The moment your product has real users, you get access to an ocean of data — and that ocean will drown you if you let it. Dashboards full of charts feel productive and tell you almost nothing. After launch, the skill isn't measuring more. It's measuring the few things that actually predict whether you have a business.
Here are the six numbers I tell every non-technical founder to watch — and the rest you can confidently ignore for now.
1. Active users — are people actually showing up?
Not signups. Not downloads. The count of people who used the product in a way that matters this week. A customer who pays but never logs in is a customer who's about to cancel. Active usage is the heartbeat; everything else is a lagging echo of it. If this number is healthy, you're probably fine. If it's quietly sliding, nothing else will save you.
2. Monthly Recurring Revenue — are they paying, and is it growing?
MRR is the one number that turns "people seem to like it" into "this is a business." It's the predictable money that arrives every month. Watch the trend, not the total — is it climbing, flat, or slipping? Early on, the absolute figure is tiny and irrelevant. The direction is everything. (If your MRR feels too small for the value you deliver, the problem is usually upstream, in how you priced.)
3. Churn — how fast are they leaving?
Churn is the leak in the bucket. You can pour new customers in the top all day, but if they drain out the bottom, you're running to stand still. A few percent of monthly churn is normal; a steady climb is a fire. And churn is honest in a way surveys never are — people don't tell you they're unhappy, they just quietly stop paying. Watch it like a hawk, and when someone leaves, ask them why. That answer is worth more than any chart.
4. Activation rate — do new customers reach the "aha" moment?
This is the number most founders miss, and it's the highest-leverage one. Of the people who sign up, what fraction actually get to the point where the product delivers its promise — the first booking, the first claim processed, the first real win? If people sign up and never reach value, your problem isn't marketing, it's onboarding. As I learned in the first 90 days after launch, fixing activation almost always beats chasing more signups.
5. Reliability — is the thing actually up?
Once customers depend on you, uptime stops being a technical nicety and becomes a business metric. You don't need a fancy system — even a simple uptime monitor that texts you when the site goes down will do. The number to know: when something breaks, how fast do you find out, and how fast do you fix it? Reliability problems compound into churn faster than almost anything else. The cost of ignoring them shows up later as a five-figure repair bill.
6. Your cost to serve — are the unit economics sane?
For each customer, roughly: what does it cost you to deliver the product, and what do they pay? You don't need an accountant — you need to know you're not losing money on every sale. This matters double if your product leans on AI or other usage-based services, where costs scale with success and can quietly eat your margin (here's how I cut one product's AI costs by 73%). A growing business with upside-down unit economics isn't a business — it's a countdown.
What to ignore (for now)
Page views. Time on site. Social followers. Feature-by-feature usage heatmaps. Net Promoter Score. These aren't useless forever — they're just noise at your stage, dressed up as insight. The analytics dashboard you think you need is almost always procrastination with a chart on it. Six numbers. Check them weekly. Make decisions. Get back to building.
The one habit that ties them together
Put these six on a single sheet and look at them every Monday. Not ten dashboards — one page, six numbers, five minutes. The goal isn't to admire data; it's to notice when one of them turns, early enough to do something. A founder who knows their six numbers cold makes faster, braver decisions than one drowning in a hundred.
Not sure which numbers actually matter for your specific product, or how to track them without building a reporting system you don't have time for? Send me what you're working on — I'll help you set up the simplest possible scoreboard that tells you the truth.
