0 to $5M ARR in 3 Months: Roy Lee's Viral Growth Playbook

Roy Lee got kicked out of Harvard, then Columbia, and turned both expulsions into rocket fuel. By engineering viral moments on Twitter and TikTok, he raised $5M in under 24 hours and scaled Cluely to $5M ARR in three months—proof that distribution can be as powerful as the product itself.

Key Facts

• Time to $5M ARR: Approximately 3 months post-launch (launched April 20, 2025) (Roy Lee)
• Seed round raised: $5.3M at $30M valuation, closed in under 24 hours (Roy Lee)
• Monthly short-form views: ~80 million views/month across Instagram and TikTok (Roy Lee)
• UGC creator network: ~400 clippers plus UGC creators; ~500 total content contributors (Roy Lee)
• Interview Coder ARR: ~$2M ARR as a subsidiary product (Roy Lee)

How Roy Lee Engineered Virality From the Start

Roy Lee identified three personal superpowers—viral sense, high risk tolerance, and thick skin for criticism—and built his entire go-to-market around them. He recorded himself cheating on an Amazon interview, posted it publicly, and let the controversy compound into a fundraise.

Most founders build first, then figure out distribution. Roy Lee did the opposite. When his B2B startup Drink (AI sales leads for liquor distributors) failed to gain traction, he returned to a demo he'd shelved: Interview Coder, a tool that overlays AI answers invisibly during live coding interviews. A half-million-view LinkedIn post had already proved the concept had viral potential.

Lee recorded himself passing an Amazon interview using the tool, posted it publicly, and watched Amazon issue copyright strikes, Columbia open disciplinary hearings, and Twitter erupt. Rather than retreat, he leaned in—live-tweeting every escalation. As he explains, the logic was simple: 'If you want to build a company, the answer always comes in doubling down on the thing you are very good at.' For him, that was viral sense.

The controversy wasn't accidental. Lee consciously baited Columbia's administration into suspending him to make the story bigger. By the time he was expelled, he had a captive audience and a fundraising narrative no pitch deck could replicate.

"I knew that I could record myself using this on a big tech interview. I knew it would go viral and I knew that I could tolerate all the hate that came from it." — Roy Lee

"I'm generally a pretty insecure person. I'm really strongly driven by the desire to prove the world that I am better than they think of me." — Roy Lee

The Twitter Algorithm: Roy Lee's Three Buckets of Virality

Roy Lee breaks Twitter virality into three categories: informational (novel news or data), relatable (broad shared experiences), and engageable (content designed to provoke a quote-post reaction). Engageable content is the most controllable and the most powerful for startups.

After roughly 3,000 tweets, Lee claims he can predict impression counts within 25% for any post he publishes. His framework is built around a single mental test: 'Imagine someone quote-posted your tweet because they have such a strong opinion about it. Can you visualize exactly what strong opinion they will have? If you cannot envision that, then your tweet will not go viral.'

Informational virality depends on having genuinely novel data—something only research labs or companies with major launches can reliably produce. Relatable virality (think: 'shipping on a Friday') casts a wide net but converts poorly. Engageable virality is the one any founder can manufacture deliberately, and it almost always requires controversy or shock.

Lee also flags a second requirement for engageability: radical clarity. 'The average American is so much stupider than you think. If a farmer from Kansas picked this up in the daily newspaper, would he be able to read your tweet and understand exactly what you're talking about? 99 percent of the time it's no.' Jargon kills reach.

"I think I might personally know the Twitter algorithm better than any single person in the world right now." — Roy Lee

"Very rarely will you quote something and say, 'I really love this.' Almost always you will quote something and say, 'I feel so strongly negative about this. I have to share my negative opinion.'" — Roy Lee

• Informational: Works if you have genuinely new data or a major product launch.
• Relatable: Wide reach, low conversion—use sparingly to maintain algorithm presence.
• Engageable: Design every post so a critic will want to quote-post it with a strong opinion.
• Clarity test: If a non-technical person can't understand it instantly, rewrite it.
• Algo pool: A viral moment keeps your content surfaced for ~40 hours—post again immediately.

Launching Cluely: Designing a Video to Go Viral, Not to Sell

Cluely's launch video never showed the actual product working. Instead, Lee designed every shot to be screenshot-worthy and every scene to invite a strong quote-post reaction. The result was millions of views and a waitlist—before the product even worked reliably.

The Cluely launch video features Lee on a date with a woman, secretly using AI assistance throughout the conversation. The concept was borrowed from the viral playbook of Friend.com: cinematic production, a Black Mirror aesthetic, and zero literal product demo. Lee's framing: 'We don't even have to show the product at all. We just have to show the idea of the product—real-time AI that nobody knows you're using.'

Every creative decision was made through the lens of engageability. The date setting breaks a social contract people feel strongly about. The 'Bananazon' reference pokes Amazon—the company that triggered his expulsion—rewarding followers who'd been tracking the story. The anime convention line at the end was a calculated appeal to the tech-adjacent Asian audience he'd identified as core to his algo pool.

Lee's advice for founders launching anything: 'Your launch video will probably go more viral than any of your feature demos because it's novel—there's information being spread.' Use that window aggressively, then keep posting reactively to every quote-post and criticism that follows.

"If you go back and watch all the shots, every single shot is screenshottable. You can imagine someone would quote-post with a screenshot and say something." — Roy Lee

"Sales is about a lot more than just being a little salesperson hopping on a call. You have to be cool. Nobody wants to watch an ad ever." — Roy Lee

Building a 400-Person Creator Army: The UGC and Clipper Playbook

Cluely drives ~80 million monthly views on Instagram and TikTok through three channels: clippers who cut podcast content into short clips, UGC creators who make original product content, and influencer partnerships. Creators under 100K followers are underpriced and highly effective.

Lee's creator network operates on a straightforward economic insight: creators with meaningful followings—even up to 100K—have rarely been paid for content. That gap makes them highly responsive to paid opportunities and willing to work on a per-video basis with view-based bonuses.

The operational model is leaner than it sounds. Creators post to fresh accounts dedicated to Cluely content, receive brief creative direction, and are evaluated purely on performance. An in-house content team reviews videos, coaches on format, and redistributes winning templates across the entire creator pool. Underperformers are cut; high performers get more assignments.

For UGC hooks, Lee advocates for absurdist disconnects between the hook and the product. His example: lead with something shocking and unrelated, then reveal it's actually about software. 'The hook is viral in itself, and you can literally do this with any product.' The method works for B2B too—Lee notes Cluely has closed million-dollar enterprise contracts directly attributable to Twitter engagement.

"People get super underpaid right now for their ability to make viral content, and right now this delta has not been captured by anyone." — Roy Lee

"Especially on Twitter, we literally signed million-dollar enterprise contracts just because people think my Twitter is funny." — Roy Lee

• Clippers: 400+ creators cutting podcast clips—output correlates with how quotable the podcast guest is.
• UGC creators: Fresh accounts, per-video pay, view bonuses, fired on poor performance.
• Content coaching: In-house team identifies winning formats and distributes them across all creators.
• Hook formula: Lead with something shocking or absurd, then connect it loosely to the product.

Product Market Fit, Retention, and Where Cluely Goes Next

Lee identified true PMF with Cluely only when live meeting notes tripled retention nearly overnight—weeks after hitting $5M ARR. Revenue growth preceded retention stability, which means virality can mask PMF gaps. Data integrations and agentic actions are the next product frontier.

Cluely's rapid ARR growth obscured a real retention problem: many users couldn't fully adopt the novel overlay interface. Lee is candid about this. The product's value becomes clear in meetings, but the habit of querying it in real time takes time to develop. The live meeting notes feature, launched recently, was the first signal of genuine stickiness—retention tripled almost immediately.

For Interview Coder, PMF was unambiguous from day one. As Lee puts it: 'If it works, you have PMF. If it will give you the answer and you won't get caught—that's PMF.' Cluely's PMF was murkier, proving that viral distribution can generate revenue well before the product is truly indispensable.

The roadmap Lee describes is ambitious: deep data integrations (email, Google Drive, Notion, Salesforce, HubSpot) that make Cluely context-aware across a user's entire work life, followed by agentic actions that let Cluely click around the computer autonomously. The long-term vision is a new graphical interface for computing—one that positions Cluely to distribute whatever hardware form factor (glasses, pendants, chips) wins the next platform war.

"We launched live meeting notes and retention tripled pretty much overnight. I think that's the point where we realized, okay, we have something that is genuinely working." — Roy Lee

"The ability to snap your fingers, record a 10-second video, and have that be seen by 100 million people—this has never before been possible in human history." — Roy Lee

Three Types of Twitter Virality (Roy Lee's Framework)