When One Medical was scaling to 100+ offices across 20+ markets, we had a data problem nobody talked about publicly.

We had plenty of data. We had dashboards. We had analysts pulling reports every week. What we didn't have, at first, was the discipline to ask the harder question: is anyone actually making decisions with this?

The moment it became undeniable that we had a governance problem was surprisingly simple. We couldn't agree on who counted as an active user.

Not a complex metric. Not a calculated field with seventeen dependencies. Just: is this person actively using the product? The business team had a definition. The technical team had a different one. Leadership was making decisions based on whichever number happened to show up in the deck that week. Two teams, two numbers, one flawed foundation.

That single disconnect, around the simplest possible metric, taught me more about data governance than any framework or methodology I've read since.

The Difference Between Data and Usable Data

There's a version of the data problem that gets a lot of attention: organizations that don't collect enough data, or don't have the tools to analyze it. That problem is real, but it's increasingly solvable. The technology is mature. The platforms exist.

The harder problem (and the one I watched play out repeatedly at One Medical) is organizations that have plenty of data but can't act on it coherently.

The root cause is almost never technical. It's organizational. It's the absence of what I've come to think of as a single version of truth: a shared, governed definition of what your most important metrics actually mean, who owns them, and how they're calculated.

Without that foundation, dashboards multiply. Reports diverge. Different departments optimize for different numbers. And leadership ends up making decisions based on whichever analyst happened to pull the data most recently.

The business team defines "active user" as someone who logged in within the last 30 days. The technical team defines it as someone with at least one completed session in the last 90 days. Both definitions are defensible. Neither is wrong on its own. But when they coexist without resolution, every conversation about user growth becomes a negotiation about whose number is right, before the actual strategic question even gets discussed.

Governance Before Analytics

The organizations I've seen do this well share one practice: they invest in governance before they invest in analytics.

That means sitting down, before anyone builds a dashboard, before anyone writes a query, and answering the foundational questions. What are the five or ten metrics that actually drive this business? Who owns the definition of each one? When two teams pull different numbers for the same metric, what's the process for resolving it? How do we make sure that the "active users" number in the board deck is the same "active users" number the product team is optimizing for?

These conversations are not glamorous. They don't produce impressive slides. They rarely get celebrated in all-hands meetings. But they determine whether data actually drives a business, or whether data simply decorates it.

At One Medical, building the analytics function forced us to have those conversations explicitly, often for the first time. The process was uncomfortable. It required business and technical teams to negotiate definitions they had previously operated independently. It required acknowledging that some of the metrics we'd been tracking for months were inconsistently defined. It required building the less visible infrastructure (documentation, ownership, change management processes) before the visible infrastructure could be trusted.

The payoff was significant. Once teams were working from a shared foundation, the quality of strategic conversations improved almost immediately. Debates about data shifted from "your number is wrong" to "here's what this metric is telling us." That's a completely different kind of discussion.

What This Has to Do with Patient Access

When I left One Medical and spent time as CIO at a PE-backed dermatology group and COO at a multi-site specialty practice, I watched the same dynamic play out in a different context.

Healthcare organizations are drowning in operational data. Scheduling systems, EMRs, billing platforms, patient communication tools: each generates its own data stream, with its own definitions, its own reporting logic, its own version of the truth. A practice trying to understand its own performance has to reconcile numbers from five different systems, each of which was built by a different vendor with a different data model.

The question "how many appointments did we book last month?" sounds simple. In practice, it depends on whether you're counting scheduled appointments, kept appointments, completed appointments, or appointments where the patient showed up but was seen by a different provider than originally booked. Each of those definitions produces a different number. Each is defensible. None is useful without agreement on which one you're tracking and why.

When I co-founded Parakeet Health, the governance lessons from One Medical shaped how we approached client reporting from the beginning. The metrics we surface to clients aren't comprehensive. They're deliberate. Appointments booked. Slots filled. Cancellations recovered. Patients re-engaged. Each one is defined the same way for every client, tracked consistently over time, and connected directly to the operational and financial outcomes that practice leaders are accountable for.

The goal isn't an impressive dashboard. It's a shared foundation, so that when a client's COO and their VP of Operations look at the same report, they're having a conversation about what to do next, not a debate about whose numbers are right.

The Lesson That Travels

I've spent 25 years inside healthcare organizations. The technology changes. The platforms change. The terminology changes. But the underlying organizational dynamics don't change as fast as people think.

The organizations that get the most value from data, whether it's One Medical in 2012 or a PE-backed dermatology platform in 2026, are the ones that treat governance as a prerequisite, not an afterthought.

Data without governance is noise. You can have the most sophisticated analytics infrastructure in your industry and still make worse decisions than a competitor who has less data and better discipline about what it means.

Governance without action is theater. A beautifully documented data dictionary that nobody consults when they're pulling a report hasn't solved anything.

The combination of both, shared definitions, clear ownership, consistent measurement, and the organizational will to act on what the data says, is how healthcare organizations actually improve. Not just in the metrics that show up in board decks, but in the operational performance that those metrics are supposed to reflect.

That's the lesson One Medical gave me. It took longer than I should have to learn it. But it's one I've carried into every organization since.