about choosing metrics that matter. We will cover the difference between vanity and value metrics, the categories of metrics most products need, and the practices that turn data into decisions.
Vanity Metrics vs. Value Metrics
Some metrics look impressive but tell you almost nothing about whether the product is working. These are vanity metrics. Other metrics are less impressive but tell you what is actually happening. These are value metrics. The distinction matters because vanity metrics are seductive.
Vanity Metrics
- Total registered users. Goes up forever. Tells you nothing about whether anyone uses the product. Includes every account ever created, including ones that signed up once and never came back.
- Total page views. Easy to inflate with marketing spend or viral content. Doesn't tell you whether users find what they need.
- Total downloads (for apps). Includes downloads that were never opened. Doesn't track usage.
- Followers, likes, mentions. Social media metrics that may correlate with awareness but rarely correlate with business outcomes.
- Press coverage and award mentions. Feel good. Don't move user behaviour or revenue.
Value Metrics
- Active users in a meaningful time window. Daily, weekly, or monthly active users, depending on how frequently the product is meant to be used. Tells you who actually uses it.
- Retention. What percentage of users come back. Tells you whether the product is actually valuable to people who tried it.
- Conversion rates. What percentage of users complete a meaningful action (signup, first use, purchase). Tells you whether your funnel works.
• Revenue per user, customer lifetime value, gross margin. Tell you whether the business
model is healthy.
- Churn rate. What percentage of users leave each period. Combined with growth, tells you whether you are really growing or just running on a treadmill.
How to Spot the Difference
Eric Ries (author of The Lean Startup) describes vanity metrics as those that go up and to the right but don't help you make decisions. Value metrics are actionable: when they change, you know what to do. The test is action: does this metric guide a decision? If not, it is decoration.
The Categories of Metrics Most Products Need
A useful metrics framework covers the main aspects of how users move through and interact with your product. Most products need metrics in each of these categories. The specific metric in each category will vary by product, but the category is universal.
Acquisition
How do users find and arrive at the product? New signups per week, traffic by channel, marketing spend efficiency. Tells you whether the top of the funnel is healthy.
Activation
What percentage of users reach the moment of meaningful value? First completion of the core action, time to first value, percentage of new users who complete onboarding. Tells you whether new users actually start using the product.
Engagement
How deeply do users engage with the product? Daily active users, sessions per user per week, depth of use (how many features used, how often). Tells you whether the product is actually useful to people who reached it.
Retention
Do users come back? Day-1, day-7, day-30 retention. Cohort retention curves. Tells you whether the product is valuable enough to make users return.
Monetisation
How much revenue does the product produce? Conversion to paid, average revenue per user, expansion revenue, churn rate. Tells you whether the business model is working.
Referral
Do users tell others? Net promoter score, viral coefficient, share of new users from referrals. Tells you whether the product is good enough to spread.
Category Question Answered
Acquisition Are users finding us?
Activation Are new users reaching value?
Engagement Are users using us deeply?
Retention Are users coming back?
Monetisation Are we making money?
Referral Are users telling others?
If you have a metric in each of these categories, you have broad coverage. Many teams over-index on acquisition and engagement and under-index on retention and referral. The ones they neglect are usually the ones that matter most for long-term success.
Choosing Specific Metrics
Within each category, you need to pick specific metrics. The choice depends on your product, your business model, and your stage. A few rules help.
Rule One: Match Frequency to Use
If your product is meant to be used daily (a chat app, a fitness tracker), track daily active users. If weekly (a team tool), weekly. If monthly (a tax tool), monthly. The metric should match how often the product is meant to matter. Tracking DAU for a product used quarterly produces tiny numbers that move randomly.
Rule Two: Define "Active" Strictly
Active should mean something specific, not just logged in . A real estate app where active means opened the app tells you less than one where active means searched for properties . Define active as the behaviour that represents real value, then track that.
Rule Three: Track the Critical Action
Every product has one or two actions that represent the core value. Sending a message in a chat app. Publishing a post in a writing tool. Closing a deal in a CRM. Track the rate at which users take this action; it is more meaningful than generic engagement metrics.
Rule Four: Use Cohort Views
Single-number metrics hide a lot. Retention is forty percent averages across all users. Cohort views (where you track each batch of users separately over time) reveal whether retention is improving or worsening for new users. Cohorts are more useful than averages for diagnosing trends.
Rule Five: Combine Counting and Behaviour
Some metrics count things (active users, transactions). Others measure behaviour (time to value, depth of use). The combination is more useful than either alone. Counting tells you scale; behaviour tells you quality.
Leading vs. Lagging Indicators
Some metrics tell you what already happened. Some tell you what is likely to happen next. Both are useful; mixing them up causes confusion.
Lagging Indicators
Outcomes that have already occurred. Revenue last quarter. Churn last month. Customers acquired last year. Lagging indicators tell you whether the strategy worked. They are definitive but slow; by the time they move, the cause is in the past.
Leading Indicators
Behaviours that predict future outcomes. Number of demos scheduled (predicts future deals closed). Number of users who completed onboarding (predicts future retention). Engagement depth in the first week (predicts long-term retention). Leading indicators tell you what to expect; they move before lagging ones.
Why Both Matter
If you watch only lagging indicators, you only know what happened after it is too late to change. If you watch only leading indicators, you know what is likely but can't see whether it actually played out. The combination tells the full story: leading indicators tell you what to expect, lagging indicators confirm whether the expectation was right.
Common Metric Mistakes
Mistake One: Tracking Too Many
Dashboards with fifty metrics. Nobody can remember them all. Nobody knows which to act on. Decisions get made on instinct anyway because the metrics are too many to synthesise. The fix is to pick the few metrics that matter most and put them where everyone can see them. Other metrics can exist in deeper dashboards for diagnosis, but they shouldn't be the headline.
Mistake Two: Confusing Correlation With Causation
Two metrics moving together doesn't mean one caused the other. Maybe both were caused by something else. Maybe it was coincidence. Drawing conclusions from correlation alone produces false confidence. Where possible, run experiments to test causation directly.
Mistake Three: Ignoring Variability
A metric that goes up ten percent might be real change or might be noise. The variability of the metric matters. Daily numbers are noisier than weekly. Small samples are noisier than large. Without thinking about variability, teams chase noise and ignore real signals.
Mistake Four: Single-Number Thinking
Reducing the product to one metric hides a lot. Active users grew ten percent sounds good but might mean new users grew while existing users churned. The single number needs to be decomposed. Cohort views, segment views, and supporting metrics tell the real story.
Mistake Five: Optimising Without Asking Why
The metric went down. Let's ship something to bring it back up. But why did it go down? Was it seasonal? Was it a deeper problem? Without understanding the cause, the optimisation may make things worse. Diagnose before treating.
Mistake Six: Goodhart's Law in Action
When a measure becomes a target, it ceases to be a good measure. Once the team optimises for a specific metric, they often hit the metric in ways that don't actually improve the underlying thing. Tracking support response time can lead to fast-but-unhelpful responses. Tracking monthly active users can lead to engagement loops that boost the number without making the product better. Be careful about what you optimise.
Building a Metrics Practice
Choosing the right metrics is half the work. The other half is using them. A metrics practice has several elements that produce consistent value over time.
A Weekly Metrics Review
Fifteen to thirty minutes per week, the team looks at the key metrics together. Not to generate reports; to spot what changed, what surprised them, what looks worth investigating. The discussion is more valuable than the numbers; it builds shared understanding of what is happening.
A Quarterly Deep Dive
Once a quarter, do a longer review. Look at trends. Slice by segment. Compare cohorts. Update dashboards if metrics have shifted. The deep dive surfaces things weekly reviews miss.
Tie Metrics to Decisions
When making product decisions, name the metric they are meant to affect. This change should improve activation by reducing time to first value. Then, after shipping, check whether it actually moved. The discipline of explicit predictions builds team intuition over time.
Be Honest About Misses
Some predictions will be wrong. Some shipped features will fail to move metrics. The honest response is to learn from the miss, not to retroactively change the goalpost. Teams that punish missed metrics produce sandbagged predictions; teams that learn from misses produce better intuition.
Resist Adding Metrics Reflexively
Every time something goes wrong, the temptation is to add a new metric to track it. Over time the dashboard becomes huge. Resist this urge. Most situations are diagnosed by existing metrics, with deeper investigation as needed. New metrics should clear a high bar.
Different Stages, Different Metrics
The metrics that matter most depend on the stage of the product. Early-stage products have different priorities than mature ones.
Pre-Product-Market-Fit
Focus on retention and engagement quality among early users. Don't worry much about acquisition or revenue scale. The question is whether the small group who has tried the product loves it. If not, scaling won't help.
Post-Product-Market-Fit, Pre-Scale
Focus shifts to acquisition and activation. The product works for some users; the question is how to find more like them efficiently. Channel-level acquisition metrics, activation rates, and time-to-value matter most.
Scaling
Acquisition is at scale; the focus shifts to monetisation, retention, and unit economics. Are we making money on the users we acquire? Are we keeping them long enough to be profitable? Cohort metrics, lifetime value, and gross margin matter most.
Mature
Growth has slowed; the focus is on expansion within existing customers and new product lines. Net revenue retention, expansion rate, and feature-level engagement matter. The challenge is finding new growth vectors when the original ones plateau.
A Final Word
Metrics are the language in which the product speaks back to the team. The clearer the language, the better the team understands what is happening. The fewer the metrics, the easier to focus. The more honest the metrics, the more trust the team can place in them.
If you take one practice from this article, take this: pick three to five metrics that matter most for your product right now. Put them somewhere everyone sees. Review them weekly. Make decisions based on them. After a quarter, evaluate whether they were the right metrics; adjust if not. Over time, the discipline produces sharper instincts about what numbers tell you the truth and which are decoration.
Key Takeaways
- Vanity metrics look impressive and don't guide decisions. Value metrics may be less impressive but tell you what is actually happening.
- Cover the categories: acquisition, activation, engagement, retention, monetisation, referral. Most teams under-index on retention and referral.
- Combine leading indicators (predictive behaviours) with lagging indicators (outcomes). Leading metrics show what to expect; lagging confirms what happened.
- Few metrics, watched closely, beat many metrics watched casually. Three to five headline metrics are usually enough.
- Tie metrics to decisions. Make explicit predictions when shipping changes; check honestly whether they came true. This builds intuition over time.