From Data Chaos to Operational Clarity: A Conversation with Emily Lemmon, Lead Software Engineer

Helping Health Plans Turn Complex Data into Action: Q&A with Emily Lemmon

Amanda Bair: You’ve spent much of your career helping health plans manage data in Pareo. What drew you to this work? 

Emily Lemmon: I’ve always been interested in data. I’m pretty detail-oriented, and I’ve always enjoyed finding the places where details match — or where they don’t.

What keeps it interesting is that data isn’t just data. We have these huge collections of information related to claims, providers, and members, but every claim represents a real encounter. There’s a patient. There’s a provider. There’s a real-world event behind every record.

I enjoy being able to zoom out and see patterns across enormous datasets while still remembering that each data point connects back to something that actually happened. It’s interesting to see how those things connect and what we can learn from them.

AB: Most users see dashboards, reports, and workflows. What does it actually take behind the scenes to move large volumes of data into and out of a platform reliably? 

EL: It takes a lot of organization, planning, and understanding how data fits together.

At a high level, we start with foundational datasets — more stable ones like claims, members, and providers, and more dynamic ones like audits from vendors. Before anything meaningful can happen in a payment integrity program, those datasets have to be brought in accurately and stored correctly.

That sounds simple, but every health plan structures data a little differently. Different organizations use different field names. They identify records differently. They have unique business requirements.

A lot of our work involves mapping. We have to translate what a client sends us into a structure that can be consistently used inside the platform. We need to make sure we’re saving the right information in the right places while preserving what makes that client’s business unique.

Over the years, we’ve seen enough patterns that we’re able to standardize more of that work. That has made implementations faster and more scalable while still allowing us to support client-specific requirements where they matter most.

AB: What are some of the biggest challenges health plans face when trying to bring data together from multiple sources?

EL: One of the biggest challenges is that different systems often have different assumptions about what makes a record unique.

For one health plan, a claim number may uniquely identify a claim. For another, you may need the claim number, member ID, and date of birth together to identify the same record correctly.

The larger and more complex the organization becomes, the more those variations tend to show up.

The challenge goes beyond just collecting data. It’s understanding the relationships within the data and maintaining those relationships as information changes over time.

The more sources you have, the more important it becomes to understand exactly how those connection points work.

AB: What do people often misunderstand about the work you do?

EL: People sometimes assume that automation is simple. They’ll say, “I want the process to do this.”

But from an engineering perspective, we have to answer a lot more questions.

How do we identify the records that belong in that process? What data points tell us which records qualify? What happens if the data is incomplete? What happens if something unexpected occurs?

One thing I’ve learned is that “it should never happen” isn’t really an answer.

We have to think through the exceptions, even when they’re unlikely. If something unexpected does happen, should the process stop or continue? Should it notify someone or make a note and then continue? Should it create an exception?

Those decisions are where a lot of the work actually happens.

AB: It sounds like there’s a surprising amount of creative problem solving involved in what many people would view as a technical role. 

EL: Absolutely. People sometimes picture software developers working in isolation, but that’s not really how this kind of work functions.

Communication is one of the most important skills I use every day. A lot of what we do involves understanding what a client actually wants, identifying assumptions that may not have been explicitly stated, and helping everyone arrive at a shared understanding of the problem.

There are plenty of situations where something isn’t broken, it simply isn’t behaving the way someone expected it to behave. In those cases, the challenge is understanding where those expectations came from and determining the best path forward.

That’s a very collaborative process.

AB: When health plans successfully automate processes that were previously manual, what has to happen behind the scenes?

EL: A lot of detailed investigation. You have to understand exactly what should happen, when it should happen, and what should happen when something goes wrong.

Because things do go wrong. Files are missing information. Connections fail. Processes encounter unexpected situations. The automation itself needs a remediation plan.

The most successful automations are the ones where we’ve thoroughly thought through — both the expected path and the exceptions.

That’s also why automation can be so satisfying. Every manual process contains knowledge. When people perform a process manually for years, they learn all the edge cases and nuances. The challenge is translating that knowledge into a repeatable system.

AB: Looking ahead, where do you see the greatest opportunities for health plans to reduce manual work and get more value from their data?

EL: Standardization is a big opportunity.

The more health plans and technology partners can align around common ways of defining and exchanging information, the easier it becomes to automate processes and move data efficiently.

But I also think there’s another opportunity. Sometimes organizations ask technology to replicate a manual process exactly as it exists today. The real question should be: does that process still need to exist at all?

Years ago, I worked on a process where data was transformed into files that were printed and physically carried somewhere else for the next step in the workflow.

That made sense in one respect because that’s how the process had always been done.

But when technology removes the limitations of the manual world, we have an opportunity to ask a different question: What outcome are we actually trying to achieve?

Sometimes the best solution isn’t automating an old process. It’s replacing it with something better.

Conclusion

As payment integrity programs grow in scale and complexity, the ability to manage data effectively becomes increasingly strategic. Success depends not only on identifying opportunities and managing workflows, but also on ensuring that the right data is connected, validated, and available when it is needed.

For Emily Lemmon, that challenge remains as interesting today as when she first started.

“The more sources you have, the more important it becomes to understand exactly how those connection points work,” she says.

In a payment integrity landscape defined by growing complexity, those connections may be the most important work happening behind the scenes.