Metric Theater: When Your Dashboard Lies to Everyone at Once
DAU up. Churn down. Everyone nods. Meanwhile a user is quietly switching to a competitor. That gap has a name — Metric Theater — and it's killing your product.
From Product Theory: The Hidden Forces That Shape User Behavior — 40+ short chapters on why users behave the way they do.
The Monday metrics meeting. Everyone presents their numbers.
DAU up 3%. Activation holding at 42%. Churn down half a point. Seven-day retention looks solid. Every graph slopes up and to the right. Everyone nods. The meeting ends on time. People go back to their desks feeling good.
Nobody mentions the support ticket from Friday. A user wrote: "I've been trying to export my data for three weeks. Nothing works. I'm switching to a competitor."
The numbers said everything was fine. The user said it wasn't. That gap has a name.
Metric Theater
Metric Theater is what happens when a team starts performing for its dashboard instead of listening to its customers. The weekly meeting stops being a diagnosis and becomes a show.
You learn the choreography without noticing. If retention looks bad this week, you zoom out to the monthly view. If the month looks bad, you segment by user type and present the one cohort that's growing. If nothing looks good, you quietly swap the metric for a friendlier one. You compare this month to the worst month, never the best. You pick the graph that goes up.
None of it is lying, exactly. It's worse than lying. It's a room full of smart people talking themselves into a story they already wanted to believe.
When you perform for the dashboard instead of the customer, the numbers look good but the product doesn't work.
How I learned to see it
I watched this happen at BuildBetter. Our dashboard showed usage climbing — more queries per user, higher engagement scores, longer sessions. The graphs looked great. We felt great.
Then I sat down and watched a customer actually use the thing.
Within two minutes she was confused. Within five, frustrated. Within ten, she gave up and went back to her old workflow. "The search doesn't work the way I think," she said. "I keep trying to ask it questions, but it just shows me random stuff."
Our dashboard said engagement was up. She said the product was broken. Both were true — and that's the trap. She was clicking around because she was lost. Her confusion registered as engagement. We had built a metric that rewarded us for a broken experience, and we were performing for it in every meeting.
The tell is obvious once you see it
Here's the fastest diagnostic I know: if your metrics meetings never produce bad news, you're watching theater.
Real measurement is uncomfortable. It surfaces problems. It shows you things you don't want to see. You look at a graph and think that can't be right — so you dig in, and it turns out it is right, and now you have to change something.
Theater is comfortable. Everything confirms the story you already told. The product is working. Users love it. Growth is coming. The meeting ends on time and nobody's stomach hurts.
| Real measurement | Metric Theater |
|---|---|
| Surfaces problems you didn't expect | Confirms what you already believed |
| Someone leaves the room with homework | Everyone leaves feeling reassured |
| Timeframes chosen honestly | Timeframes chosen to flatter |
| Ends in a question | Ends on time |
Building for the dashboard kills the product
Once the dashboard becomes the audience, you start building features to move metrics instead of solving problems. "This will increase activation" becomes the justification for everything — never mind whether it helps anyone, never mind the complexity it adds. It'll move the number.
I've watched teams bolt on pop-ups, notifications, gamification, streaks, badges — anything to juice the graph. Activation goes up. The product gets worse. And here's the thing nobody says out loud:
Users don't want to activate. Nobody wakes up wanting to hit your activation milestone. They want to accomplish something. Activation is just your proxy for whether they're getting value. The moment you optimize the proxy instead of the value, you snap the wire between them — and now your number can go up while your product falls apart underneath it.
Rule of thumb: If you can't explain your metric in terms of user value, you're measuring the wrong thing.
The antidote is pairing numbers with people
The fix is embarrassingly simple. Never let a number travel alone.
- For every graph, read one user quote.
- For every dashboard, watch one session recording.
- For every metric that moved, find out what a real person said about it.
Numbers tell you what. People tell you why. You need both, and the order matters less than the pairing.
When we rebuilt our metrics meetings at BuildBetter, we made one rule: you cannot share a metric without sharing a user story that explains it.
DAU is up? Show me a message from someone who just started using it daily, and tell me what changed for them. Activation is down? Show me a session recording of someone who bounced, and point to where they got stuck.
This kills Metric Theater on contact. You cannot perform for the dashboard when the price of admission is actual user evidence. The meetings got longer. They got uncomfortable. They got much, much better.
What we only saw because we looked
Pairing the numbers with people surfaced a pattern we'd been staring past for months. The dashboard said our onboarding was steadily improving. The session recordings told a different story: people were skipping onboarding entirely and going straight to search.
They weren't "failing to activate." They already knew what they wanted to do and didn't need the tour. We had been quietly congratulating ourselves on onboarding gains that were mostly noise — and we would have kept optimizing that flow forever if all we ever did was watch the graph.
The limit: sometimes you have to trust the numbers
Don't overcorrect into anecdote worship. At scale you cannot watch every session or read every ticket. You need aggregated data to spot trends a single conversation would never reveal. The point was never to throw out quantitative measurement.
The point is to stop performing for it.
Use dashboards to find problems. Use people to understand them. When a metric looks weird, resist the reflex to explain it away — go look at what actually happened. The dashboard is a starting point, not an ending point. It tells you where to point the flashlight, not what you'll find.
Rule of thumb: Metrics meetings that never lead to user research are just theater.
If it made your last metrics meeting look a little different — or made you want to go watch one session recording before the next one — that's the whole point. Subscribe to Context Limit and I'll send the next idea along when it's ready.
