The Customer-Product Fit Loop
Alternate customer conversations with focused product changes
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 99%
Armstrong says Y Combinator taught founders without product-market fit to spend their time on two linked activities: talk to customers and improve the product from what they learn. Coinbase's first hosted Bitcoin wallet attracted sign-ups, but people did not return. Armstrong contacted about ten inactive users. One explained that the wallet was of little use because he had no Bitcoin; he said an easy way to buy some would likely have kept him engaged. Coinbase built a buy button, despite the difficult banking and licensing work behind it, and Armstrong says organic daily growth followed. The loop is customer evidence, focused build, observed behaviour, and repetition. Capital follows when demand, rather than a flat product, becomes the scaling constraint.
Origin
Armstrong learned the loop at Y Combinator and applied it to inactive users of Coinbase's first hosted wallet. Their feedback led to the buy button he associates with the start of organic growth.
Core principles
- 01Early-stage work should centre on customer evidence and product improvement
- 02People who tried and abandoned a product can reveal the missing utility
- 03A feature matters when it removes a real barrier to continued use
- 04Fundraising is stronger after product demand is visible
How to run it
- 1
Find the stopped users
Identify people who signed up, tried the product, and did not return. Their behaviour provides a concrete starting point for discovery.
Pro tip Contact a small batch personally before automating research.
Watch out Respect consent and communication preferences.
- 2
Ask about the barrier
Learn what they hoped to do and what prevented continued use. Avoid pitching or defending the current product during the conversation.
Pro tip Ask what would have made the product useful in that moment.
- 3
Select a consequential change
Translate a credible barrier into the smallest product change that could remove it. Prioritise utility over superficial engagement tactics.
Pro tip Prefer changes that complete the user's core job.
Watch out One interview is a clue, not automatic proof of broad demand.
- 4
Ship and observe
Release the change and measure whether users now complete the intended behaviour and return. Use behaviour alongside what customers say.
Pro tip Define the target behaviour before launch.
- 5
Repeat the loop
Return to customer conversations with the new evidence and improve the product again. Keep attention away from activities that do not advance learning or use.
- 6
Raise for a proven constraint
Seek growth capital when demonstrated demand cannot be served with current resources. Explain the bottleneck the capital will remove.
Pro tip A capacity problem is stronger evidence than a flat usage chart.
Watch out Capital needs and timing vary; this episode does not establish a universal fundraising rule.
In the wild
Coinbase's early wallet gained sign-ups but little repeat use. An inactive user told Armstrong that he liked the app but had no Bitcoin. Coinbase built a simple purchase experience, and Armstrong says daily organic growth began after it launched.
→ Customer feedback identified a missing core action rather than a cosmetic retention feature.
As purchases grew, Coinbase had to use corporate cash to buy customers' Bitcoin before bank debits settled. Armstrong says roughly $550,000 of a $600,000 raise was being used to service daily buying, which made the need for more capital concrete.
→ The fundraising story addressed a demand-driven capacity constraint.
Common mistakes
Substituting startup theatre
Conferences and fundraising do not repair a product that customers cannot use or do not retain.
Building from one comment alone
A conversation should guide a test, while observed behaviour determines whether the change generalises.
Raising before the product works
Armstrong's example favours capital for a demonstrated scaling bottleneck rather than as a substitute for fit.
Is it for you?
Best for
Early-stage teams with sign-ups or trials but weak retention and no clear product-market fit.
Not ideal for
Mature products where representative research, reliability, regulation, or portfolio trade-offs require a broader process.
From the transcript
“talk to your customers and then improve the product based on their feedback”
“the minute that we launched that buy button it started to grow every day organically”
From the episode
Coinbase Founder: The Crazy Journey Of Building A $100 Billion Company: Brian Armstrong