Hey — It's Nico.

This week I'm testing a new format for the newsletter. It'll have two sections:

• A startup story. This week, how Segment failed, pivoted, found PMF, and exited for $3.2B.
• Weekly picks (startup-related news & resources).

Can you do a favor for me? Reply to this email with "YES" if you like this new format. Otherwise, reply "NO".

Your feedback helps me keep improving Failory. Feel free to share further comments in your reply.

​

​

How Segment Found PMF

Segment is a B2B customer data platform that raised $600k after participating in Y Combinator in 2011, raised more than $280M from venture capital, and got acquired for $3.2B by Twilio – the dream startup story.

Yet, shortly after their early-stage fundraising success at Y Combinator, the Segment team was in deep trouble. After two unsuccessful attempts to make their ideas work, they were running out of money and had to figure out what to do.

Paul Graham, the founder of YC, told them: ​“So you’ve spent half a million dollars... and have nothing to show for it“.

So how did they go from a failing startup to a $3.2B exit?

Peter Reinhardt, co-founder and CEO, likes to talk about their PMF experience. Finding strong indications of PMF shortly after their meeting with Paul was the main determinant of their success. He likes to quote Jason Lemkin, saying:

“Getting from $0–1m is impossible. Getting from $1–10m is unlikely. And getting from $10–100m is inevitable.”

Obviously, “getting from 0-1m” is the hardest part, and it’s the one where finding PMF is the key. Yet, this is something that’s not intuitive for a lot of people in early-stage businesses:

“As engineers who had never done this before, talking to people didn’t seem like real work. Real work was coding. But in reality, 20 hours of great interviews probably would’ve saved us an accrued 18 months of building useless stuff.” – Peter Reinhardt (​source)

​

Segment’s PMF Journey

Idea 1: ClassMetric, a tool to help students signal they were confused during a lecture

The idea Segment started with was an app that allowed students to press a button whenever they felt confused and didn’t understand the lecture. The lecturer would then be able to see a real-time graph of the number of confused students. This (along with lecture note analytics) was the idea they used to raise their initial round of $600k at YC.

Yet, shortly after raising the money, they realized that their vision was at odds with reality. When they tested the tool during a lecture, they noticed that initially, 60% of students were browsing Gmail and Facebook instead of using the app, and by the end of the lecture, this number rose to 80%.

Students were just not using the product.

The team had to call their investors, tell them the idea didn’t work, and ask what to do with the money. The investors told them to look for a new idea because they had invested in the team.

​

Idea 2: Segment.io, an analytics tool focused on segmentation

The new idea was to build an analytics tool a bit like Google Analytics but focusing heavily on segmentation. After 12 months of building, they had no real customers.

That’s when the meeting with Paul Graham happened, and that’s when the team knew they had a drastic decision to make: they had $100k left in the bank - would they test out a new idea, or would they stick to segment.io until they were forced to shut down?

​

Idea 3: analytics.js, a library for user data

While the analytics tool didn’t have any traction, they had an open-source library that collected user data and was able to send it to different analytics tools that saw some interest (they created it while working on ClassMetric, and refined it during segment.io). Some team members thought that this library had the potential to grow into a big business, while Peter thought this was a bad idea. In order to test it, they created a landing page with an email signup form, posted it on Hacker News, and waited to see what happened.

Unlike their other ideas, this one instantly attracted attention. The team noticed the real problem that their open-source library was solving (their MVP) and expanded on the solution a great deal to build a big business on top of it. In retrospect, $500k and 18 months of coding resulted in nothing, but $100k, a landing page, and a post on Hacker News led to a $3.2B exit.

​

Segment’s PMF Lessons

Lesson 1: Having and not having PMF feels different

In his talks about Segment’s PMF journey, Peter loves to share a statistic he found: founders that don’t find PMF once aren’t more likely to find it the second time they try (the chance remains ~22%). However, founders who are successful at finding it once are more likely to find it a second time (~32% chance).

One reason this might be true is that before you’ve experienced a strong pull from the market, it’s very easy to delude yourself that what you are experiencing is PMF, while in reality, it is not.

In the case of Segment:

• Bullying your customers into using your product isn’t PMF​: Some professors agreed to try their ClassMetric product, but most likely, they did so out of pity rather than out of desire or necessity. This could also be true if you have a good salesperson on the team: you might convince your customers to buy what you are selling, but this doesn’t necessarily indicate they need it.
• Idle interest isn’t product-market fit: ​Some people were curious about segment.io. Yet, those people didn’t convert into real customers. This kind of vague interest is dangerous - it was enough to deceive Peter and his team into thinking they were building something people wanted, while in reality, they weren’t.

If you’re wondering if you have PMF or not, you most likely don’t have it.

True PMF feels much more intense – like a loss of control. The market pulls, rather than you having to push.

​

Lesson 2: You need to discard not-good-enough ideas extremely fast

Peter advises that before PMF, you need to “be a cockroach.” You need to constrain your burn as much as possible, code as less as possible, and talk to customers as much as possible.

This way, you can test and try to validate the highest number of ideas, increasing your chances of success a great deal. Assuming that there is a ~22% chance to find PMF, on average, you would need to test five sufficiently different ideas before finding one that has strong enough PMF indicators. You need to plan to have the time to test at least that many ideas.

If you've read up to here, can you help me with a click here? It's my way of knowing if people find this kind of content useful.

​

Lesson 3: Your vision and intuition are irrelevant, all that matters is the market

Peter had a vision of the world and a strong belief that the classroom tool and the analytics tool were solving a real problem. It turned out not to be true.

He also believed strongly that the analytics.js library had no future as a big business. It turned out that this was wrong as well.

Business is all about the wants and needs of people in the market. It’s not about your vision of the world and your desired place in it. Because of this, being skeptical and humble in order to find real problems that people have is just as important as building good solutions.

“If you think about odds of a team failing to find product-market fit, it might look something like this (for the sake of illustration only):

• Non-technical team: 15% odds of being able to build it, 60% odds of solving a real problem.
• Technical team: 90% odds of being able to build it, 10% odds of solving a real problem.”

Peter believes that actively validating the problem is even more important for technical founders because this process is unintuitive for them. After you’ve found a real problem, building the solution is the easy part.

Returning to the $0-$1M quote, finding the first problem to solve is the hard part, but also arguably the most important part. Once you have a foothold, your new context will give you a new way to see your market. This means you will be in a unique position to see adjacent problems that you can solve much easier than other people, which is at the heart of innovative startups.

​

Lesson 4: Mocks before code

Having found PMF, Segment went on to develop new product features for the different needs of their customers. One such feature was Personas - a product for audience management. While Personas currently has a great PMF, while in development, the team ran into serious doubt and confusion​.

“If we ruthlessly prioritized our customer problems and solutions, instead of embarking on an infrastructure project too soon, we never would have let the runway question creep up on us.” - Tido Carriero

The principles of PMF apply to each new product or feature even after you have a big business. Tido recommends to ​“prototype, iterate, and try not to code” ​when developing new features, which isn’t at all different from what you’d do in a new, idea-stage startup.

​

​

This Week’s Picks

Ramp’s Product VP on how the startup became the fastest-growing SaaS (Link).

How B2B startups came up with their ideas (Link).

54 VC-backed startups have filed for bankruptcy this year (Link).

A newsletter on how to build a successful venture fund (Link).

The ‘Stripe for Insurance’ startup that just raised $35.5M (Link).

A breakdown of Drake’s eCommerce store (Link).

The possible economic impact of Generative AI (Link).

A $4M/year SaaS founder shares his midyear financial update (Link).

OpenAI files trademark application for “GPT-5” (Link).

Founders are using “Brown Noise” to stay productive (Link).

Startups are copying Google’s usage of “.new” domains (Link).

Lessons from Keith Rabois (Link).

​

​

Remember to reply with "YES" if you like this new format and "NO" if you dislike it or prefer the old one.

That's all of this week.

Cheers,

Nico