Torch Dental’s Data Moat Strategy: From Zero to 15 Million Transactions

Torch Dental’s Data Moat Strategy: From Zero to 15 Million Transactions

Every B2B founder claims to be building a “data moat.” Most are lying to themselves.

The harsh reality: collecting data isn’t a moat. Having a database full of customer information doesn’t protect you from competition. Real data moats require something more fundamental—product architecture where accumulated data makes your offering materially better for every user, creating advantages that compound exponentially over time.

Khaled Boukadoum, Founder of Torch Dental, built one of these rare, genuine data moats. In a recent episode of Category Visionaries, he shared how Torch went from zero to processing $5 billion in patient transactions—and why that number represents something competitors simply cannot replicate.

The Architecture of a Real Data Moat

Most SaaS companies collect data as a byproduct of their service. Torch Dental architected their entire platform around data accumulation as the core value driver.

“We’ve processed about $5 billion of patient transactions over our life, which is about 15 million unique transactions,” Khaled explains. But the raw numbers only tell part of the story. What matters is how that data feeds back into the product to create compounding advantages.

Torch’s platform handles the most complex part of dental practice operations: the post-treatment financial experience. When a patient completes treatment, Torch automates insurance verification, payment posting, claims submission, and collections. Every single one of these interactions generates data that trains the AI models powering the automation.

Here’s where it gets interesting: the more transactions Torch processes, the better its AI becomes at predicting insurance coverage, identifying payment patterns, flagging potential claim issues, and optimizing collection strategies. Better AI means better automation. Better automation attracts more practices. More practices generate more transactions. More transactions improve the AI further.

This isn’t a marketing claim about “network effects”—it’s mechanical reality built into the product architecture.

Why Transaction Volume Creates Irreplicable Advantages

Transaction data in healthcare has unique properties that make it particularly valuable for building moats. Unlike simple usage data or behavioral analytics, each transaction contains dense, structured information about insurance plans, payment behaviors, claim adjudication patterns, and reimbursement outcomes.

When Torch processes an insurance verification, they’re not just confirming coverage—they’re learning how that specific insurance plan handles specific procedure codes, what their typical reimbursement rates are, which claims require additional documentation, and how long adjudication typically takes.

Multiply this learning across 15 million transactions, and you have something extraordinary: a probabilistic model of the entire dental insurance ecosystem. This model allows Torch to automate tasks that would be impossible with rule-based systems or limited datasets.

A competitor launching today would need years of transaction volume to reach Torch’s current baseline capability. And during those years, Torch continues accumulating data, pushing the goal line further away. The gap doesn’t narrow—it widens.

The Cold Start Problem and How Torch Solved It

Every data moat faces a chicken-and-egg problem: you need data to make the product valuable, but you need a valuable product to attract customers who generate data.

Torch solved this by focusing their initial product on workflows where even basic automation provided immediate value, then using that foothold to accumulate the data needed for more sophisticated capabilities.

“Our bread and butter at Torch is really using AI to take all these manual workflows and automate them,” Khaled notes. The early workflows Torch automated—insurance verification, payment posting—delivered value from day one, even with limited training data. Practices adopted Torch because it solved immediate pain points, not because of some future promise of AI improvement.

But every practice that adopted Torch began generating transaction data. As the dataset grew, Torch could tackle increasingly complex workflows: predictive collections strategies, anomaly detection in claim adjudication, intelligent patient payment plans, revenue cycle forecasting.

This staged approach allowed Torch to build momentum while simultaneously constructing their data moat. Early customers got a valuable product immediately. Later customers got an increasingly powerful product that early customers also benefited from through ongoing improvements.

Making Data Accumulation Core to Product Strategy

The difference between companies that claim to have data moats and companies that actually have them comes down to product design choices made early.

Torch didn’t just build software that happens to collect data. They architected their entire platform around the principle that accumulated data should make the core product better for all users simultaneously.

This manifests in specific ways. When Torch’s AI learns that a particular insurance plan consistently denies claims for a specific procedure code without prior authorization, that knowledge benefits every practice that works with that plan. When Torch identifies payment collection strategies that work better for specific patient demographics, every practice gains that insight.

The product literally improves every day without Torch shipping new features—simply because the AI trains on more data. This is the hallmark of a genuine data moat.

The Strategic Implications for GTM

Building a real data moat fundamentally changes your go-to-market strategy. When your product gets measurably better with every customer, acquisition becomes an investment in product quality, not just revenue growth.

This influenced how Torch approached market expansion. Rather than optimizing purely for customer lifetime value and acquisition costs, they considered each new practice as both a revenue source and a data contributor. High-volume practices became especially valuable—not just because they paid more, but because they generated more transaction data that improved the product for everyone.

“We’re now a couple hundred practices live and definitely growing quickly,” Khaled shares. Each new practice doesn’t just add to transaction volume—it adds diversity to the dataset, exposing Torch’s AI to new insurance plans, payment patterns, and operational scenarios.

This creates a strategic advantage in market expansion. As Torch enters new geographic markets or practice segments, they’re not starting from zero. The accumulated knowledge from existing practices transfers, giving them a head start competitors cannot match.

Why Most Data Moats Fail

The brutal truth is that most claimed data moats are illusions. Companies collect data but never build the product mechanics that turn data accumulation into compounding advantages.

The failure patterns are predictable: data sits in warehouses without feeding back into product improvement; the product provides value independent of accumulated data; or the data becomes stale and loses relevance over time.

Torch avoided these traps through architectural decisions made early. The AI models that power automation are continuously retraining on new transaction data. The product value proposition is intrinsically tied to automation quality, which is intrinsically tied to data volume. And transaction data remains perpetually relevant because insurance plans, payment behaviors, and claim patterns constantly evolve.

The Future State: System of Record as Ultimate Moat

Khaled’s long-term vision for Torch reveals something crucial about data moats: the ultimate moat isn’t just accumulating data—it’s becoming the system of record that everyone else integrates with.

“Our ambition is really, again, to be the operating system, to be that piece of connective tissue that connects the insurance companies with the patients and with the practices and sits at the center of all those relationships,” Khaled explains.

When you become the system of record, data doesn’t just improve your product—it becomes the authoritative source that other systems depend on. This creates a different category of moat entirely: not just competitive advantage through better automation, but structural advantage through being the central data infrastructure.

The practices using Torch today aren’t just customers—they’re participants in building this infrastructure. Every transaction they process contributes to the dataset that makes Torch more valuable, more accurate, and more indispensable.

For B2B founders obsessing over building moats, Torch Dental’s journey offers the clearest blueprint: architect products where accumulated data makes the core value proposition materially better for all users; solve the cold start problem by delivering immediate value while building toward data-driven advantages; and position not just for competitive advantage but for becoming infrastructure that others integrate with. That’s how you build something competitors can’t copy, can’t catch, and can’t displace.