“I’ll need to get back to you on that.”
Words that no one wants to hear in a meeting.
You’ve prepared a data-rich slide deck. Management comes back with, “Well, what does it all mean?”
You begin frantically pulling together small pieces from each slide to craft a story on the spot. Only to realize you don’t have the answer and will have to “get back” to the leadership team.
This is a common scenario in biopharma, and we’ve all probably been in that exact situation at some point. The truth is, your data systems should be set up so that the most common questions already have answers ready.
The big picture should always be relatively available, but all too often it’s not.
Many companies have an abundance of data, but lack a shared understanding of what it all means. Nobody truly knows what’s happening with projects or the organization, and how resources are actually being allocated.
What Happens When Biopharma Data Lacks Context
Most teams operate as if data alone will help them make sound decisions. So they do the experiments, tracking, and data collation. But without a purposeful flow of information and digital architecture, data alone is not enough.
When there is no flow of information, the team begins to make decisions based on a single piece of data they can see, rather than building a complete picture. Getting the whole picture takes several people to pull together manually, time that could be spent elsewhere.
Even then, there’s no assurance that the data will tell the full, true story. Because when there is no single source of truth, information is the wild, wild west. People struggle to find what they need, so they spend time reconciling and figuring it out.
This leads to what we call a “navigation nightmare,” in which teams become frustrated and give up. They don’t want to have to click through a thousand different windows to get the information they need.
The trouble is that this problem only compounds as the company and projects grow. The volume of data increases and without a structured data architecture, clarity decreases. This is a common growing pain that shows the team did not consider how the data structure would look as they grew.
A perfect example of how this plays out in real life is the shared drive for a company that’s been in business for ten years. There are a thousand nested folders with unlimited information, you can spend days in there and still not even come close to finding what you need.
Why Biopharma Data Architecture Fails as Organizations Scale
Organizations don’t set out to build fragmented systems. It happens because they are adding tools reactively, based on the needs at the time, causing the digital infrastructure to become a patchwork.
Teams are building as they go, which becomes a normalized behavior. As Oscar puts it: “It becomes this badge of honor: we don’t know how to prioritize anything, but we’re just gonna keep on moving through.”
In data architecture, building as you go creates a patchwork that becomes increasingly difficult to scale.
When each team customizes its setup without a shared architecture, the company ends up with one-off setups that aren’t similar to anything else. You end up with data flowing in separate directions rather than into a unified structure. This makes it hard for anybody to compare apples to apples.
Here’s how the patchwork scenario plays out in real life:
- Naming conventions break down. One group calls it a program, another a project, and a third an initiative. This means AI tools, dashboards, and reporting systems will produce unreliable outputs.
- Teams lack shared definitions. When one hour of Sally’s time logged in the system does not mean the same thing as one hour of Bob’s time, resource projections become unreliable. For example, Bob and Sally may have different access levels, conventions, and assumptions about what the entry covers.
- Data loses credibility. When definitions and names differ, data collection isn’t set up correctly, which typically comes to light once the data has been collected. Rather than restarting, teams retroactively redefine what the data means, leading people to question the data altogether.
The trouble is when this questionable data becomes the basis for risky decisions, such as scaling up or down in staff. The data suddenly becomes “really accurate” when it supports adding headcount, but gets questioned the moment it suggests cutting it.
What Teams Need to Build Before Adding More Tools
The answer isn’t to add more tools or to rebuild during phases of growth. And it’s certainly not to count on AI-powered tools because they’re only as good as the data you feed them. If the data is disorganized, the output will be, too.
The answer is to design the architecture so that when systems change, the underlying structure holds. With the right architecture, data flow, and people managing their systems, the data should always tell an accurate story.
As Oscar explains: “You have to expect that the system that you’re building might change, but that architecture can very much stay the same. And that’s how you build resilient systems.”
Building resilient systems starts with listening to and understanding what matters most to the company. Then, designing a data structure based on the questions you know leaders will ask. It’s about building a system that people will feel empowered to use.
Here are a few considerations when building a resilient data architecture.
- Build for the data that matters. At Sigma, one of the best ways we can understand what matters to a company is sitting in on two or three of their meetings. By listening to those conversations, we can identify the eight or nine questions that keep coming up and build data dashboards to answer them.
- Make it easy. Our goal is to design the data structure so it’s easy for both the person entering the data and the person reading it. If you want certain data points so you can make a specific decision, you have to make it easy for the person to provide that information. If the system is too complicated to use consistently, the data will always be unreliable — no matter how good the architecture is.
- Users need to own their software. The primary user of a system should know their software better than anyone else. They should actively coordinate with adjacent teams on naming and structure so when the company grows, they’ll be able to connect the various systems.
How to Start Diagnosing Your Digital Operations
Every team has strengths and weaknesses in the context of how data is organized. The best thing to do is take some time to think about how the data is organized at your organization.
Ask yourself questions like:
- What does it feel like at your organization?
- Is the data accessible?
- Are you constantly searching for things?
- Does everyone know where all the data lives?
- Are you pulling things out that you thought were sources of truth, only to find out they were drafts that were never continued or builds in Monday, Trello, Smartsheet, or Notion that just stopped?
Digital operations risk is something we regularly work on with biopharma teams. So we’ve taken the topic and put it into one of our Risk Assessments. The goal of the Risk Assessment is to help companies identify their biggest issue, which could be anything from a lack of data integrity to a broken data-to-decision flow. Once you identify the right issue, you can create more mature, healthier digital operations.
The Risk Assessment is an opportunity for your entire team to reflect on how well things are set up and to gauge whether everyone perceives the strength of your digital operations the same way. Often, they don’t.
When your digital operations are strong, decisions stop being made on gut feelings and incomplete pictures. They start being made on data people can actually trust, which moves the organization forward and the therapeutics one step closer to patients.
Key Takeaway
The system you’re building today will change, but the underlying architecture doesn’t have to. Design for where you’re going, not just where you are.
Enjoy this blog? Be sure to listen to the accompanying podcast episode: 0304. When Data Exists but No … – Lean By Design – Apple Podcasts
