The Pattern Library · cited by Sensei
Patterns are universal in shape, specific in example. The companies named are public; the patterns are how Sensei thinks. Read into every Sensei reading. Cite freely.
Is this worth my time?
When Sensei weighs whether a product is worth a founder's time, these are the studies already on the shelf.
Features get cloned in a quarter. Workflows take a year. Habits compound over three. The moat is the muscle, not the tool.
Figma's files inside Figma became a professional muscle before the multiplayer was matched. Notion's databases became a personal muscle before the block system was matched.
High-growth markets pre-consolidation reward the third player, not the first. Once one incumbent clears a Series B, the cost of entry roughly doubles and the window for a challenger closes in 12–18 months.
Calendly raised $350M only after Doodle had spent a decade teaching schedulers existed. Linear raised Series B 4 years into a market Jira had owned for 20.
Founders underprice because the number makes them flinch. Customers don't flinch — they discount reflexively, then renew.
Superhuman launched at $30/mo when the market was free. Linear at $8/user when Jira was $7. Both held their price through the next round; the founders who started lower never successfully raised it.
Year-one projections are usually wrong by 30% in the wrong direction. Plan for 70% of your top-line estimate — and make sure the math still clears.
ConvertKit's first public milestone: $98K/mo promised, $65K/mo hit. Baremetrics published the same gap. The plans that survived assumed the miss.
Pioneers teach the market a vocabulary the second or third entrant gets to sell into. Category creators usually don't win their own category.
Friendster taught the world 'social network.' Facebook sold it. Evernote taught 'second brain.' Notion sold it. The first mover funds everyone else's marketing.
Where do I play and how do I win?
When Sensei reads for the lane a founder should pick, these are the priors the reading leans on.
Every successful early motion looked embarrassingly narrow at the time. The founders who broadened before fit died politely.
Stripe's first motion: 7 Bay Area developer teams. Linear's first: 3 engineering teams inside YC companies. Slack's first: 45 people at Tiny Speck. Breadth is the tax you pay after fit, not before.
The sentence a customer uses to describe your product has to match the sentence they use to describe the problem. When they don't match, every ad dollar pays the translator.
Dropbox's 'your files, everywhere' mirrored the problem sentence exactly. Color's 'proximity-based photo sharing' did not — and burned $41M learning it.
Companies that ship in quarters compound. Companies that ship in 'phases' drift. Pick one bet per quarter and forget the others for ninety days.
Shopify's 'one bet' seasons became institutional. ClickUp's 'everything at once' shipped 400 features and hit retention problems that took two years to unwind.
A thirty-day plan survives first contact when every week ends with something a customer touches. Any internal-only week is a week the plan drifted.
Basecamp's Shape Up cycles end at customer delivery by design. Every startup post-mortem that cites 'we built for three months before talking to users' describes the same failure.
Am I seen the way I want to be?
When Sensei judges whether a founder is read the way they intend, these are the rules of the three-second test.
The heroes that convert name three nouns: who it's for, what it replaces, what happens if they buy. Never an adjective doing a noun's job.
'Remote job board for designers.' 'Email that clears itself.' 'The invoice you actually send on Monday.' The failed hero patterns always start with adjectives: 'Seamless.' 'Intelligent.' 'Next-gen.'
Conversion killers don't compound linearly — they cap your upside. Fix the biggest one and three others that looked independent clear with it.
Intercom's 2018 pricing-page rewrite lifted trial-to-paid 24%. The other 'fixes' they A/B tested that year averaged 2%. One root cause, one fix.
Visitors decide in three seconds of scroll-feed attention — not three thoughtful seconds. If your hero has more than one idea, none of them land.
Linear's hero has said one thing since 2019: 'The issue tracker you'll enjoy using.' Notion has rewritten theirs eleven times chasing more ideas — they convert worse per Ahrefs data than they did in 2020.
Generic trust signals — 'used by top teams,' '10,000 customers' — cost nothing and prove nothing. Named logos with a quote from a named person convert 2–3× better.
Gumroad's testimonials name the creator and the number ('I made $62,000 my first year'). Unnamed carousels on competitor sites convert half as well on the same traffic.
Who wins if I don't?
When Sensei sizes up a rival field, these are the patterns that decide who carries the clock.
An incumbent raising a Series B buys roughly 18 months of runway to clone you. Treat every competitor round as a clock starting.
Airtable raised Series B in 2018; Coda spent the next 18 months racing to ship collaboration before Airtable caught up. The loser on that clock usually pivots.
You don't beat a funded incumbent by matching their features — you beat them by compounding a habit they can't flex to copy without breaking their business model.
Figma couldn't outspend Sketch. It shipped real-time multiplayer — a habit Sketch's file-based model could not mimic without a rebuild. Same playbook: Shopify vs. Magento; Slack vs. HipChat.
Most founders read competitors on one axis (features). The winning read is two: how much threat they are, and how vulnerable they are. High threat + high vulnerability is the wedge.
When Loom launched against Zoom + Wistia, both were high threat; only Wistia was high vulnerability on async use cases. Loom attacked the gap and hit $180M ARR in four years.
The wedge that wins is one sentence a customer says without your help. 'I use X because Y does the one thing my team cannot skip.' Two sentences means no wedge yet.
Superhuman's wedge sentence was 'keyboard-only email.' Notion's was 'one doc, one database.' Every company that needed a paragraph to explain their edge is in the AngelList graveyard.