In your first years you will be buried in dashboards, and almost none of them will tell you whether the business is actually working. The discipline that separates founders who compound from founders who churn is choosing one number that means "we are creating real value," then surrounding it with a few guardrails that stop you from gaming that number, even by accident. This lesson shows you how to pick your north star metric and the guardrails that keep it honest.
What a north star metric is
The idea was popularised by Sean Ellis. Your north star metric (NSM) is the single measure that best captures the core value your product delivers to customers and that predicts sustainable growth across your whole base. It is an output metric and a leading indicator of revenue: when your product, pricing, and distribution are genuinely working, the NSM rises before the bank balance does.
Notice what it is not. It is not a vanity metric such as app downloads, registered users, gross merchandise value (GMV), or social followers, and it is not last quarter's revenue, which is a lagging result. A good NSM answers one question: how many customers experienced the value we promised, this week or this month?
How to choose yours
- Tie it to delivered value. Think "orders delivered on time," "kirana stores that reordered this week," or "businesses that ran payroll this month," not "total sign-ups."
- Make it a live count or rate, not a cumulative total. A running total of "users ever" only goes up and hides churn. Active value events per period tell the truth.
- Confirm it correlates with revenue. Ellis treats revenue correlation as a test. If your NSM can rise while revenue stays flat, it is probably a vanity metric in disguise.
Why you need guardrails
Goodhart's law warns that when a measure becomes a target, it stops being a good measure. Push any single number hard enough and your team will find shortcuts that lift it while quietly hurting the business. Guardrail metrics, also called counter-metrics, are the small set of numbers you promise not to damage while you chase the NSM.
The India honesty check
The most common way Indian founders fool themselves is confusing gross activity with real revenue. Under Ind AS 115, if your platform only arranges a sale between a buyer and a seller, you are an agent and must recognise revenue net, meaning only your commission or take rate, not the full order value. You may book gross revenue only when you are the principal, controlling the good or service before it reaches the customer. So treat GMV as a scale indicator, never as revenue.
Two more India-specific honesty checks:
- Booked revenue is not cash. Under GST, your output tax liability is generally triggered at the time of supply, broadly the invoice date, so you can owe tax before the customer pays. Business customers also deduct TDS, so the cash that lands is smaller than the invoice. Track collections and days sales outstanding, not just billings.
- Advances are a liability, not revenue. Money received before you deliver is deferred revenue, recognised only as you perform the service. Booking it early inflates profit and creates restatement risk later.
A starter guardrail set
- Contribution margin per unit or per order, so growth does not quietly lose money on every transaction.
- Cash runway in rupees, the months of survival at current net burn. This is the guardrail that keeps you alive.
- Burn multiple, defined by David Sacks as net burn divided by net new annual recurring revenue. Lower is better, and a rising burn multiple is an early warning that growth is getting expensive.
- Retention or churn, because an NSM that grows only through new acquisition while old customers leak is not real value.
- One quality or trust metric, such as complaint rate, refund rate, or on-time delivery, to catch shortcuts that flatter the NSM at the customer's expense.
Write your NSM and three to five guardrails on a single page. Review them on the same day each month, and change them rarely. One honest number, protected by a few guardrails, will guide better decisions than fifty dashboards nobody trusts.