These are my notes from “How to Find Product Market Fit” by David Rusenko. An entry of my Startup School series where I follow the curriculum in Y-Combinator’s Startup School
David is the founder of Weebly, which sold to Square.
- Feb 2006 - first line of code
- Oct 2006 (6 months in) - still haven’t launched
- Oct 2006 - 30 users signed up by hustling forums
- Nov 2006 - got into YC, Tech Crunch press release
- Jan 2007 - dropped out of school, moved to SF, started YC (11 months in)
- April 2007 - trend is wrong direction! we don’t have product market fit (14 months in)
- April 2007 - raised 650k series A on demo day
- July 2007 - News Week Press Release (15 months)
- Aug 2007 (18 months in) - Times Magazine press release still no product market fit
- Oct 2007 - trend changes! product market fit!
This is a chart of new user signups over time.
A few notes about this graph
- Each spike was result of one of the press releases.
- After each spike, signups settled down at a much lower level, but slightly higher than it used to be.
- After the signups settled, the signups still would slowly trend downwards, this is the wrong direction!
- This is not product market fit. Because the trend leads to 0 signups eventually.
finally in Aug 2007, the trend turned upwards! Signups were growing over time instead of declining. This is Product Market Fit!
The best companies create a market
- How can you create a market? By definition, market research won’t help.
- Find a hidden need — what are you a substitute for, what need are you serving better? What job are you being hired for?
- Where are you getting pulled? Where are people hacking the thing you got and using it in a way you didn’t intend? Double down on that.
Everyones’ going to tell you the idea is dumb. We demoed in front of 1000 people we pitched. The founder told us it was the dumbest idea he’s ever heard.
How to Find Product Market Fit
1. Talk to customers. Develop a market thesis.
2. Listen to their problems not their solutions.
When they talk about their pain, listen to that. when they talk about their solutions, you can ignore that.
“Some people say, “Give the customers what they want.” But that’s not my approach. […] I think Henry Ford once said, “If I’d asked customers what they wanted, they would have told me, ‘A faster horse!’” People don’t now what they want until you show it to them. That’s why I never rely on market research.” - Steve Jobs
It’s absolutely critical to talk to customers and understand their pain. No amazing product was ever created in a vacuum and delivered perfectly on V1. Listen to their proposed solutions, but dig deeper. What underlying job are they trying to accomplish?
3. Rapid prototyping & user testing.
Building a fully functional product is the most expensive possible way to test your hypothesis. It’s going to take about 10x the iterations you think it will take to get there, so trying to build the product every iteration will take too long. The best thing you can do is get to a functional prototype as fast as possible. Don’t worry about scaling, monetizing. We were worried a bunch of people would come at once, that will not happen, it never happens.
- Focus on getting to a functional prototype as quickly as possible that you can get in front of users.
- Don’t worry about scaling, monetizing, etc. This will come later.
- Expect it will take 10x the number of expected iterations:
- Keep your burn low.
- Build a team that can do this quickly (this is what makes outsourcing difficult in the early days).
4. Build the solution to their problems.
5. Test the solution with them.
Two important rules:
- Make sure you are talking to your target customers (but be open to changing your view on who those are). If a bunch of people show up hacking your product, they’re now your customer
- Don’t overthink it. Anecdotal is OK.
Most helpful tools in the early days:
- Customer interviews (5-10 needed). Talking to a customer for an hour. understanding everything, their problems, their pains
- UX testing sessions (3-5 needed).
Metrics (you’ll never have as large a sample size as you’d like — beware of telling too many stories with too little data).
UX testing sessions are simple & invaluable:
- Get someone to use your app/service in front of you.
- Encourage them to give open and honest feedback.
- Ask them to perform a task (you are not allowed to touch the phone/keyboard).
- DO NOT SAY OR DO ANYTHING.
- Watch in extreme agony as they struggle to figure it out.
You only need 3-5 testing sessions to uncover the most critical issues.
example: We wanted to make it as easy as possible for people to sign up. So we got rid of password confirmation field cause they can just reset if they type it wrong, we got rid of everything other than email and password. Then in 100 point font on the front page we added the text: “Signup Here”. Got my friend to come downstairs, said “signup”, after 45 secs he gave up. I asked why he couldn’t figure it out, there’s this huge 100 point font in front of your face that says signup here.” He said, “Oh, that makes sense, I was looking in the top right corner for the signup button. I though this form was to login since it was just email and password.” We added a non functional field for “name”, and then people started to see it as a signup form.
6. Did it work? If false, GOTO 1.
7. By the time you’ve reached this step, you’ve likely looped ~27 times.
When Should We Launch
Minimum Remarkable Product - Launch when your product is better than what’s out there.
- only one thing matters. don’t go to conferences, don’t write blog posts, don’t do anything that isn’t getting you to your next milestone. Which is probably at this point getting to product market fit.
- optimize for learning - What’s the biggest unknown right now. Do that thing first. It might seem low priority, or low effort. When you learn that thing, you end up throwing out all your old priorities, and now you have a new list.
How do I know when I achieved PMF
1. return usage. Day 1, 3, 7, 30 retention).
2. Net Promoter Score (NPS) > 50 is product market fit.
NPS stands for Net Promoter Score. It is most often captured through surveys or in-app experiences, like the smiley-face ratings you have probably seen. That is an NPS in action. We sent a survey early October to direct (non-MSSP/OEM) customers. The survey asked a few questions around product, customer work flow and one key question - on a scale of 1-10 “How likely are you to recommend us to a friend or colleague?”. Then, the math happens:
Categorize survey responses by scale into 3 buckets:
Detractors: 0-6 (unhappy, potentially damage brand, negative word-of-mouth) Passive: 7-8 (satisfied, but not thrilled, more easily swayed to competitors) Promoters: 9-10 (loyal, refer others, fuel growth)
Tally the number of responders in each category and calculate the percentage from all responders. Then, subtract the percentage of Detractors from percentage of Promoters to get the NPS score. The score can range from -100, indicating every customer is a detractor to 100, where every customer is a promoter. Our Detractors composed of 10% of total responders vs 47% Promoters, resulting in 37 NPS (47%-10%).
3. Paying customer renewal rates.
Renewal rate is the defined as the percentage of eligible subscriptions who renew, which is cohort-based. Churn is typically defined as the number of churned customers / total customers, which is not. Churn, while easier to measure, can be deceiving especially if your growth is accelerating (it will look lower than it actually is).
Metrics not included
- Signups. - Signups are more of a reflection of market fit than product market fit. You’ve identified something a lot of people want — but did you build the right solution?
- Conversion rate. - Conversion rate improvements can often be tactical and unrelated to product market fit. All conversion rates start low
- Full Video - https://www.youtube.com/watch?v=0LNQxT9LvM0
- How to Find Product Market Fit (SUS 2017) by Peter Reinhardt - https://www.startupschool.org/videos/13
- The Real Product Market Fit by Michael Seibel - https://blog.ycombinator.com/the-real-product-market-fit/