"Dumbest idea I've heard" to $100M ARR: Inside the rise of Gamma | Grant Lee (co-founder) - YouTube
Gamma hit $100M ARR in 2 years with 30 people by doing everything backwards: rebuilding after a "successful" launch, betting on micro-influencers over ads, and staying profitable when everyone else was burning cash—in a category investors called impossible.
Read OriginalMy Notes (4)
Caring deeply for 5-10 years
- Start with the problem you actually care about
- Before you do that, ask: can I invest 5-10 years solving this? Do I care deeply enough?
- If you can't answer yes, you'll never overcome the hurdles from other startups and incumbent tools.
Owning the end-to-end workflow
- You need your product to live in customers' brains as the default tool. When they think "I need to create a presentation," they come to you.
- The experience from when they start creating to when they ship it to their boss needs to feel magical. You can't do that unless you either know the workflow deeply yourself or care deeply enough to solve it for someone else.
Don't let paid acquisition become your core growth engine
Don't invest in performance marketing until you have word of mouth working. Don't fool yourself into thinking you can solve other problems by ramping up paid ads.
Get the word of mouth piece first so you're coming into performance marketing with tailwinds. Then start ramping it up.
Set a constraint: never let more than 50% of your acquisitions come through paid channels. If that's happening, your core growth engine is broken.
If your core growth engine is broken, you have a leaky bucket. You're spending money building top of funnel but people aren't making it all the way through.
Something else needs to be fixed before you dial up paid marketing. You can spend a little money, but don't dial it up until your core growth engine actually works.
If you rely on paid acquisition as the main growth engine, CAC will keep going up. The more you try to reach new audiences, the more expensive it gets.
Don't assume CAC will stay flat. You'll end up on a treadmill that runs faster and faster.
It's easy to get hooked early when you're investing small amounts. But it becomes almost impossible to get off that treadmill once you're too far in.
For Gamma, even at scale, over 50% of new signups still come through word of mouth. That's a sign the product is working—people are using it and telling others.
The first 30 seconds matter more than anything
They focused on making the first 30 seconds feel magical.
The goal: if someone goes through that onboarding, they need to tell all their friends. That's the bar.
- New users are selfish, vain, and lazy.
- They have no desire to learn a new tool and their attention span is incredibly short.
What can you give them in the first 30 seconds that earns you the next 30 seconds? And then the next 30 seconds?
Give them one egg
Throw a consumer one egg, they can probably catch it. Throw them four or five eggs, they're going to drop all of them.
Founders want to talk about their 5-10 features. The consumer gets totally confused about why they need this thing.
Give them one egg - one first experience. For Gamma: "Create a slide in seconds." That's the egg. Is that compelling? Some people will still opt out, but for people who catch it, you've solved a real problem.
You've given them enough so they'll stick around and keep playing with your product.
Pull forward what is the most magical thing about your product. Sometimes founders think about 5-10 features, but maybe there's only one thing that actually differentiates you.
Focus on that one thing in the first 30 seconds. You can build on it over time once you've earned the right to keep their attention.
They had an internal mantra: make the first 30 seconds dead simple for someone to create content and dead simple for them to share it.
Everything they did for those first few minutes was about removing friction so users could create and share.
Forced them to think: what is the absolute most magical thing we can show someone immediately? Not eventually, not after they learn the tool - immediately.
Scott Belsky talks about the first smile in the first 15 minutes. Grant took it even further - the first 30 seconds.
If you can't earn those first 30 seconds, you're never going to build a word of mouth machine. Users will leave before they experience the value.
Summary used for search
TLDR
• After winning Product Hunt product of the month, they knew they didn't have PMF because growth plateaued—spent 3 months rebuilding the entire onboarding to make the "first 30 seconds magical," then hit 20K signups/day organically
• Influencer playbook: personally onboarded thousands of micro-influencers ($100-1K each), let them tell the story in their voice, open-sourced the entire brand so creators could easily use it—90% of reach came from 10% of creators
• Over 50% of growth is still word-of-mouth; set hard constraint that paid acquisition can never exceed 50% or "your core growth engine is broken"
• Rapid testing loop: idea in morning → functional prototype → 20 real users testing via Voice Panel by afternoon → iterate by evening (saves months of building wrong things)
• "GPT wrapper" success formula: use 20+ models orchestrated across the workflow, go deep on one problem you care about for 10 years, not broad AI application
In Detail
Grant Lee's journey with Gamma challenges nearly every conventional startup playbook. When they won Product Hunt's product of the day, week, and month, most founders would celebrate and scale. Instead, Grant recognized they didn't have product-market fit—growth was plateauing, word-of-mouth wasn't happening. They made a bet-the-company decision to spend 3-4 months completely rebuilding the onboarding experience, focusing obsessively on making the first 30 seconds "magical." The framework: treat users as "selfish, vain, and lazy"—give them one egg to catch (one clear value prop), not five. After relaunching with AI at the core of onboarding, they went from hundreds to 20,000 signups per day, all organic.
Their growth engine centers on a counterintuitive approach to influencer marketing. Instead of paying big creators $50K for scripted posts, they personally onboarded thousands of micro-influencers at $100-1,000 each. Grant jumped on calls with each one early on, helping them understand Gamma deeply so they could tell the story authentically in their own voice. They open-sourced their entire brand (brand.gamma.app) so creators could easily replicate their visual style. The math: 90% of reach came from 10% of creators, but you can't predict which 10%—so cast a wide net. LinkedIn converted 4-5x better than other platforms. The key insight: influencers want to be in their audience's "Dunbar number" of 150 trusted people—when they recommend something, it doesn't feel like an ad, it feels like a friend's suggestion.
The company maintains an iron rule: over 50% of growth must come from word-of-mouth and organic channels. If paid acquisition ever exceeds 50%, "your core growth engine is broken." This forced them to nail the product experience first. They also pioneered a rapid testing loop: have an idea in the morning, build a functional prototype (often with AI coding tools), recruit 20 real users through platforms like Voice Panel, watch them struggle with it by afternoon, and iterate by evening. This saves months of building features nobody wants.
On the "GPT wrapper" criticism: Gamma uses 20+ different models orchestrated across their workflow—different models for outlining, visual generation, editing, etc. The defensibility comes from going incredibly deep on one specific workflow (presentations) they care about solving for 10+ years, not from model ownership. They stumbled into monetization when users demanded to pay after burning through free credits, ran a Van Westendorp survey, and landed at ~$20/month (anchored to ChatGPT's pricing). They hit $1M ARR and profitability within months.
Their hiring philosophy: "hire painfully slowly." All 10 of their first employees are still there after 5 years. They hire generalists who can wear multiple hats, and everyone is a "player-coach"—no pure managers, all leaders still do IC work. About 25% of the early team were product designers (unusually high) because they knew they needed to invent new UX patterns for AI. When they find exceptional people, they "bet big"—give them more resources, harder problems, more playing time.