Humans Vs. Systems: How Karchem Approaches Data Horror Stories

Whether it was human error or your laboratory’s systems gone awry, data mishaps can feel daunting. At Karchem Consulting, we’re familiar with even the scariest horror story moments and how to make them right. 

When our consultants meet with a new client, they often present us with a system that doesn’t work for them, isn’t streamlined, or is generally misunderstood. Even after we’ve implemented a new system, we pride ourselves on sticking with the client through potential growing pains. A mistake users sometimes make is going into a new system thinking it can “read their minds,” rather than understanding how the system works and how to use it efficiently. 

We’re here to make sure your laboratory software not only works for you, but that you understand the why behind the way the data is recorded. Whether it’s a need for proper user training or a misunderstanding due to a system having too much flexibility, we can organize the data, identify the cause, and set you up for success. 

The Junk Drawer Effect 

Often, having too much flexibility within your laboratory data system creates chaos, not scalability. In the past, scientists have asked us if we can create a separate schema, or model, to track the same entities as the system we’re trying to implement across their lab. Our consultants were able to address that knowledge gap and explain why data should be tracked in a uniform way. With multiple entities tracking essentially the same data, it creates a recipe for disaster and confusion when it comes to downstream querying and analysis. When tracking results through a laboratory system, Karchem Consulting will always make sure your data is “FAIR” -- Findable, Accessible, Interoperable, and Reusable. Like all systems in your lab, the goal should be to enhance efficiency, scalability, and reproducibility. This consistency in data management ensures downstream data is usable moving forward, not just thrown in the junk drawer and forgotten. 

Similarly, some clients come to Karchem with a seemingly robust data model but make the mistake of defaulting to a “catch-all” entity schema. Common mistakes such as adding a free-text field when scientists aren’t sure what else to use and tacking on additional fields and schemas for each new-use case or team, are only a band-aid solution. These steps don’t consider the bigger picture and scalability. For example, if a scientist goes to look for results for sample data downstream, they may only know to look for “Sample Type A,” but if they have a random catch-all, there could be Sample Type A’s recorded under a general catch-all schema. Over time, it becomes a tangled mess that needs to be addressed, affecting reproducibility and accuracy. 

Results Without Rules 

When Karchem Consulting builds data models, we consider what fields should be required versus optional depending on their downstream use. One of the common mistakes we see in biotechs is users not taking into account the usability for not only the person using the results, but also for the person recording the data. When entering a value into a results table, if required fields aren’t defined, users could omit a corresponding unit field or record the value in a unit different from the one set for that field, thereby rendering any values collected ultimately useless without downstream investigation. In this investigation, a user may see that technically, the data exists, but practically, it can’t be compared, analyzed, or trusted. 

Alignment Issues: The Human Bottleneck 

It’s an issue that can occur in any team environment: internal disagreements derail the adoption of a strong system. When you agree to partner with Karchem Consulting, we define priorities and help teams find common ground. Whether automation engineers are setting up similar worklists in different ways, or a biology team has dozens of ways of recording the same assay, these issues are far from abnormal but end up being a human bottleneck. Because of Karchem’s cross-team work model, we tend to notice the ways various teams disagree or lack communication and can bring the team together. 

When a team doesn’t communicate, it can lead to data mishaps, including finding out your system has failed you. Maybe you found a system element that isn’t built to meet your goals, or your backup process didn’t back up as expected due to a technical error. And while that may be true, we’ve often found the real bottleneck was human error and a lack of reporting problems among teams. This problem can be hard to avoid, but getting Karchem involved in projects early can help mitigate risk. 

Few data horror stories start with your system failing. As we’ve discussed, most start with assumptions, over-flexibility, misalignment, and a lack of communication and understanding among teams. The good news is Team KC is here to help. With the right combination of training, thoughtful data modeling, and cross-team alignment, we can implement systems that support your science instead of slowing it down, making your next “uh oh” moment an easy lesson instead of a horror story. 

Ready to transform your lab's data infrastructure? Contact us at Karchem Consulting today to get started.