पाठशाला Pathshala · संचालन Sanchālan, Operations · Lesson 02 · Start

The north star metric and its input tree

One number that measures the value customers actually received this week, and the three or four inputs beneath it that a team of ten can move. How to choose it, break it down and know when to change it.

Pathshala, The Founder Library · 11 October 2026 · 7 min read

Ask a founder what the company is for and the answer takes five minutes. Ask what number went up last week and the answer is usually revenue, which tells you what happened two months ago to people who have already decided. A north star metric is the number that tells you what the company did for its customers this week, and the input tree beneath it is the part a team of ten can actually move on Monday.

This lesson covers what the metric is and why it is rarely revenue, the categories successful companies draw theirs from, how to break it into inputs that specific people own, the interactive tree that shows why the breakdown matters, and the two cases in which the right answer is a different metric altogether.

What a north star metric is

Amplitude’s North Star Playbook defines it as a single, crucial metric that represents the value your product provides its customers. Three words in that sentence do the work. Single: one number, with the rare exception of a second metric added to guard quality. Value: the metric counts something the customer received, not something the company did to them; registered users and downloads are activity, not value. Customers: it is measured on the people who pay or are meant to, in the unit in which they experience the product, which is why a marketplace counts transactions and a messaging platform counts messages sent.

The companion Amplitude note adds the properties a good one has: it aligns to customer value rather than vanity, it is understandable as a statement of the company’s strategy, and it is a leading indicator of future results rather than a lagging one like monthly revenue or revenue per user. It also offers a useful sorting question. Which game is the product playing? An attention game asks how much time customers are willing to spend in the product. A transaction game asks how many transactions they make. A productivity game asks how efficiently someone gets their work done. Most Indian startups a founder will build are in the second or third game, and the metric follows from which.

What forty growth-stage companies chose

Lenny Rachitsky surveyed current and former employees at more than forty growth-stage companies and published the results as a guide to north star metrics. Six categories emerged. About half the companies used a revenue metric such as ARR or GMV. About a third used customer growth: paid users or market share. About a third used consumption: messages sent, nights booked. About a third used engagement: daily or weekly active users. One in ten used growth efficiency such as LTV to CAC or margin, and one in ten used user experience such as NPS. The shares add to more than one hundred because some companies carried two.

The examples are the useful part. Marketplaces chose transaction volume: nights booked, rides taken, orders placed. Platforms chose usage: Twilio’s messages sent, Plaid’s bank accounts linked. Freemium B2B companies chose paid activity: Airtable’s weekly paid seats, Asana’s weekly active paid users, Slack’s number of paid teams. Consumer subscriptions chose engagement: Duolingo’s daily active users, Netflix’s median view hours per month. Shopify chose active merchants. Patreon chose creators earning above a set amount, which is a metric about the customer’s success rather than the company’s. The pattern is that the metric sits one step before revenue in the chain of cause and effect, close enough that it predicts revenue and far enough that the team can move it this week.

Rachitsky’s own warning about the largest category is worth quoting: revenue is spiky and thus hard to make operational, focusing on it early can drive poor pricing decisions and it rarely motivates a team. A north star of ₹2 crore ARR tells a product engineer in Bengaluru nothing about what to build on Tuesday. A north star of weekly active paid seats tells her exactly where to look.

The input tree

The north star is an output. Nobody can walk into the office and increase it directly. What a team can do is move the inputs, which Amplitude defines as a set of three to five influential, complementary factors that most directly affect the north star and that the team can influence through the product. The recommended way to find them is to consider four dimensions. Breadth: how many customers are active. Depth: how much of the product each one uses, or how much value each transaction carries. Frequency: how often they come back. Efficiency: how quickly or reliably they get the result. Most north stars decompose into two or three of these as a multiplication, and the multiplication is the test. If the inputs do not multiply or add up to the output, they are not its inputs; they are other metrics.

Sequoia’s note on measuring product health describes the same structure from the bottom up. A drop in sessions is the earliest leading indicator for a drop in daily actives; a drop in daily actives predicts weekly and then monthly; early retention is a strong predictor of long-term retention. Each layer leads the one above it, which is what makes the tree worth drawing: the input that moves first is the one the weekly page should watch most closely.

The tree does two things once drawn. It assigns ownership, because each branch belongs to the function that can move it: sales owns accounts, onboarding owns seats per account, product owns the share active. And it decides priority, because a ten per cent improvement in each input does not produce the same lift; the branch with the most room and the cheapest lever is the quarter’s objective, which is where [OKRs](/library/okrs-for-team-of-ten) come from. Three inputs is right for a company of ten. Five is the ceiling. Beyond that the tree is an org chart, not a strategy.

Revenue tells you what customers decided two months ago. The north star tells you what they received this week. Only the second one can be acted on Monday.

Two cases where the answer is different

Before product-market fit. Rachitsky’s advice for this stage is to focus on cohort retention and ask whether people keep using the product, and Sequoia calls retention the best indicator of product-market fit. A company with 400 users and no evidence that month-two users become month-six users does not need a north star; it needs a retention curve that flattens, which [the product-market fit lesson](/library/what-product-market-fit-looks-like-in-the-numbers) shows how to read. Choosing a growth metric before that point measures how fast the bucket is being filled and ignores the hole.

When the strategy changes. The metric should evolve as strategy changes. Rachitsky’s line is that your north star metric is your strategy and your strategy is your north star metric, which cuts both ways: a company that changes its metric every quarter has no strategy, and a company that keeps a metric after the strategy has moved is steering by a star that has set. The test is annual at most. A D2C brand that spent two years on repeat orders and now sells through quick-commerce platforms has a different customer and probably a different north star.

A worked example

A Jaipur company sells a subscription of fresh staples to households, delivered weekly. Revenue is ₹1.4 crore a month. The founders propose GMV as the north star. Walk the test. Is it value delivered? Partly; a household that received a late or incomplete delivery still counts. Is it leading? No; it lags the decision to subscribe by a month and the decision to cancel by two. What game is this? A transaction game with a frequency component. The honest north star is households receiving a complete on-time delivery this week, and the tree beneath it is active subscribed households × deliveries per household per week × share delivered complete and on time. Operations owns the third branch, growth owns the first, the product team owns the second through the plan sizes it offers. In the first quarter the third input is 91 per cent and the cheapest ten per cent of improvement in the whole tree is there. That is the quarter’s objective, and GMV will follow it with the usual two-month delay.

Choosing yours, in one sitting

Block two hours with the founders and whoever owns the data. First, write the sentence: the value a customer gets from the product is …, and the unit it arrives in is …. Second, name the game, and list three candidate metrics that count that unit. Third, run each candidate through the checklist: value not activity, leading not lagging, understandable as the strategy, measurable weekly with the data you have. Fourth, for the survivor, draw the tree: which three or four inputs multiply or add to it, and which function owns each. If the arithmetic does not close the metric is wrong or the inputs are. Fifth, put the north star in the first tile of the [weekly page](/library/weekly-metrics-review-one-page-one-hour) and the inputs in the next three, and do not touch the definition for a year.

Then, every quarter, ask one question in the planning meeting: which branch of the tree has the most room, and is that where the quarter’s effort is going? If the answer is no, one of the two has to change.

The figures in the tree are illustrative. The sources are below; Rachitsky’s survey and Amplitude’s note between them cover the ground in half an hour.