Zack Hargett
Where to hear and read about Zack Hargett2 episodes and tests
SuperwallCOPY this ai app's $6.7m/yr marketing strategy
Sub Club$6.7M ARR, No Paid Ads, and an Exit to Quizlet in 2 Years – Brett Bauman & Zack Hargett, CoconoteZack Hargett is a product builder and co-founder of Coconote, an AI lecture note-taking app he built with Brett Bauman, both former Loom employees, and sold to Quizlet. The lesson most worth taking from him is to frame a product as the solution to a real problem for buyers with intent, because attention alone does not pay.
if you frame your product as a novel toy, people are going to treat it like a toy.
$6.7M ARR, No Paid Ads, and an Exit to Quizlet in 2 Years – Brett Bauman & Zack Hargett, Coconote
Zack Hargett: the story and the lessons
1Zack Hargett's road to Coconote
Hargett came to Coconote after shutting down a venture-backed studio. In 2022 he had just closed Scout, a venture-backed studio that built for high school and college students. He founded it solo with no pressure to monetize.
Earlier side projects included Crew, a music app, and Essence, a Chrome extension built on early language models that summarized blog posts and videos.
Crew taught him that a good product does not bring its own users. Crew used UGC-style TikTok and Reels to drive hundreds of thousands of sign-ups, and he concluded that 'if you build it, they will not come'. He then taught himself marketing.
At Scout in 2021, the first TikTok on a new account got 8M views and drove 150,000 signups in two days.
Coconote blended small experiments built with one student in mind. He pictured his step-sister Emma, entering college as a civil engineering major, when he built the summarizer. In 2023 he and Brett tried replacing Siri with GPT-4 and saw the power of voice.
Existing note-takers targeted business, and the student web apps were not taken seriously. He calls Coconote Charlie Munger's 'simple ideas taken seriously'. Apple gave trouble approving the Siri replacement and it went nowhere.
A Loom offsite put him together with co-founder Brett Bauman. Mutual friends at a Loom offsite in Los Angeles told the two to meet, though they had not overlapped at the company. Hargett brought design, product and marketing, and Brett brought engineering.
He pitched ideas and Brett answered 'maybe' until he agreed to build one. One source describes Hargett as an iOS engineer at Loom, and the other says he left earlier and helped start the mobile team.
2How Hargett launched Coconote and got to revenue fast
Hargett charged from day one with a free trial and a scoped-down app. Coconote launched in April 2024 with one free note and then a trial, which works as a hard paywall. Some audio recordings were lost at first, and much of the product still looks as it did then.
He watched the RevenueCat chart from the first day, until his wife staged an intervention. Paying customers told him more than usage events in Mixpanel did.
Coconote compounded from $100K ARR in 45 days to $6.7M. It passed $100K ARR in 45 days, $1M in 4 months, about $2M in month five and $5M after a year, with no VC. It ran at over 50% EBITDA, and each founder put in about $10K up front.
These figures are the founders' own claims, with no outside check.
He treats early momentum as fuel for the next move. Hargett cites a line he attributes to Sam Altman: momentum is like oxygen for a startup. Early revenue paid for creators and kept energy high.
Slow Apple payouts were manageable because the free tier was tight and costs stayed small.
He did not optimize AI costs early. Free use was minimal, startup credits covered the first year, and audio transcription cost more than the language model calls.
Model costs fall over time, and worrying about margins early risks poor decisions. Giving away tokens early was part of acquisition cost.
Coconote launched into a market with almost no AI study apps. AI note-takers on the App Store were aimed at professionals, and the student web apps were not serious. The team chose to execute properly.
A podcast host noted the app could not have been built two years earlier, which gave a first-mover edge. How much growth came from early timing is not shown.
3Hargett's creator-led marketing machine
Hargett hires undiscovered marketers, not influencers or agency-repped creators. He looked for creators with 5 to 10K followers, good engagement and a personal Gmail in their bio. An agency email means you are too late.
In the first two months he tried about 30 creators and only three made a big difference. One of them ran the main account and generated hundreds of millions of views. Talent also came from Handshake, DMs, referrals and a Spanish tutor who became a top creator within nine months.
He keeps the creator team small, aiming for Navy Seals, not the Navy. The team peaked at 10 to 12 contractors before the acquisition, with one coach per 12 creators. Hargett met each creator individually, usually deciding within 15 to 30 minutes.
The Navy Seals line comes from Brian Chesky. Paying per view can inflate bills on funny videos that do not convert.
He tells creators what converted and measures during the viral spike. A video with 20 to 30 million views shows its effect on signups and revenue within a couple of days. Creators then learn which creative paid and which did not.
Coaches mostly review content and walk creators through metrics, with no fixed curriculum.
Creators are judged on a pyramid: want to win, study virality, create. The base is showing up, engaging in Slack, being on time and keeping the camera on. The middle is guessing why videos went viral with the view count hidden, and the top is creating your own scripts and trends.
There are no quizzes, and top creators are trendsetters.
Creators got eye-catching, creator-only app screens to film. The recording and transcribing screens in the viral videos were not real app screens. A lecture-hall video using them hit 18 million views, about 99% organic.
Creators sit in Slack with engineers and ask for more hidden screens.
4What converts: Hargett on toys, problems and identity
Hargett prefers 10 million views that convert over 40 million that do not. A PDF to brain rot feature drew 100 to 200 million views and about 3,000 trial starts, which converted to roughly $25,000 against a $360,000 ceiling.
Novelty pulls in low-intent traffic. In his words: 'We would rather have 10 million views on a video that converts well, because it's actually targeting our target audience'.
A 41M-view video drove no revenue because it made the app a toy. The video had 41M views and 4.5M likes. People expect digital things to be free, so a toy gets treated like one.
PDFtoBrain was meant as a Trojan horse into the product. It has since been sunset.
He matches content style to where the product is in its life. A shock hook plus demo worked in March to May 2024 while the product was novel. The brain rot videos 6 to 8 months later reached a younger audience with less disposable income.
He calls the early format lazy but effective, and said it only works while the product is new.
Identity content and problem-solving content convert best. People pick products that match what they say about themselves, such as 'I am fit, so I use Ladder' or 'I am a runner, so I use Strava'.
Hargett says producing UGC hooks that convert is hard right now.
He marketed to the buyer, not only the student user. Early viral videos used 'my mom just changed my life'. Creators were told to picture a mother and daughter at orientation and ask who would be offered the $130 app.
The team later stepped back from moms, since the audiences overlap and the real aim is whoever has the most buying intent. $130 a year looks small next to $40K to $100K in yearly tuition.
He says customers' own words back to them. Most customers described Coconote as a way to never miss a key detail, and that line went on the first App Store preview screen. The first screenshot also says 'great for ADHD'.
He compares it to political campaigns repeating what resonates, and said he had never listened to customers this way before.
Coconote's audience turned out to be lifelong learners, not just 18 to 22 year olds. Some customers study outside accredited universities, and others are at a university but outside the usual age range.
The aim is a great tutor for anyone with a phone. Flashcards and notes suit health and biology better than computer science or math, so those subjects are underserved.
5Hargett on pricing and paid ads
He charged a premium to students, a group seen as hard to monetize. Coconote launched at $99 a year and $19.99 a month. An early test at $129 a year brought more users and more revenue, and about 80% of subscribers take the annual plan.
Comparable utilities cost about a third as much. He did not let the student archetype cap the price, and fewer higher-paying customers suit a bootstrapped company. No sample sizes are given for the $129 test.
Premium pricing rests on trust that recordings are not lost. Users stop taking their own notes, so a lost recording is costly. Cheaper apps can lose audio when a phone call comes in, since Apple's audio APIs make reliability hard.
The app also gives value passively, as peace of mind while it records.
He sees pricing power as scarcity, and clones as lead generation. Two years earlier one great AI model had pricing authority, and now comparable models exist. Hargett calls clones like 'Cocoonote' mosquitoes that may send users back after they try a cheaper one.
Apps are never true commodities, so there is always some difference to compete on.
Paid ads never worked because neither founder owned the channel. The goal was first-purchase profitability on ads, and the team never reached it. Neither founder gave ads half their time.
Clips cut by an outside agency from raw creator footage beat the organic-style creative but still fell short. His rule: 'there's very few growth channels that you can halfass if you want them to work'.
6Hargett's onboarding, paywall and cancel-flow playbook
Doubling onboarding length raised trial starts by 16%. Testing one or two screens at a time gave small wins. A big jump from about five screens to a longer flow, adding personalization and social proof, lifted trial starts.
The sources give about 13 screens in one case and about 15 in the other. They could not say which screens helped most, and no baseline is given.
Moving login after the paywall was the biggest onboarding win. Apple and Google accounts already identify buyers, so no new account is needed to pay. Login as the first screen had caused about 10% drop-off.
It needs plumbing because store purchase state and account state are separate. Engineers build login first and it stays at the front out of habit.
He shows the paywall after the first core action. Users can record a full lecture, an hour or more, before the paywall appears, and that was a big win over gating the recording.
Letting users invest in the product first lifts trial conversion.
A seven-day trial extension at cancel beat discounts and pauses. He tested a 30% discount, a 3-month pause aimed at students who stop paying in May or June, and a prompt asking whether they just need more time. The extension won by far.
Web cancellation flows kept 25% of would-be cancellers in one account, and the extension saved 27% in another, though a smaller share went on to pay. Keeping people on auto-renew matters because it is hard to win them back.
7How Hargett handled the sale to Quizlet
Quizlet reached out when Coconote was only a few months old. Hargett's first instinct was to stay heads down and he doubted anything could come of the call. Brett slept on it and argued for taking it, and they talked with a handful of companies.
His advice is to take such calls only for a company you admire, and to ask for a ballpark range after a few calls.
He hired an investment banker so the founders could keep building. Chris Parker at Creative Partners ran the process through closing. Hargett said that losing focus on customers, building and marketing shows up in negotiations.
He called the banker 'our third co-founder a little bit'. The founders took every call for a company they respected.
The sale took about a year while revenue roughly doubled. Most of it was waiting, with the banker preparing outreach while the founders did nothing active. The deal closed in December.
The sources describe the exit as coming about 18 months after launch in one case and about two years in the other. Deal terms are never shared.
He told no one how far the talks had gone, not even his wife. Only the two founders knew. Hargett told his wife in a Taco Bell parking lot the moment the deal closed, and engineers and creators learned on the day of close.
He took advice from founders who had been through M&A: 'a great leader shelters other people'. Deals can fall apart in the last week.
Quizlet gave Hargett room to think longer term. As an independent company they expected returns on marketing within a month or two. A larger company allows bigger bets, and Hargett now focuses on the health he ignored for two years.
Brett still works on Coconote engineering and on subscription growth at Quizlet. Bootstrapping had real trade-offs on reinvestment.
8How Hargett picks products and reads AI
He asks what AI can do now that it could not 3 to 6 months ago. Hargett uses this test to find new products. Runna's real-time running coach is his example, since before it a human would have had to coach from your metrics.
The best new bets are products that were impossible a few months earlier.
Without a deep moat, he says to keep shipping wow moments. A hit can get a hundred copies within a week. Unless you hold one or two of the seven powers, such as brand or network effects, team execution is the growth engine.
Wow matters most for products sold through short-form video, since the aha is what the ad shows. He names Quizlet's brand, built over almost 20 years, as one such moat.
Coconote was built around a daily habit that piles up notes. Recording lectures happens daily or a few times a week, and notes accumulate across courses such as physics 1, 2 and 3, which locks users in.
Brett looks for repeatable use cases tied to identity, citing Ladder, AllTrails and Strava.
He shelters early ideas from smart skeptics. Smart friends would have said no one will pay for a student app. He asks 'what happens if it does work' instead of why it will not.
Pricing norms formed on products like weather apps may not apply to a new AI product. Coconote charged a premium and grew to $6.7M ARR.
Coconote applies the big labs' models to learners instead of competing. The mission is to give learners superpowers, deliberately vague, while foundation model makers do the core work. The goal is a great tutor for anyone with a phone.
He pictures a child of Jeff Bezos getting a top tutor at the door, and says everyone else should get one too.
A four-engineer team of flexible generalists shipped native apps. An iOS engineer also does back-end work, and a web engineer does too. AI on top of experience speeds the work, and all apps are native with no cross-platform code.
The team supports millions of users at over 50% EBITDA.
COPY this ai app's $6.7m/yr marketing strategy
Two ex-Loom employees explain how they grew Coconote to 6.7 million ARR in about 18 months, bootstrapped, using a small team of part-time creators, targeted short-form video, a longer onboarding and trial-extension anti-churn flow. The result: 100K ARR in 45 days, 5 million in a year, over 50% EBITDA, and an acquisition by Quizlet.
Hargett came to Coconote after shutting down a venture-backed studio. In 2022 he had just closed Scout, a venture-backed studio that built for high school and college students. He founded it solo with no pressure to monetize.
Earlier side projects included Crew, a music app, and Essence, a Chrome extension built on early language models that summarized blog posts and videos.
Crew taught him that a good product does not bring its own users. Crew used UGC-style TikTok and Reels to drive hundreds of thousands of sign-ups, and he concluded that 'if you build it, they will not come'. He then taught himself marketing.
At Scout in 2021, the first TikTok on a new account got 8M views and drove 150,000 signups in two days.
Coconote blended small experiments built with one student in mind. He pictured his step-sister Emma, entering college as a civil engineering major, when he built the summarizer. In 2023 he and Brett tried replacing Siri with GPT-4 and saw the power of voice.
Existing note-takers targeted business, and the student web apps were not taken seriously. He calls Coconote Charlie Munger's 'simple ideas taken seriously'. Apple gave trouble approving the Siri replacement and it went nowhere.
A Loom offsite put him together with co-founder Brett Bauman. Mutual friends at a Loom offsite in Los Angeles told the two to meet, though they had not overlapped at the company. Hargett brought design, product and marketing, and Brett brought engineering.
He pitched ideas and Brett answered 'maybe' until he agreed to build one. One source describes Hargett as an iOS engineer at Loom, and the other says he left earlier and helped start the mobile team.
Coconote compounded from $100K ARR in 45 days to $6.7M. It passed $100K ARR in 45 days, $1M in 4 months, about $2M in month five and $5M after a year, with no VC. It ran at over 50% EBITDA, and each founder put in about $10K up front.
These figures are the founders' own claims, with no outside check.
He treats early momentum as fuel for the next move. Hargett cites a line he attributes to Sam Altman: momentum is like oxygen for a startup. Early revenue paid for creators and kept energy high.
Slow Apple payouts were manageable because the free tier was tight and costs stayed small.
Coconote launched into a market with almost no AI study apps. AI note-takers on the App Store were aimed at professionals, and the student web apps were not serious. The team chose to execute properly.
A podcast host noted the app could not have been built two years earlier, which gave a first-mover edge. How much growth came from early timing is not shown.
Hargett hires undiscovered marketers, not influencers or agency-repped creators. He looked for creators with 5 to 10K followers, good engagement and a personal Gmail in their bio. An agency email means you are too late.
In the first two months he tried about 30 creators and only three made a big difference. One of them ran the main account and generated hundreds of millions of views. Talent also came from Handshake, DMs, referrals and a Spanish tutor who became a top creator within nine months.
He keeps the creator team small, aiming for Navy Seals, not the Navy. The team peaked at 10 to 12 contractors before the acquisition, with one coach per 12 creators. Hargett met each creator individually, usually deciding within 15 to 30 minutes.
The Navy Seals line comes from Brian Chesky. Paying per view can inflate bills on funny videos that do not convert.
Creators are judged on a pyramid: want to win, study virality, create. The base is showing up, engaging in Slack, being on time and keeping the camera on. The middle is guessing why videos went viral with the view count hidden, and the top is creating your own scripts and trends.
There are no quizzes, and top creators are trendsetters.
Creators got eye-catching, creator-only app screens to film. The recording and transcribing screens in the viral videos were not real app screens. A lecture-hall video using them hit 18 million views, about 99% organic.
Creators sit in Slack with engineers and ask for more hidden screens.
Hargett prefers 10 million views that convert over 40 million that do not. A PDF to brain rot feature drew 100 to 200 million views and about 3,000 trial starts, which converted to roughly $25,000 against a $360,000 ceiling.
Novelty pulls in low-intent traffic. In his words: 'We would rather have 10 million views on a video that converts well, because it's actually targeting our target audience'.
He matches content style to where the product is in its life. A shock hook plus demo worked in March to May 2024 while the product was novel. The brain rot videos 6 to 8 months later reached a younger audience with less disposable income.
He calls the early format lazy but effective, and said it only works while the product is new.
He says customers' own words back to them. Most customers described Coconote as a way to never miss a key detail, and that line went on the first App Store preview screen. The first screenshot also says 'great for ADHD'.
He compares it to political campaigns repeating what resonates, and said he had never listened to customers this way before.
Coconote's audience turned out to be lifelong learners, not just 18 to 22 year olds. Some customers study outside accredited universities, and others are at a university but outside the usual age range.
The aim is a great tutor for anyone with a phone. Flashcards and notes suit health and biology better than computer science or math, so those subjects are underserved.
Doubling onboarding length raised trial starts by 16%. Testing one or two screens at a time gave small wins. A big jump from about five screens to a longer flow, adding personalization and social proof, lifted trial starts.
The sources give about 13 screens in one case and about 15 in the other. They could not say which screens helped most, and no baseline is given.
Moving login after the paywall was the biggest onboarding win. Apple and Google accounts already identify buyers, so no new account is needed to pay. Login as the first screen had caused about 10% drop-off.
It needs plumbing because store purchase state and account state are separate. Engineers build login first and it stays at the front out of habit.
He shows the paywall after the first core action. Users can record a full lecture, an hour or more, before the paywall appears, and that was a big win over gating the recording.
Letting users invest in the product first lifts trial conversion.
A seven-day trial extension at cancel beat discounts and pauses. He tested a 30% discount, a 3-month pause aimed at students who stop paying in May or June, and a prompt asking whether they just need more time. The extension won by far.
Web cancellation flows kept 25% of would-be cancellers in one account, and the extension saved 27% in another, though a smaller share went on to pay. Keeping people on auto-renew matters because it is hard to win them back.
Quizlet reached out when Coconote was only a few months old. Hargett's first instinct was to stay heads down and he doubted anything could come of the call. Brett slept on it and argued for taking it, and they talked with a handful of companies.
His advice is to take such calls only for a company you admire, and to ask for a ballpark range after a few calls.
He hired an investment banker so the founders could keep building. Chris Parker at Creative Partners ran the process through closing. Hargett said that losing focus on customers, building and marketing shows up in negotiations.
He called the banker 'our third co-founder a little bit'. The founders took every call for a company they respected.
The sale took about a year while revenue roughly doubled. Most of it was waiting, with the banker preparing outreach while the founders did nothing active. The deal closed in December.
The sources describe the exit as coming about 18 months after launch in one case and about two years in the other. Deal terms are never shared.
Quizlet gave Hargett room to think longer term. As an independent company they expected returns on marketing within a month or two. A larger company allows bigger bets, and Hargett now focuses on the health he ignored for two years.
Brett still works on Coconote engineering and on subscription growth at Quizlet. Bootstrapping had real trade-offs on reinvestment.
He asks what AI can do now that it could not 3 to 6 months ago. Hargett uses this test to find new products. Runna's real-time running coach is his example, since before it a human would have had to coach from your metrics.
The best new bets are products that were impossible a few months earlier.
Without a deep moat, he says to keep shipping wow moments. A hit can get a hundred copies within a week. Unless you hold one or two of the seven powers, such as brand or network effects, team execution is the growth engine.
Wow matters most for products sold through short-form video, since the aha is what the ad shows. He names Quizlet's brand, built over almost 20 years, as one such moat.
Coconote was built around a daily habit that piles up notes. Recording lectures happens daily or a few times a week, and notes accumulate across courses such as physics 1, 2 and 3, which locks users in.
Brett looks for repeatable use cases tied to identity, citing Ladder, AllTrails and Strava.
Coconote applies the big labs' models to learners instead of competing. The mission is to give learners superpowers, deliberately vague, while foundation model makers do the core work. The goal is a great tutor for anyone with a phone.
He pictures a child of Jeff Bezos getting a top tutor at the door, and says everyone else should get one too.
A four-engineer team of flexible generalists shipped native apps. An iOS engineer also does back-end work, and a web engineer does too. AI on top of experience speeds the work, and all apps are native with no cross-platform code.
The team supports millions of users at over 50% EBITDA.
$6.7M ARR, No Paid Ads, and an Exit to Quizlet in 2 Years – Brett Bauman & Zack Hargett, Coconote
Two ex-Loom builders launched an AI lecture note-taking app with a free trial from day one, grew it through a small team of content creators, and used a long onboarding and trial-extension cancellation offers to convert and retain students. They reached $100K ARR in 45 days, $1M in 4 months and $6.7M ARR with no paid ads, then sold to Quizlet in about two years.
Hargett came to Coconote after shutting down a venture-backed studio. In 2022 he had just closed Scout, a venture-backed studio that built for high school and college students. He founded it solo with no pressure to monetize.
Earlier side projects included Crew, a music app, and Essence, a Chrome extension built on early language models that summarized blog posts and videos.
Crew taught him that a good product does not bring its own users. Crew used UGC-style TikTok and Reels to drive hundreds of thousands of sign-ups, and he concluded that 'if you build it, they will not come'. He then taught himself marketing.
At Scout in 2021, the first TikTok on a new account got 8M views and drove 150,000 signups in two days.
Coconote blended small experiments built with one student in mind. He pictured his step-sister Emma, entering college as a civil engineering major, when he built the summarizer. In 2023 he and Brett tried replacing Siri with GPT-4 and saw the power of voice.
Existing note-takers targeted business, and the student web apps were not taken seriously. He calls Coconote Charlie Munger's 'simple ideas taken seriously'. Apple gave trouble approving the Siri replacement and it went nowhere.
A Loom offsite put him together with co-founder Brett Bauman. Mutual friends at a Loom offsite in Los Angeles told the two to meet, though they had not overlapped at the company. Hargett brought design, product and marketing, and Brett brought engineering.
He pitched ideas and Brett answered 'maybe' until he agreed to build one. One source describes Hargett as an iOS engineer at Loom, and the other says he left earlier and helped start the mobile team.
Hargett charged from day one with a free trial and a scoped-down app. Coconote launched in April 2024 with one free note and then a trial, which works as a hard paywall. Some audio recordings were lost at first, and much of the product still looks as it did then.
He watched the RevenueCat chart from the first day, until his wife staged an intervention. Paying customers told him more than usage events in Mixpanel did.
Coconote compounded from $100K ARR in 45 days to $6.7M. It passed $100K ARR in 45 days, $1M in 4 months, about $2M in month five and $5M after a year, with no VC. It ran at over 50% EBITDA, and each founder put in about $10K up front.
These figures are the founders' own claims, with no outside check.
He treats early momentum as fuel for the next move. Hargett cites a line he attributes to Sam Altman: momentum is like oxygen for a startup. Early revenue paid for creators and kept energy high.
Slow Apple payouts were manageable because the free tier was tight and costs stayed small.
He did not optimize AI costs early. Free use was minimal, startup credits covered the first year, and audio transcription cost more than the language model calls.
Model costs fall over time, and worrying about margins early risks poor decisions. Giving away tokens early was part of acquisition cost.
Coconote launched into a market with almost no AI study apps. AI note-takers on the App Store were aimed at professionals, and the student web apps were not serious. The team chose to execute properly.
A podcast host noted the app could not have been built two years earlier, which gave a first-mover edge. How much growth came from early timing is not shown.
Hargett hires undiscovered marketers, not influencers or agency-repped creators. He looked for creators with 5 to 10K followers, good engagement and a personal Gmail in their bio. An agency email means you are too late.
In the first two months he tried about 30 creators and only three made a big difference. One of them ran the main account and generated hundreds of millions of views. Talent also came from Handshake, DMs, referrals and a Spanish tutor who became a top creator within nine months.
He keeps the creator team small, aiming for Navy Seals, not the Navy. The team peaked at 10 to 12 contractors before the acquisition, with one coach per 12 creators. Hargett met each creator individually, usually deciding within 15 to 30 minutes.
The Navy Seals line comes from Brian Chesky. Paying per view can inflate bills on funny videos that do not convert.
He tells creators what converted and measures during the viral spike. A video with 20 to 30 million views shows its effect on signups and revenue within a couple of days. Creators then learn which creative paid and which did not.
Coaches mostly review content and walk creators through metrics, with no fixed curriculum.
A 41M-view video drove no revenue because it made the app a toy. The video had 41M views and 4.5M likes. People expect digital things to be free, so a toy gets treated like one.
PDFtoBrain was meant as a Trojan horse into the product. It has since been sunset.
Identity content and problem-solving content convert best. People pick products that match what they say about themselves, such as 'I am fit, so I use Ladder' or 'I am a runner, so I use Strava'.
Hargett says producing UGC hooks that convert is hard right now.
He marketed to the buyer, not only the student user. Early viral videos used 'my mom just changed my life'. Creators were told to picture a mother and daughter at orientation and ask who would be offered the $130 app.
The team later stepped back from moms, since the audiences overlap and the real aim is whoever has the most buying intent. $130 a year looks small next to $40K to $100K in yearly tuition.
He says customers' own words back to them. Most customers described Coconote as a way to never miss a key detail, and that line went on the first App Store preview screen. The first screenshot also says 'great for ADHD'.
He compares it to political campaigns repeating what resonates, and said he had never listened to customers this way before.
Coconote's audience turned out to be lifelong learners, not just 18 to 22 year olds. Some customers study outside accredited universities, and others are at a university but outside the usual age range.
The aim is a great tutor for anyone with a phone. Flashcards and notes suit health and biology better than computer science or math, so those subjects are underserved.
He charged a premium to students, a group seen as hard to monetize. Coconote launched at $99 a year and $19.99 a month. An early test at $129 a year brought more users and more revenue, and about 80% of subscribers take the annual plan.
Comparable utilities cost about a third as much. He did not let the student archetype cap the price, and fewer higher-paying customers suit a bootstrapped company. No sample sizes are given for the $129 test.
Premium pricing rests on trust that recordings are not lost. Users stop taking their own notes, so a lost recording is costly. Cheaper apps can lose audio when a phone call comes in, since Apple's audio APIs make reliability hard.
The app also gives value passively, as peace of mind while it records.
He sees pricing power as scarcity, and clones as lead generation. Two years earlier one great AI model had pricing authority, and now comparable models exist. Hargett calls clones like 'Cocoonote' mosquitoes that may send users back after they try a cheaper one.
Apps are never true commodities, so there is always some difference to compete on.
Paid ads never worked because neither founder owned the channel. The goal was first-purchase profitability on ads, and the team never reached it. Neither founder gave ads half their time.
Clips cut by an outside agency from raw creator footage beat the organic-style creative but still fell short. His rule: 'there's very few growth channels that you can halfass if you want them to work'.
Doubling onboarding length raised trial starts by 16%. Testing one or two screens at a time gave small wins. A big jump from about five screens to a longer flow, adding personalization and social proof, lifted trial starts.
The sources give about 13 screens in one case and about 15 in the other. They could not say which screens helped most, and no baseline is given.
Moving login after the paywall was the biggest onboarding win. Apple and Google accounts already identify buyers, so no new account is needed to pay. Login as the first screen had caused about 10% drop-off.
It needs plumbing because store purchase state and account state are separate. Engineers build login first and it stays at the front out of habit.
A seven-day trial extension at cancel beat discounts and pauses. He tested a 30% discount, a 3-month pause aimed at students who stop paying in May or June, and a prompt asking whether they just need more time. The extension won by far.
Web cancellation flows kept 25% of would-be cancellers in one account, and the extension saved 27% in another, though a smaller share went on to pay. Keeping people on auto-renew matters because it is hard to win them back.
Quizlet reached out when Coconote was only a few months old. Hargett's first instinct was to stay heads down and he doubted anything could come of the call. Brett slept on it and argued for taking it, and they talked with a handful of companies.
His advice is to take such calls only for a company you admire, and to ask for a ballpark range after a few calls.
He hired an investment banker so the founders could keep building. Chris Parker at Creative Partners ran the process through closing. Hargett said that losing focus on customers, building and marketing shows up in negotiations.
He called the banker 'our third co-founder a little bit'. The founders took every call for a company they respected.
The sale took about a year while revenue roughly doubled. Most of it was waiting, with the banker preparing outreach while the founders did nothing active. The deal closed in December.
The sources describe the exit as coming about 18 months after launch in one case and about two years in the other. Deal terms are never shared.
He told no one how far the talks had gone, not even his wife. Only the two founders knew. Hargett told his wife in a Taco Bell parking lot the moment the deal closed, and engineers and creators learned on the day of close.
He took advice from founders who had been through M&A: 'a great leader shelters other people'. Deals can fall apart in the last week.
Quizlet gave Hargett room to think longer term. As an independent company they expected returns on marketing within a month or two. A larger company allows bigger bets, and Hargett now focuses on the health he ignored for two years.
Brett still works on Coconote engineering and on subscription growth at Quizlet. Bootstrapping had real trade-offs on reinvestment.
Coconote was built around a daily habit that piles up notes. Recording lectures happens daily or a few times a week, and notes accumulate across courses such as physics 1, 2 and 3, which locks users in.
Brett looks for repeatable use cases tied to identity, citing Ladder, AllTrails and Strava.
He shelters early ideas from smart skeptics. Smart friends would have said no one will pay for a student app. He asks 'what happens if it does work' instead of why it will not.
Pricing norms formed on products like weather apps may not apply to a new AI product. Coconote charged a premium and grew to $6.7M ARR.
Coconote applies the big labs' models to learners instead of competing. The mission is to give learners superpowers, deliberately vague, while foundation model makers do the core work. The goal is a great tutor for anyone with a phone.
He pictures a child of Jeff Bezos getting a top tutor at the door, and says everyone else should get one too.