Ready to build your own Founder-Led Growth engine? Book a Strategy Call
Frontlines.io | Where B2B Founders Talk GTM.
Strategic Communications Advisory For Visionary Founders
For the first three years, ZoomInfo structured deals where customers paid a services fee to get custom data built for their specific TAM — with 90 days of exclusive access before it rolled to all users. The customer paid to expand the database; ZoomInfo kept the asset. "We were using our customers effectively to pay us to build net new parts of that asset." This is a replicable capital-efficient flywheel for any data or content business pre-scale: let early customers subsidize the infrastructure that makes you defensible to the next customer.
Henry's entire early market was companies selling to the CIO or anyone in the CIO's org. Not HR buyers. Not CFO buyers. One org chart. That constraint is what made the data asset specific enough to be valuable and the sales motion specific enough to be repeatable. "If you sold to the chief information officer or somebody in the chief information officer's organization, then you were a fit for us. Otherwise you weren't." Horizontal expansion came only after the 2019 ZoomInfo acquisition — 12 years in.
Henry's early instinct when closing a staffing firm: immediately identify every competitor in that vertical and run the same play. Today it's automated. "When a seller closes a deal, we built a go-to-market agent that goes out, takes that company, understands everything about them, looks to see what other customers we have that are clients like that, and then looks at the delta and then sends it to them and says, here are eight other companies that are just like this deal that you just sold that you should also now prospect into." Most teams treat a closed deal as a finish line. Henry treats it as a trigger.
The pattern Henry sees across his customer base: companies pilot AI SDRs for outbound, get a bump for two to three months, then it dies. The structural reason most teams miss — the TCPA prohibits automated outbound voice calls unless the recipient has explicitly opted in. "There's a law in the United States called the TCPA that says you cannot call someone outbound with an automated voice unless they've opted in to receive an automated voice call from you." Inbound AI SDRs work. Outbound automated voice without consent does not. Most teams conflating the two are either burning budget or running legal risk.
Henry's framing for why most SaaS businesses are in trouble: "The LLMs have sucked down every piece of publicly available information that exists in the world. And if you just have an application that sits on top of some unique public information that you've gathered, you have no real advantage." The only defensible position: data generated through a contributory network or flywheel that LLMs cannot access. Every customer you acquire that adds to that network makes the asset harder to replicate. "Every dollar you spend that gets you the net new customer makes the data asset more unique, more proprietary, better than the next person who gets in."
Customers no longer want one interface to one data source. They want ZoomInfo data plus first-party CRM data plus call recording data plus niche vendor data — surfaced in their SaaS app, in Claude, in ChatGPT, in their CRM, in internal vibe-coded tools. "Historically, one interface. Today, lots of surface area to plug into — and you need a unified context layer that brings all of that go-to-market context together." Seat-based models are structurally trapped. Extensible data layers are not.
The check was for $14,500. Henry Schuck had it framed.
It came six months after he started DiscoverOrg in 2007 on a $25,000 credit card in Columbus, Ohio. The buyer was an SVP of sales at Comsys, a publicly traded staffing and recruiting firm, who responded to an email marketing campaign and moved through the sales cycle fast. But what looked like a clean, repeatable first win was about to reveal a structural problem Henry would spend three years solving.
The customer loved the database. Then they asked for data DiscoverOrg didn’t have.
Comsys wanted access to DiscoverOrg’s existing data across roughly 9,000 companies. But their real TAM was larger. They came back and said, in Henry’s words, they loved what he’d built “across these 9,000 companies, but we really need the next 10,000 companies.”
It was a pattern that would repeat with nearly every early customer. The product was genuinely valuable — but its coverage was always smaller than what buyers needed. Henry had two choices: raise capital to fund data expansion, or make customers pay for it themselves.
He chose the latter. The structure he built around it became the foundation of DiscoverOrg’s early moat.
In a recent episode of Unicorn Builders, Henry explained how that first deal set the template. DiscoverOrg would give customers access to its existing database, then charge a services fee to build the additional data they needed. Henry couldn’t recall the exact per-account rate, but the mechanics were clear: customers got three months of exclusive access to the custom data. After that, it rolled to all users on the platform.
“We were using our customers effectively to pay us to build net new parts of that asset,” Henry said.
The model’s elegance: every services dollar a customer spent made DiscoverOrg’s database more valuable to the next customer. The company was getting paid to build its own moat.
The customer-funded model only worked because DiscoverOrg stayed narrow on who it sold to. For the company’s first decade-plus, the entire ICP fit on one line: companies selling to the CIO or anyone in the CIO’s organization.
Not HR buyers. Not CFO buyers. One org chart.
Henry described the qualification process in physical terms: “I would print out stacks and stacks of effectively like the first page in a company’s homepage. And I would go look through those and I’d go, okay, this one sells to IT. This doesn’t sell to IT.”
That constraint created two compounding advantages. First, the database became genuinely deep in a specific area rather than shallow across many. Second, the sales motion became pattern-matchable. When a staffing firm signed, Henry immediately identified every competitor in that vertical and ran the same play. “My mind always went to, okay, who are the other companies that also focus on the same thing? Because if it works for them, then it should work for everybody else.”
Today that motion is automated. When a rep closes a deal, a go-to-market agent analyzes the new customer, identifies lookalike companies, finds the gap, and sends the rep eight target accounts. The manual insight Henry had in 2007 is now a system. But it only became systematizable because the ICP was tight enough to generate clean signal.
Horizontal expansion — selling to any company that sold to any business — came only after DiscoverOrg acquired ZoomInfo in 2019. The constraint held for over a decade before Henry let it go.
Henry is now watching a version of his own early pattern play out across his customer base — and most teams are making an avoidable mistake.
The pitch: deploy AI SDRs for outbound prospecting, replace human labor, scale the motion. What Henry is seeing instead: customers pilot it, see a bump for two or three months, then watch results collapse. Most attribute the decay to message fatigue or list quality. The actual problem is structural and legal.
“There’s a law in the United States called the TCPA that says you cannot call someone outbound with an automated voice unless they’ve opted in to receive an automated voice call from you.”
Inbound AI SDRs — where the prospect has already initiated contact and consented through a form — work within the law. Outbound automated voice without prior opt-in does not. Henry is having this conversation with customers regularly, watching teams either burn budget or run legal risk without realizing it.
The principle behind the example: new tooling gets adopted for the use case that sounds most valuable before anyone checks whether that use case is actually viable. The teams that win stress-test the constraint before building the motion around it.
Henry’s framing for what makes a SaaS business defensible is the starkest version of an argument that has been building across the industry: “The LLMs have sucked down every piece of publicly available information that exists in the world. And if you just have an application that sits on top of some unique public information that you’ve gathered, you have no real advantage.”
The corollary is specific. The only data that creates durable competitive advantage is data generated through a contributory network or flywheel that no model can access — signals that only exist because customers are actively generating them. “Every dollar you spend that gets you the net new customer makes the data asset more unique, more proprietary, better than the next person who gets in.”
This reframes the VC question entirely. Henry’s case for raising venture capital isn’t growth at all costs. It’s a specific argument: if your customers generate the proprietary data that makes your product defensible, then every dollar spent acquiring a customer is also a dollar invested in an asset that compounds. Market share and moat-building become the same activity.
The founders who miss that distinction are spending acquisition dollars on growth that doesn’t compound. Henry spent over a decade making sure every dollar did both.