How to pick the one metric that matters in your first year
Too many numbers, not enough attention. A one-hour workshop to find the number that tells you the product works.
On this page
Every startup has too many numbers and not enough attention. In the first year, the most useful thing you can do with your analytics is ignore almost all of it and pick one number that tells you whether the product is working.
This guide explains how we choose that number with the teams we work with, and the mistakes we see most often.
Why one number
A dashboard with thirty charts feels like control. In practice it is a way to avoid deciding what matters. When every number is on the screen, there is always one going up, and there is always a reason to feel fine.
One number forces a conversation. It makes the team agree on what "working" means. It makes trade-offs visible: if a change moves the number, it was a good change, and if it does not, it was not, however nice it looked.
It also makes meetings shorter. That alone is worth it.
What makes a good first-year metric
A good number for your first year has four qualities.
- It measures value, not activity. "Logins" measures activity. "Projects with at least one report shared" measures value. People log in for many reasons; they share a report because it was useful.
- It moves weekly. If it only changes once a quarter, you cannot learn from it fast enough.
- The team can move it. Revenue is important, but in month three it depends on too many things outside the product.
- It is hard to fake. If a pop-up can double the number overnight, it is the wrong number.
A good metric is one where, if it goes up and nothing else changes, you would still be happy.
Maya Chen
A simple way to find it
We run this as a one-hour workshop. You need the founders, whoever talks to customers most, and a whiteboard.
Step 1: write the moment of value
Finish this sentence: "A customer gets real value from our product when they …". Write down every answer, then argue until you have one. For Tally, it was "when they make a decision based on a chart".
Step 2: find the event closest to that moment
You probably cannot measure "made a decision". But you can measure something close to it. For us, it was sharing a chart or a report with someone else. People do not share charts they do not believe.
Step 3: make it a rate
Raw counts grow when you add users. A rate tells you whether the product is getting better. Our number became "the share of active teams who shared at least one chart this week".
Step 4: check it against your best customers
Look at the customers you would most like to clone. Do they score high? If your best customers score low, the number is measuring something else.
Examples
| Product | Weak metric | Better metric |
|---|---|---|
| Analytics tool | Weekly logins | Teams sharing a chart each week |
| Online store builder | Stores created | Stores with a first sale in 14 days |
| Team chat | Messages sent | Teams with 3+ active members on day 7 |
| Language app | Minutes practised | Learners returning 4 days of 7 |
Common mistakes
Picking revenue too early. Revenue is the result of the product working, not a way to tell whether it does. In year one, a small number of large deals can hide a product nobody uses.
Changing the number every month. If the number does not move, the answer is usually to change the product, not the number. Give it at least a quarter.
Keeping it secret. The number should be on the wall, in the weekly email and in every planning meeting. A metric only the founders know about does not change anyone's work.
Forgetting the guardrails. One number can be gamed, even by accident. Pick two or three "guardrail" numbers, such as support tickets and refunds, and check them every month to make sure you are not buying growth with pain.
When to change it
Your first-year metric will not be your third-year metric. Change it when the product changes shape, when you enter a new market, or when the number is so high that moving it further does not matter. Until then, keep it, and keep looking at it.
What we chose, and how it went
For the record, here is how this played out for Tally itself. In our first year, our number was "teams that shared at least one chart in the week". It started at 9%. We ran the workshop again every quarter and kept the number. By the end of the year it was 31%, and it had changed almost every product decision we made, from the share button on every chart to the scheduled reports we launched this month.
We also made mistakes. For two months we tried "dashboards created" instead, because it was higher and felt better. It went up nicely while retention went down. That was the most expensive lesson of the year, and the reason the section on common mistakes above exists.
A template for your wall
Write your number like this, and put it where the whole team sees it every day:
- The number: share of active teams who shared a chart this week
- Why it matters: sharing means someone trusted the data enough to act on it
- This quarter's goal: from 24% to 30%
- Guardrails: support tickets per team, refunds, page load time
