There's a rule almost everyone in B2B SaaS repeats like gospel: "Don't sell to startups." They're small, they die overnight, and if you keep saying yes to their custom feature requests, you slowly turn into a systems integrator instead of a software company.
Stripe and Mercury did the exact opposite. They went after startups on day one. One of Stripe's earliest customers, Shopify, is now worth $180B+. Mercury today banks half of every single YC batch.
Everyone says startup customers are a dead end
Honestly, the logic isn't wrong. A widely-shared analysis of Korean B2B SaaS startups lays out exactly why selling to startups goes bad.
Startup customers have a high risk of vanishing overnight since survival rates are low, and contract values stay small so revenue never really compounds. Chase that revenue and customize a feature for every account, and you quietly stop being a standardized SaaS product — you become a custom dev shop. The prescribed fix: find the shared needs of larger enterprise accounts and build for those instead.
Reasonable advice. Except if that were the whole story, Stripe and Mercury shouldn't exist.
Except Silicon Valley's unicorns grew the opposite way
a16z partner James da Costa put a name on this pattern in a recent podcast episode: the "greenfield" go-to-market strategy. Here's how he framed it.
"If you attract all of the new companies at formation and then grow with them, as your customers become big companies in their own right, so will you."
— James da Costa, a16z partner
Except this isn't new. Stripe and Mercury proved it over a decade ago.
When Patrick and John Collison built Stripe in 2010, their first customer was a friend, Ross Boucher. Their first 10 to 30 customers all came from inside the YC ecosystem — every one of them a brand-new startup.
That's where the "Collison Installation" came from. Paul Graham recalled it this way: "Other founders would ask 'wanna try our beta,' but the Collison brothers would say, right then, give me your laptop". It only worked because there were one or two founders in the room making the call.
What happened to those early customers is the whole point. Shopify, an early customer of Stripe Connect, is now worth $183B as of August 2025. Stripe itself went from 100 customers and 10 employees in 2011 to over $1B in revenue by 2017.
Mercury ran the same play on purpose. As of March 2023, over 50% of every YC batch banked with Mercury. By 2024, 40% of new startups opened their first account there. Today it's a $500M revenue business with 200,000+ customers. When Silicon Valley Bank collapsed in March 2023, $2B flowed into Mercury within six days — because the startup network was already there.
| Enterprise-first | Greenfield (startup-first) | |
|---|---|---|
| Decision structure | 5-10 stakeholders, approval chain | 1-2 founders, instant yes |
| Switching cost | Strong incumbent resistance | No tool in place yet |
| Initial deal size | Large | Small |
| Long-term upside | Customer is already big | Compounds as the customer scales |
So why do some startup-first bets turn into an SI trap?
The Korean analysis isn't wrong, exactly. The real failure isn't "having startup customers" — it's saying yes to every individual request until the standard product disappears. Neither Stripe nor Mercury won by customizing. They won by holding the line on a standard product.
Rillet makes the distinction even clearer. The AI-native accounting platform didn't target day-one startups at all — it targeted the opposite moment: the exact point a scaling startup outgrows spreadsheets or QuickBooks and needs to move to something like NetSuite. Its early customers were already-scaled AI startups: Windsurf, acquired by OpenAI for $3B, and Decagon, valued at $1.6B.
The result: Rillet doubled annual recurring revenue in 10 weeks and signed 200 customers, all while keeping a standard ERP product. CEO Nicolas Kopp put it simply: "work that used to take weeks now takes hours".
So greenfield only works under two conditions: nail the moment when stakeholders are few and switching cost is zero, and build only for needs that repeat across customers, never one-off requests. Miss the second condition, and you land right in the SI trap the Korean analysis warned about.
How to run the greenfield play in your own product
- Diagnose the timing first
How many decision-makers does a company need at the moment your product becomes relevant to them? If 1-3 founders can say yes on the spot, that's a greenfield signal. - Find the zero-switching-cost moment
Don't try to rip out an incumbent tool. Find the company, or the moment, where nothing is in place yet. Mercury targeted the instant a startup opens its first bank account. - Lock the standard feature set, skip the customization
When a request comes in, ask: "does this repeat across at least 3 other customers?" If yes, build it in. If it's one-off, decline politely. This is where SaaS and SI diverge. - Track your customer's growth as your leading indicator
Watch seat count, transaction volume, and headcount growth at your customer accounts. Your revenue naturally compounds as they scale. - Design a second wave around the scale-up moment
Like Rillet, you don't have to be the first tool a company adopts. You can target the second transition point — when their current tool starts to break.
Go deeper
Big Ideas 2026: New Infrastructure Primitives The original a16z podcast episode that seeded this piece. James da Costa introduces the greenfield strategy in a few minutes. a16z.com
Investing in Rillet a16z's announcement explaining why it led Rillet's Series B and the $500B market opportunity in AI-native ERP. a16z.com
Rillet raises $25M from Sequoia TechCrunch's reporting on Rillet's founders, early customers, and its NetSuite-switching strategy. techcrunch.com
The First Few: Stripe A detailed account of Stripe's first customers, the Collison Installation, and its early YC-fueled growth. justgogrind.com
Mercury: Business Breakdown & Founding Story A research report on Mercury's founding, YC batch share, and growth after the SVB collapse. research.contrary.com
The Korean B2B SaaS Startup Dilemma Kakao Ventures' analysis of why startup customers are considered risky, and how the SI trap forms. kakao.vc
