Last week, our team at SlaySchool.com passed 100,000 users in under 6 months! What started as a meme AI idea for medical students is now being used by learners of all ages & backgrounds globally. We’ve been profitable for the past few months, but more importantly I have learned a ton. Building a consumer startup is hard, it’s 10x harder in education. Here are some of our trade secrets. Please keep them to yourself 😉

The beginning: 0 Users

YC’s common startup advice is build something you want, something you’d use. We ignored that advice..sorta. We built an app that converts PDFs into flashcards because it’s something we wish we had when we were in school. We knew about Quizlet and Anki, but didn’t know about the hundreds of other companies in this space. 🤦🏽

We thought this could be a great tool for high school students, so we called it Slay. A name so embarrassingly silly, that somehow you can’t help but talk about it.

Reddit hates us: 0-10 users

We didn’t want to do some Gas-type social media shenanigans to reach high school students, so we decided to go to Reddit. Reddit has communities that talk openly about their problems and ask for suggestions. It’s the best place to go to peek into startup ideas. We decided to do a number of posts in various Anki sub-reddits about AI generated flashcards. We told users the product was free and they should DM us if they were interested. It needed to be free to get people’s attention, and we needed a DM to talk to them. Our goal wasn’t to get users at this point, it was to understand their needs.

In DMing users we learned a ton about what they used flashcards for, who they were, and why what we had built would be useful to them. We learned quickly that premed students, and medical students loved what we had built. But not everyone was a fan. For every 5 reddit groups we posted in, we were banned or kicked out of one. Flagged for spam, sales etc, but it was worth it because we got 10 hardcore users. They used our app 2-3x a week and gave us copious amounts of feedback.

Listening to feedback, setting up discord: 10-100 users

Everyone has opinions and feedback, the tricky thing is always deciding what to act on. This is why it’s important to understand what your user is trying to achieve. When we spoke to users we didn’t focus on what they told us they wanted, we focused on their study patterns. We tried to answer:

• When/where did they study?
• How did they use flashcards?
• How did they evaluate what to study?
• How all the tools they used came together?
• And what an AI flashcard maker really did for them?

These questions helped us decide what to build and how to make our product even better. At this early stage of building an MVP speed matters the most. If you’re waiting on the next customer call, email, etc, that’s too slow in the consumer world. Discord was essential for us to get dozens of private DMs with our users and get rapid feedback. We could send Figma files, invite them to our test accounts, and even screenshot stupid ideas in seconds instead of days. And the best thing about students is, they are native to these applications so its easy to use. Like most things worth attaining in life, great product development goes down in the DMs! 🤫

As more people answered our Reddit DMs, the more our discord grew. We knew we were on to something so we started to set up a low price. $8 a month or $60 a year. First month free to capture more users.

Exam season, social media: 100-1000 users

Around November we started to generate some revenue and decided to get on social media. We got all the main accounts - TikTok, Youtube, Instagram. We avoided Facebook because that’s for boomers and stalkers, duh 🙄. I started posting medical terms and screenshots of our app to get users. It was taking up too much time, but I learned a lot about video editing and putting out content. I used to make fun of influencers, but I get it - it’s work. We decided to hire an influencer to put content on our socials. The goal was consistency, not quality. One post daily on our socials. Post once a week about our product, and all the other times post tips and memes to keep the users entertained. We noticed traffic to our website would jump on weekends, so we posted content about Slay School every Friday. This, alongside our blog, our continuous cat and mouse with Reddit, got us to 1000 users.

We also got to learn more about the medical student market and identified interesting ways we can help this community. For example, many US med students love Anki, but finding Anki decks is annoying. They are in different search forums and websites with broken links and spam. We spent a few hours putting together all the Anki decks med students, dental students, nurses, etc would need and added it to our blog. These links alone generated half our traffic during the Fall.

Christina, the first influencer: 1000-5000 users

As I started to see our revenue and user numbers go up, I started to obsess over growth. Day and night I thought about how I can get thousands of users. I thought of crazy cheap growth hacks - going to libraries with flyers, posting google reviews of medical schools with a picture of Slay, hosting hackathons, doing webinars, writing application reviews and so on. All of these things were extremely high-effort with unclear payoff. The one thing we did see working however, was TikTok. We started to see more and more users come from TikTok to our website, sign up and pay. It was time to test this channel. I looked up various hashtags on tiktok and found a number of influencers. I had dozens of tabs open of these influencers and their emails and I reached out to them one by one. 80% ignored me. 19% charged me a rate that was insane. Only one decided to give me a chance. She was a medical student and she loved our product. She offered us a huge discount on our pitch and together we crafted a perfect script. She put the video out and for a few days nothing happened. Then over the weekend we hit a spike. Hundreds of people were signing up at once! It worked, and we realized we were on to something.

Meme or value creator?: 5000-10,000 users

In January we felt pretty good about ourselves. We had a marketing channel that we could leverage, and a cash flow positive product. The med school market was massive, and we knew flashcards were applicable to law students, engineers, language learners, and anyone studying hard to memorize topics.

Here is the part where I admit I have a coach. I am not a genius entrepreneur by any means, but through the YC community I have met actual geniuses and theres a few I talk to often. When I told them I was crushing it but just needed to boost my retention numbers, here was his reply:

I sat down with my team and asked for names. "What about Jessica? What about Mark?" We couldn’t think of a single person who loved our product so much. We had users, we had revenue, but we weren’t essential to solving anyone's problem to the point that they’d be disappointed if we disappeared. The closest we came to such a user was Kevin in San Francisco. It was time to make a trip.

In January, we flew to San Francisco with the sole purpose of meeting Kevin, some professors, and discreetly entering Stanford and UCSF medical libraries to interview students. We did exactly that. We had a 3-hour lunch with Kevin to understand how he uses Anki, how he studies for the USMLE, and what other tools he had at his disposal. We then visited Stanford and engaged every student who seemed mildly distracted, learning about their study habits. We didn’t pitch Slay School. We didn’t even mention our users and revenue. We simply gathered as much information as we could and only at the end did we discuss what we were working on. On a scale of 1-10, with 10 being the hardest—if approaching someone at a bar is a 10, this was a 9. Why were two non-med students building an app for med students? We felt out of their league! 😏

But that wasn’t all. We also added Intercom to our app, not for customer support but to converse with users. The student market is notoriously difficult to connect with digitally for feedback because they rarely check their emails. The only viable way to communicate with digital users was through the app itself. While we were learning, we didn’t stop marketing, so our growth continued.

Experiments and iterations: 10k - 30k users

Gathering all this feedback helped us realize a multitude of features we needed to build. For starters, we created a Chrome extension because students often use multiple screens, and another app isn’t always the best solution. An extension sits atop their computer screen and is easy to manage. We also integrated this with other web links, allowing students who use applications like Uworld or Khan Academy to create flashcards in seconds without needing to log in and open another app.

We enhanced document storage and added more upload features so that students could use this tool more frequently. We even introduced multiple-choice questions to allow students to practice quizzing themselves, rather than simply absorbing information. Additionally, we created a public repository of cards where, with student consent, they can share their work with others, making studying a collaborative effort. All of these improvements led to mini explosions of growth. We also discovered student clubs, schools, and multiple Facebook groups talking about us, and we provided bulk discounts to those who signed up with us.

While this may sound straightforward, it wasn’t. We spent most of January and February in constant testing. We connected so many events to Posthog that we exceeded the usage limit twice! I had to kindly ask their support team for help a few times. Intercom was also busy, as many of our releases were not tested properly. This led to frequent breakdowns, and March was spent fixing bugs. Each member of our team was on Intercom, even those coding. You can’t build a product with second hand information. Everyone on the team needs to talk to users. Everyone🫡.

Breakthrough: 30k - 80k

Alongside product development, I was building a marketing flywheel. I couldn't keep emailing influencers and putting out content myself. I had posted twice on LinkedIn to hire interns who could help, and fortunately, I found a few. One intern was particularly outstanding; she took over all tasks related to contacting influencers, negotiating, and rolling out campaigns. She did a phenomenal job finding free and low-cost influencers who advertised our products weekly. Gradually, I was able to step back from this work and leave it entirely to her, freeing up my time to focus on bigger deals and customer feedback.

Another factor that helped us was adding multiple languages. After reading Chegg’s annual report and speaking with many former Quizlet employees, I learned that both companies heavily focused on North America and the UK, with only 15% of their customers coming from abroad. This presented a golden opportunity. One of the best uses of large language models is translation, allowing our app to be localized to any country and language within minutes. Influencers in certain parts of the world are also less expensive, creating what felt like business arbitrage, and we capitalized on it fully.

As an aside, Googling your competitors is not enough. Former employees, blog posts, and speeches by their executive team is the best place to get juicy competitive intel 😋.

We achieved a significant breakthrough in Poland, with thousands of students signing up and expressing gratitude for what we built. Our feedback forms consistently showed us how disappointed students would be if we ceased operations. Retention numbers also significantly increased in these markets. We now had a solid product and could name several users who would be disappointed if we stopped. Additionally, we had data on which segments, in which countries, really appreciated our product.

Upping the price: 80k - 100k

While things looked promising, churn remained a challenge. The price was low enough that a user could utilize our product for a month, create the flashcards, and then leave. This also reduced their incentive to provide feedback and help improve the product. Although we were one of the most affordable options on the market, this strategy didn’t contribute to long-term customer retention. We decided to quadruple the price for monthly subscriptions, introduce a cost-effective semester plan, and offer a ludicrously cheap annual plan. Additionally, we implemented a paywall for several features, while offering "easter egg" discounts on others.

Reading blog posts like this might give the impression that all these changes were implemented simultaneously, but that's never the case. Throughout 2024, we ran experiments weekly. One of the best pieces of advice I received from Furqan Rydhan (Founders Inc, Thirdweb) was to cohort each experiment weekly to measure results accurately. This approach allows for roughly 50 significant experiments per year, unless you're testing different variables. We tested free trials, 50% discounts, and variable pricing based on school type and country. Each week brought a new experiment for a new cohort, and what proved most effective was our current pricing strategy. This also meant that at the end of every week, I was analyzing data from Posthog, Stripe, and various internal tools. Being a former product manager sometimes comes in handy. 😅

Raising the prices while continuing the marketing flywheel led to remarkable outcomes. People signed up for longer terms, the annual plan became popular, and revenue doubled every month. Although some students expressed concerns, we offered our product for free to many nonprofits, students from emerging economies, and teachers. The majority of students not only paid but also sent us emails expressing gratitude for building this tool. We initially maintained a testimonial page with thank-yous from students, but after accumulating dozens, it became too cumbersome to keep updated. Was that a humble brag? 😘

The great beyond: 100k + users

When I look at other AI education tools in this space, I see that we have a long way to go. We are constantly iterating on marketing and the product, and we understand that the consumer educational market is extremely seasonal. We will soon release a mobile app and focus on standardized tests over the summer. AI is also improving, which means we can personalize the content more effectively and provide a broader range of learning options.

Paul Graham wrote a blog post about how to think for yourself. Part of the post centered around the theme of curiosity:

Curiosity led me to experiment with large language models, sparking my interest in learning how to code and eventually develop small applications. This same curiosity introduced me to some of the people I work with today. It also drove me to explore marketing and medical curriculums, ultimately leading to the creation of a business. When I look at the Stripe charts or user graphs, I realize that I'm not just measuring the success of the company, but also how much I have learned. Ultimately, a life well-lived is one where you pursue your curiosities until you no longer can. Perhaps that’s what will keep us distinct from our AI overlords 🤖.

I hope you’ve gained some insights from my post. We have been approached by schools, educational institutions, nonprofits, and even corporate L&D teams eager to utilize our features for their stakeholders. We have already formed partnerships with a few, and things are progressing well. I have never transitioned a consumer company to a B2B model, but I am curious to see if I can make it work. If you’re interested in what we offer, I’d love to chat!

Slay School is an AI-powered learning platform. Our technology transforms PDFs, handwritten notes, videos, and any digital medium into personalized study aids. Designed to optimize learning through scientifically-backed spaced repetition and active recall, it also provides immediate feedback, helping students refine their learning strategies. Slay School supports any language, on any device, and incorporates a community aspect, blending study with competition. This makes learning a fun and engaging race rather than a tedious chore. Learn more at SlaySchool.com.

Thanks to Matt, Daniel, Demi, Sophia, Tuba, Raouf, Daniel A, Anson, Amy, Reese & Raisa

‍