The second in a series of posts about building companies and teams.
Preceeded by "What I learned at Venmo", and followed by "What I learned at Google".

By the time I left Venmo in 2013, I had started thinking about education - what I thought was the "root solution" to the disparate problems in the world.

Though I hadn't set out to start a company, I soon met someone I instantly and deeply connected with (at a job interview - he was the interviewer), and we decided to build something new together.

Over the next 7 years, we built two products that have been used billions of times by hundreds of millions of students, assembled an amazing team and built a company culture we loved, executed a pivot that completely changed the product while retaining our core mission, got acquired by Google, and started powering their biggest products.

From the beginning, and through multiple products, we focused on the same user and moment: a student studying by themselves at home, looking for help with their schoolwork. This experience was universal, a daily need, and terribly broken in 2013. There remains much room for improvement today.

Our philosophy of caring about this problem over any particular solution helped us make company defining decisions with relative ease. The biggest shift was two and half years in, right after the Series A that we raised on the strength of our community-driven Q&A site (StackOverflow for school). The community was healthy, traffic was growing, SEO was working. But our reliance on SEO prevented us from building a direct connection with our users. To build that direct connection, we needed to be a mobile app that could quickly and reliably answer most of their questions - which our community-driven model could not.

So a couple of us started a new project. We put out a test app called 'Homework Genius' that did one thing: take a query and show Bing search result pages side-by-side, making it easier for a student to compare and consume them. Even that started getting traction. We quickly built conviction that we could build an app that delivered better educational results than Google, and was far easier to use: take a picture of your problem, and we’ll teach you how to solve it. We shifted more of the team to work on it, and moved fast. We added photo input, math OCR, sped up results, built classifiers that detected the concepts within your query, wrote almost a textbook worth of content, and within a year we had a top ranked and rated education app.

There were a few ideas and phrases we would repeat endlessly at the company. "Most of our users are in the future" gave us the freedom to change our product for the better as needed. "My job is to put myself out of a job" encouraged me to actively coach engineers so I could be free to work on non-engineering problems.

Some of these phrases were key product principles, and one such enduring principle was: "if you want someone to do something, make it easy for them." We wanted students to learn how to solve their problems, not just find and copy an answer. So how could we make learning easier for them?

When we were building our community-driven Q&A site, that root principle led to the first set of "three easys." For our content to be used, it had to be "easy to find." For students to learn from it, content had to be "easy to understand". And because correctness mattered but a crowd-sourced answer often started out a little imperfect, the content had to be "easy to improve." These principles guided a lot of our product development in the early years, and we had them painted on the wall.

For our second product, the AI-powered search app, these became: "easy to access", meaning input should be easy, and the experience should be mobile-native and fast. "Easy to trust" meant that students should be able to quickly decide if the results were worth their time. And "easy to understand" remained critical - the results had to be easy to learn from, something many educational publishers seem to forget. (This post describes our app's product philosophy in more detail.)

These and other oft-repeated principles equipped the team with rules by which to make quick and confident decisions, and ever since then I’ve tried to have simple principles for our team to remember and use.

Once a year, we’d take a break from focused execution, slow down, and go on a trip together. Our first year, it was a month in Brazil, the next year, two weeks in Costa Rica, then Nicaragua, then Morocco.

These trips brought us much closer together. In the city, we had our own friends and relationships. But on these trips we could get to know each other deeply, and this closeness helped us trust, debate, and build together.

We would surf or explore in the morning, hack on new ideas during the day, and cook and clean together at night. We'd look out a few years into the future, sketch, prototype, and scratch our collective itches. Each trip led to breakthroughs that informed and altered our course in the coming years.

I had known that luck was a big part of any startup's story, but I hadn't appreciated how much of a difference one person, or one moment could make. While prototyping our crowdsourced Q&A site, we interviewed a few contributors to other educational forums. When we launched our site a month later, one of those interviewees, Ernest, showed up unprompted, started answering questions, and came back every single day for the next three years. He wrote almost 10,000 answers which have been seen nearly 50 million times. Another community member, Stefan, joined a bit later, contributed even more knowledge and time, and eventually took over management of the entire site. A few fans carried us a long way.

A huge moment for the app, which may have felt insignificant to the person responsible, was when The Daily Mail featured us on their Snapchat Discover (the 20th story posted by them that day in 2016). That one story got us nearly half a million downloads in one day, and hundreds of thousands of downloads in the days that followed. Our usage grew by orders of magnitude and stayed there, as did our place on the charts. No amount of PR or prominent AppStore features by Apple matched this impact.

Despite our success at building great education products, there were two tough lessons we had yet to learn: first, that a great product doesn't necessarily make for a great business. And second, to make it, you have to last long enough.

The Socratic app was consistently top ranked on the education charts, answered millions of queries a day, had a 4.9 star rating, and received endless loving reviews. We made the mistake of believing that we’d easily translate this love and use into a sustainable business. When we finally decided to tackle monetizing the app, we realized that a subscription business required a different, more focused content strategy, and that our current path was best suited to ads, but required an order of magnitude more traffic to cover our burn.

By that point, we were five years into the life of the company, Chris had just had a baby, the toughest years lay ahead, and we didn’t know if the US market was ready to pay for educational products. At the same time conversations with two potential acquirers were trending positive, so instead of signing a Series B term sheet, we signed one to get acquired by Google and continue our work there.

Most of the team moved to California, where we joined forces with a small team within Google that had been circling around similar ideas. We rebuilt and relaunched our apps using the best technologies we could find. We stopped contributions to the Q&A site but kept the content up (traffic grows every year with no effort). And our technology and perspective on education now touch many significant education efforts within the company.

It’s been a great home for the company’s mission. But in the following years, which have been the best and most impactful yet for edtech, I’ve sometimes wondered what we would have done with $10 million in the bank and a few more years.

//////

There were endless other lessons over the years. Some others:

One thing that helped us execute well without brutal hours was articulating clear goals at multiple granularities. After our Series A, we made a mock deck for our next fundraise. What story would our future investors want to hear? With that as a goal for the round, we would set “episode” goals for the ~6 weeks between board meetings. And using those, we’d set goals for the week - what did we need to get done to stay on track?

The most valuable part of a board meeting would be the preparation beforehand. Of course the conversation with the board was always helpful, but the real challenge was in taking a step back from the day-to-day, asking ourselves the hardest questions we could, checking-in on our progress and strategy, the money in the bank, the happiness of the team, and all the other things that we’d set aside while executing on the goals of the week.

On getting our first users: Chris bumped into a high school friend at Grand Central, who told him that her brother was now making Chemistry videos on YouTube, and was getting popular. We instantly hit it off with Tyler - he cared deeply about students, understood their struggle, and had great insights on how to make good learning content. He offered to post a message under every one of his videos, that if they wanted more help, they should visit Socratic. This instantly gave us enough traffic to start learning from our early product. Later in the “AI” days, he helped us build a classifier that identifies the specific concepts within a query.

I learned the hard way that it’s in our nature to wait too long to do the hard, but obvious thing, especially when it affects other people. I’ve often known within a few weeks of hiring someone that something isn’t working. But between feeling like it’s my responsibility to give actionable feedback, set clearer expectations, or change something in the environment to make it work, or feeling uncertain about this gut truth I know, I’ve often waited too long to make a call that I should have made a month or two earlier.

On the flipside, I also learned that people are often capable of much more than they’re given. Every time we increased someone’s responsibilities (usually because we had to), they surprised us by quickly growing to meet the task. We often wished that we’d trusted someone with greater responsibility sooner. The benefits of faster decisions compound over time, whether the decision is to let someone go or to trust someone with more, so make the decision, and take action sooner.

It had been almost a decade since most of us were in high school, and the world had changed a lot since then, so we worked hard to stay in touch with the high school experience. We would regularly bring in students who had never used Socratic, and would show them prototypes and learn about their struggles and needs. We developed a good relationship with a nearby school so we could learn from students throughout the semester. And we built tools to send messages to targeted students or cohorts on our app, so we could ask them specific questions about their usage.

We were lucky that the early team liked surfing or the beach, so the first three team trips were to small beach towns in Brazil, Costa Rica, and Nicaragua. Each time, we would get a house near the beach but away from the town. We learned that having this quiet space to think, talk, and dream together was critical. The last trip was to Marrakech, where the old city was right out the door and the intensity of those streets made it harder for us to get into a good thinking groove.

//////

A lot of people were a part of this journey. Thank you...

Chris for being an amazing partner, for leading with warmth and creativity and for teaching us all to dream big and never lose sight of the user.

Kuba and David for joining us before we could afford to pay you and for sticking with us throughout. And Becca, Gen, Ahmed, Lili, Serdar, Jess, Rosie, Tyler, Will, Jackie, Sarah, Ian, Olivia, Grant, Emily, Evy, Anthony, Chaoyi for going on the journey with us.

Andrew Parker for believing in us instantly, and being a supporter till the end, and also Sean, David Tisch, Rafa, Kortina, John Maloney, Matt Hartman, Adam Reuben, and Randall. Joe Cohen for introducing us to our early investors, and connecting us with multiple people on the team, including connecting me to Chris.

Avni, Shantanu, and Rich for helping us find a home at Google. And Asha, Paddy, Manju, Basant, Shipra, Staub, Kate and so many more amazing people for support over the years.

This is a follow-up to "What I learned at Venmo" - my first startup experience.
I would love to hear from you!