UzCombinator
UZ
EN
Startup School
Curriculum0/19
Module 1
Idea and problem
  1. 1What a startup is, and how it differs from a small business4 min
  2. 2Finding a real problem4 min
  3. 3Judging an idea: market, timing and team4 min
  4. 4Co-founders and the team4 min
Module 2
Customers and validation
  1. 5How to run a customer interview5 min
  2. 6Who your first users are4 min
  3. 7Testing demand cheaply5 min
Module 3
Building the MVP
  1. 8What an MVP is, and what it is not4 min
  2. 9Building fast: code, no-code and ready-made tools4 min
  3. 10Launching and collecting feedback4 min
  4. 11Signs of product-market fit4 min
Module 4
Growth and metrics
  1. 12Choosing the metric that matters4 min
  2. 13Unit economics: CAC, LTV and margin4 min
  3. 14Your first growth channels4 min
  4. 15Keeping your users4 min
Module 5
Fundraising and pitching
  1. 16Do you need funding? Stages and sources4 min
  2. 17Building a pitch deck4 min
  3. 18Meeting investors4 min
  4. 19Terms and valuation basics4 min
Module 4Lesson 154 min

Keeping your users

Growth and metrics
Why users come back or leave, and how to read a cohort chart.
Winning new customers is hard and expensive. If they leave quickly, that effort is wasted. Retention is the most reliable sign that a product delivers real value, and the foundation of sustainable growth. Picture a leaking bucket: however much water you pour in, it will not fill until you plug the hole.

Why retention matters

  • Customers who stay bring in revenue every month, and new customers add on top of them.
  • They recommend you to others, which is the cheapest growth channel there is.
  • They give you the most information for improving the product.
  • For investors it is one of the main pieces of evidence that the product has found its market.

Cohort analysis

A cohort is a group of users who joined in the same period, say the same month. Cohort analysis shows what share of each group stays active over time. In a simple spreadsheet you do it like this:
  • each row is the month people joined, for example March, April, May;
  • each column is the months since joining: month one, month two, month three;
  • each cell holds the percentage of that group still active in that month.
Decide in advance exactly what "active" means: for example, a user who completed at least one core action during the month.

How to read the curve

Usually some users leave in the first months; that is natural. What matters is where the line stops. If it keeps sliding towards zero, the product is not holding anyone for long. If it levels off at some point, the product has real value for someone. A good sign is when newer cohorts are retained better than older ones: it means the changes you made to the product are working.
What counts as "good" varies by industry, so watch how your own cohorts change rather than comparing yourself with other people's numbers.

Why people leave

  • They never reached the core value. They signed up, but stopped before seeing the first result.
  • The product does not solve the problem well enough.
  • The price does not match the value they get.
  • Seasonality. Some businesses are less active in certain months of the year.
  • Their champion left. With business customers, if the employee who backed your product internally leaves, the customer may leave too.

Talk to customers who left

This is one of the most uncomfortable conversations, and one of the most useful. Send a short message or call: "We noticed you stopped using our service. If you could tell us why, it would help us improve the product." Many people answer honestly. Keep a record of the reasons that repeat, and group them: product, price, service or outside reasons. Then it becomes clear which problem to tackle first.

Ways to improve retention

  • Shorten the path to first value. A new user should see the main result on day one.
  • Send useful reminders. Not spam, but messages that help at the moment they are needed.
  • Respond quickly. A fast reply on Telegram often keeps a customer.
  • Notice people going quiet. Spot when a customer's activity drops, and get in touch before they leave entirely.
  • Embed the product deeper into their work. The more of their data people keep in your product, the harder it is to walk away.

Retention with business customers

If you sell to companies, track revenue as well as the number of customers. Some customers leave, while others start paying more by adding staff or branches. If the group of customers you had a year ago is paying more today than it did then, that is a very strong sign. Hold short, regular check-ins with every important customer: what is working, what is missing and what might stop them renewing.

Try this

Build a cohort table for the last three to six months. Work out whether the line is levelling off. Then contact five customers who left last month, ask them why, and plan one change that addresses the most common reason.