The $100,000 Per Day Mistake: Why Domain Knowledge Trumps Technical Skills

The $100,000 Per Day Mistake: Why Domain Knowledge Trumps Technical Skills

2025-05-09
RMRichard Makara

The $100,000 Per Day Mistake: Why Domain Knowledge Trumps Technical Skills

Imagine making a mistake that costs your company $100,000 per day until it's fixed. This isn't hypothetical—it's the reality for data professionals in regulated industries like healthcare when they lack critical domain knowledge.

When most people think about data analytics, they picture SQL queries, dashboards, and Python code. But technical skills alone aren't enough. What truly separates exceptional analysts from average ones is deep domain knowledge—understanding the business context that gives data meaning.

I recently chatted with Brad Lumley, who has extensive experience bridging the gap between business context and technical delivery in healthcare analytics.

His journey offers valuable lessons for anyone working with data in complex environments. And while his expertise is in healthcare, I'm 100% certain that domain knowledge is the most critical asset for data professionals, regardless of industry.

The High-Stakes Reality of Regulated Analytics

Let's look deeper at that $100,000 per day scenario. This costly consequence isn't just a scare tactic—it's the daily reality in regulated industries like healthcare. It's a real scenario Brad has seen:

There's one state where if they sanction you for breaking any of their hundreds of pages of rules and regulations, you can face up to $100,000 per day fine until you've remedied the situation.

Working in this environment means being "ultra aware of all the requirements" where everything is managed by strict regulations that vary from state to state.

For national organizations, this creates layers of complexity:

"Not only do you have this enterprise-level compliance department, but for every health plan in the enterprise, you have compliance departments because different layers of regulations vary from state to state. Some rules supersede federal rules and some don't. There's always this balancing act."

The stakes are incredibly high. It's not just about doing good analytics—it's about ensuring compliance at every step:

"When you're audited by any number of regulatory agencies, you'd have to be able to produce what you did, why you made those decisions, all that kind of stuff."

This level of scrutiny means every data model, every query, and every dashboard needs to be built with a deep understanding of the compliance landscape. One wrong assumption about how data should be treated could trigger massive penalties.

The complexity goes even deeper with healthcare MCOs (Managed Care Organizations) where you're dealing with state-by-state regulations plus federal oversight. There's no "one size fits all" approach because regulatory requirements create unique data needs in each state.

Being the Translation Layer

Brad's unofficial title is "IT Delivery Liaison" — he stands between customer experience leadership and IT delivery groups to translate needs between them.

I have enough business context and enough IT context to be able to be that person who can talk to both groups. Around here, that's a pretty specialized talent.

This ability to speak both "languages" is increasingly valuable as systems grow more complex. When technical teams don't understand the business domain, projects fail, regardless of how technically brilliant the implementation might be.

The translation role isn't just about vocabulary—it's about deeply understanding the challenges, constraints, and goals of both sides:

"We can do a lot of creative thinking, come up with ideas that may not work at a regulatory level, but we have teams that can help evaluate that. I'll say, 'here's the idea, here's how we think we would implement this idea. Can you begin to shop this around to the different health plans and see if it's something they would approve?'"

The process involves a delicate dance between innovation and compliance. You need enough technical knowledge to grasp what's possible and enough domain expertise to know what's permissible.

This bridging role has become crucial in creating solutions that are technically sound and business-appropriate.

Brad's approach involves:

  1. Understanding both languages fluently - Not just technical jargon, but the actual thought processes and priorities of both groups

  2. Creating structured evaluation processes - "We have a project plan template and we'll start going out to the stakeholders in different health plans saying, 'will this work? will it not work?'"

  3. Finding the common patterns - "If this regulation is pretty close across states, how do we get to this happy medium that satisfies everything?"

  4. Owning the follow-through - The liaison isn't just a messenger but tracks the entire process from idea to implementation

For data professionals, developing this translation skill means your value extends far beyond your technical abilities. You become the bridge that makes technical solutions work in highly specialized business contexts.

How to Acquire Domain Knowledge

So how do you build this domain expertise? Brad's approach is refreshingly practical:

1. Leverage Subject Matter Experts (SMEs)

"Having them explain how things work" is crucial.

But don't just ask generic questions. Come prepared with specific scenarios: "If we wanted to implement this feature, what regulations would apply?" or "Can you walk me through exactly what happens when a member needs this service?"

The right SMEs are worth their weight in gold. Find the people who've been in the role the longest or who have worked across multiple parts of the organization. They often carry institutional knowledge that isn't written down anywhere.

Brad suggests: "Put a meeting on the books with them. I was like, I'm gonna brain dump all of my dumb ideas onto you and you're gonna tell me whether or not this makes any sense in this context."

2. Use the "Wrong Answer" Technique

"I'm gonna brain dump all of my dumb ideas onto you and you're gonna tell me whether or not this makes any sense in this context."

It's like that joke on the internet—post the wrong answer and someone will quickly correct you. As Brad confirms: "Man, are they fast to correct you."

This technique works brilliantly for three reasons:

  • People enjoy correcting others (it's human nature)
  • You learn exactly why your assumption was wrong, not just that it was wrong
  • The conversation becomes more interactive and memorable

Don't worry about looking foolish—the goal is learning, not impressing. The best domain experts appreciate curiosity more than they expect perfection.

3. Start in the Trenches

Before his current role, Brad learned the business by working on the front lines.

"I knew I was going to do what it took to get in the door. I took a job on the phones and really learned how to do the work by doing the work—answering phone calls, reading through the knowledge articles and standard operating procedures."

This hands-on experience is invaluable. You learn:

  • The actual day-to-day workflow (not the idealized version)
  • The pain points users experience
  • The edge cases and exceptions that make up real business
  • The unwritten rules and shortcuts people actually use

If you can't literally take a front-line job, volunteer to shadow one for a day. Sit with customer service reps, join sales calls, or watch operations teams do their daily work.

4. The Gemba Walk

"Go to where the work is being done. Watch it happen. See if you can participate."

Gemba is a Japanese term meaning "the actual place" where value is created. In data analytics, it means getting out from behind your desk and observing the actual business processes that generate your data.

Brad emphasizes: "Even more than walking the work, if there's any way to get your hands dirty, in the actual work, do it. Tactile experience of a process beats observation every time."

This might mean:

  • Spending a day with claims processors to understand how they categorize and handle exceptions
  • Watching how marketing teams plan campaigns that will eventually show up in your attribution models
  • Observing warehouse operations that feed into your supply chain metrics

What you'll discover is that the neat, clean data model in your head rarely matches the messy reality of how business actually happens. This understanding is gold.

5. Test Your Knowledge in Unfamiliar Contexts

Brad emphasizes the importance of applying your knowledge in new situations:

"Look for opportunities to push what your domain expertise is into a place where it's never been tested before."

This might mean volunteering for projects in adjacent business areas or helping another team with their analytics challenges. Each new context forces you to question your assumptions and deepens your understanding.

The Limits of AI in Building Context

Despite having access and experience with Ai tools, Brad consciously relies on human interaction for critical context.

Here's why:

Number one, trust. Most of us have been burned by AI already from their hallucinations. I don't totally trust it — I'm always double checking it.

But there's something deeper:

This is a relationship business. I need people to help me, and I need to help people, and I need to have that interaction.

Key Limitations in Regulated Environments

The limits of AI in regulated environments are particularly pronounced:

  1. The hallucination problem - In regulated fields, being "mostly right" isn't good enough. An AI that confidently states incorrect regulatory information could lead to those $100,000/day fines.

  2. The human judgment gap - Brad notes that his company tested various AI approaches and found something surprising: "The more human they tried to make the [AI system], the less it worked for people." Users actually preferred more obviously robotic interactions because they knew when to take control back.

  3. The relationship deficit - "This is a relationship business," Brad emphasizes. Building trust with SMEs means they'll come to you with problems before they become crises. No AI can build those relationships.

  4. The experiential knowledge gap - "I could tell you all the specs of a Ferarri 296 but I don't know what it's like to sit in one, drive it, feel the acceleration and handling."

AI can serve as a powerful augmentation tool—organizing documentation, suggesting relevant regulations, or highlighting potential compliance issues. But it still needs human oversight and validation, especially in high-stakes regulated environments.

As Brad puts it: "There's no way I'd just like copy and paste an [AI] answer to my boss. No way. Wouldn't happen."

The Career Advantage

The most successful data professionals I've talked to usually come from the domain they're analyzing. In CRM analytics, it's often former sales reps. In marketing analytics, it's marketers who learned data.

Brad confirms this pattern: "The best case scenario is to have really walked the walk."

For analytics professionals looking to level up, his advice is gold:

Do your craft in different circumstances.

It's not enough to understand something theoretically—you need to test your knowledge in new environments where the rules change.

"I thought I was really good at all things health insurance until I got to a place where all the rules were different. A lot of the knowledge is transferable, but until I'm actually doing that in this totally different environment with this different rule set, I don't really know it. I know it in theory only."

Major Career Benefits

The career advantage of domain expertise comes down to four major benefits:

  1. Rare combination of skills - Brad notes that his specialized talent of understanding both business and IT contexts is "not many people have it." This scarcity creates career opportunities and job security.

  2. Better solution design - When you deeply understand the domain, you can design analytics solutions that actually solve the right problems. This creates a track record of success that propels your career forward.

  3. Organizational trust - Domain experts earn the trust of business stakeholders, which means they get invited into strategic conversations earlier. Instead of being handed requirements, they help shape them.

  4. Resilience to automation - As AI increasingly handles the technical aspects of analytics, domain knowledge becomes your competitive advantage. The contextual understanding that takes years to develop can't be replicated by models.

The domain expert doesn't just execute analytics tasks—they shape how analytics is applied to business problems. That's a far more valuable and durable career position.

The Documentation Challenge

One of the toughest challenges in regulated analytics is documenting decisions. When audited, teams must show not just what they did but why they made specific choices.

This creates a documentation burden that goes beyond just recording technical implementations—you need to capture decision rationale:

"You just have to be clear about what your intentions were, what your plans were, and how from inception to delivery, your product or idea or change was going to continue to meet all the regulatory requirements."

It's similar to refactoring a data warehouse—changes must be "as good and better than the previous version" while still maintaining compliance.

The documentation challenge in regulated environments has several critical dimensions:

  1. Justification, not just implementation - "A reason could be reducing or eliminating waste," Brad notes. But whatever the reason, "that doesn't change the fact that you still have to meet all the same requirements."

    You need to document not just what you built, but why you built it that way and how the decision process ensured compliance.

  2. Audit readiness - "We have to have data that if we were to get audited by any number of regulatory agencies, we'd have to be able to produce what we did, why we made those decisions, all that kind of stuff."

    This means documentation can't be an afterthought—it needs to be integrated into your development process from day one.

  3. Granular detail requirements - "It can be super granular," Brad explains. "Third party auditors can come in and say, 'we need you to supply us a thousand claims and they need to include all of these procedures and they need to include all of these diagnoses.'"

    Your documentation and data systems need to support this level of detail extraction.

  4. Balancing business objectives with regulatory requirements - Brad explains: "I don't want to paint this picture that it's like the only thing we care about is whether or not the customer is happy. You can absolutely say 'we thought we would save a shitload of money by doing it this way.'

    Good documentation captures both the business goals and compliance considerations.

The solution isn't just "more documentation"—it's strategic documentation. Brad's team uses project templates that explicitly address regulatory concerns from the start. They also involve compliance teams early in the ideation process so requirements are baked in, not bolted on.

For analytics professionals, this means developing a documentation approach that:

  • Captures key decisions and their rationale
  • Links implementations to specific regulatory requirements
  • Anticipates audit scenarios
  • Is consistently maintained over time

Tools like version control for queries, data catalogs with governance capabilities, and contextual documentation systems are becoming essential in regulated environments.

Why Domain Knowledge Matters More Than Ever

Three major trends are making domain expertise increasingly valuable:

1. AI is Commoditizing Technical Skills

AI tools now make technical skills more accessible than ever. Anyone can generate SQL or code with the right prompts. But what AI can't replace is deep domain knowledge and business context.

Reusing Brad's wise words here again:

I could tell you all the specs of a Ferarri 296 but I don't know what it's like to sit in one, drive it, feel the acceleration and handling. There's absolutely this difference between sitting, looking at a screen versus actually putting your hand on the item.

2. Regulatory Complexity is Increasing

From GDPR and CCPA in privacy to industry-specific regulations, compliance knowledge is increasingly critical. In healthcare specifically, Brad explains:

"In the US, Medicaid is 100% funded by the government. There's no obligation for members to pay anything. Engagement becomes super important because you're trying to make sure that people are actually using the benefits that are available to them."

Understanding this business context shapes everything from what data you collect to how you structure analytics. Without domain knowledge, you'd miss these critical nuances.

3. Business Leaders Demand Strategic Partners

Data professionals are increasingly expected to contribute to business strategy, not just provide reporting. Those who will thrive in the coming years won't just be technically skilled—they'll be those who:

  • Deeply understand their business domains
  • Can translate between technical and business needs
  • Know how to navigate complex regulatory environments

This combination of skills is the gold mine most data professionals miss.

Getting Started: 10 Actionable Steps

If you're looking to build domain expertise, here are ten specific actions you can take today:

1. Shadow Business Users

Don't just ask for requirements—ask to observe their actual workflow. See what they do before and after using the analytics you build.

2. Join Cross-Functional Projects

Projects that span multiple departments force you to learn how different parts of the business operate and what they value.

3. Ask "Why" Questions

When someone requests a report or dashboard, ask why they need it, what decisions it will inform, and how it fits into their broader goals.

4. Document Your Thinking Process

Tools like reconfigured can help capture the context behind your analytics work, making it retrievable when you need it.

5. Test Knowledge in New Contexts

Volunteer for projects outside your comfort zone, where you'll need to apply your skills in new contexts with different rule sets.

6. Build Relationships with Experts

As Brad emphasizes, "This is a relationship business." Invest time in getting to know the people who understand the domain best.

7. Study the Regulatory Landscape

Set aside time specifically to understand the regulations that govern your industry. Don't wait for compliance issues to force the learning.

8. Create a Personal Knowledge Base

Maintain your own notes about domain-specific information, especially unwritten rules and context that isn't documented elsewhere.

9. Use the "Wrong Answer" Technique

When meeting with domain experts, propose your understanding and let them correct you—it's often more efficient than open-ended questions.

10. Look for Translation Opportunities

Identify places where technical and business teams are miscommunicating, and practice bridging those gaps.

As machine learning and AI handle more of the technical heavy lifting, your unique value will increasingly come from domain expertise and business context that no model can replicate.

Start Your Domain Knowledge Journey Today

Remember Brad's advice: "Do your craft in different circumstances." Your domain expertise becomes truly powerful when tested across different contexts and rule sets.

Thanks to Brad Lumley for sharing his insights on bridging the gap between business and technology in regulated healthcare analytics.

If this resonated, chat me up on LinkedIn, either via DMs, or directly on the feed đź«¶