The Case Study Behind the Work: Why Some Lists Can Only Be Built Manually

Private equity firms rely on clean, reliable information to make decisions. But in certain markets, even identifying the companies that actually belong in a sector requires more than data extraction. It requires judgment. It requires context. And sometimes, it requires the kind of deep, manual research that no tool, platform, or outsourced solution can replicate.

A recent project illustrates this clearly.

A PE firm approached me to build a landscape of diagnostic laboratories in the United States. But instead of asking me to start from scratch, they shared an Excel file containing roughly 4,000 entries: names, addresses, cities, states, zip codes, and phone numbers. On the surface, it looked comprehensive. In reality, it was just a starting point.

The list wasn’t “wrong.” It was simply unrefined. It mixed together entities that looked similar on paper but were fundamentally different in practice. And this is where the real work began.

The First Pass: Sorting What Is from What Only Looks Like It Is

As I examined the file row by row, it quickly became clear that a large portion of the dataset didn’t belong. Many entries were temporary COVID-testing tents that had been set up during the pandemic and taken down since. Others were CROs or CDMOs—credible companies, but irrelevant to the client’s thesis. Some entries didn’t have websites, making them unviable targets from a diligence standpoint. Many were diagnostic labs tied to hospital systems, universities, and health networks, which the firm explicitly did not want.

Each category required different reasoning. Some were eliminated because they didn’t match the client’s strategic goals. Others because the “company” no longer existed. Others because they operated in adjacent but non-relevant industries.

By the time this manual vetting process was complete, the dataset had been refined from roughly 4,000 entries to about 1,000 actual, independently operating diagnostic laboratories that matched the client's definition.

And that was only Step One.

The Second Pass: Turning ~1,000 Raw Names Into Actionable Intelligence

Once the irrelevant entities were removed, the real research began.

For each of the ~1,000 validated diagnostic laboratories, I had to determine:

·      The specific tests they offered

·      How those tests mapped to categories defined by the client

·      Ownership structure (PE-backed, VC-backed, privately held, publicly listed, etc.)

·      Any recent changes in leadership

·      Growth indicators or expansion signals

·      Any contextual nuances that might impact acquirability

None of this information existed neatly in one place. Some laboratories only provided partial details; others required digging through archived pages, state filings, or scattered references. Many required judgment calls about how to categorize tests or evaluate business models.

This is where manual research truly matters. Automation can pull data, but it cannot evaluate ambiguities, reconcile contradictions, or distinguish between meaningful detail and noise.

Why This Couldn't Have Been Done Any Other Way

The PE firm didn’t send this dataset to me because they lacked tools. They sent it because they needed clarity.

Here’s the reality:

·      AI tools and automated sourcing platforms can only work with what is already clean, structured, and consistently formatted. Diagnostic laboratory data in the U.S. is none of those things.

·      Investment banks and buy-side brokers could have produced something similar, but at a significantly higher cost, and without the granularity the client wanted at this early stage.

·      Freelancers or generalist research teams couldn’t have done it because they lack the context, the pattern recognition, and the judgment that only comes from having built dozens of PE-focused target lists before.

·      Doing it in-house would have been prohibitively time-consuming, and deal teams simply cannot afford a 150-hour research detour while running processes and evaluating opportunities.

The only viable path was a bespoke, fully manual research approach.

When Bespoke Research Becomes A Competitive Advantage

Projects like this make something clear: high-quality target lists aren’t assembled. They’re constructed. They require domain familiarity, second-order thinking, and constant collaboration with the deal team to know what matters and what doesn’t.

They require someone who can:

·      interpret ambiguity

·      refine definitions mid-process

·      recalibrate when the thesis evolves

·      spot patterns that only emerge through hands-on work

This is not “data entry.” This is diligence before diligence. And it only works when the person building the list is close enough to the work to notice what tools and scaled teams cannot.

A List Is Only As Good As The Thought Behind It

By the end of this project, the client didn’t just receive a list. They received clarity, that allowed them to:

·      understand the true structure of the diagnostic laboratory market

·      identify high-value segments

·      prioritize outreach

·      avoid spending time on irrelevant or unacquirable entities

This is the kind of outcome that isn’t possible through shortcuts. It can only come from work that is deliberate, careful, and grounded in real understanding.

If your firm needs this level of depth for an upcoming thesis or market exploration, I’d be glad to start a conversation. You can reach me directly at yash@agathonrp.com.