Spring is a season for opening windows, clearing out what accumulated over the winter, and taking stock of what still belongs. Marie Kondo built a methodology around that instinct.  Pick up an item and ask whether it still serves you.  If it does, keep it.  If not, thank it and let it go.

Your data deserves the same assessment.

Many organizations are not working with overtly broken data. They are working with data that has quietly stopped being useful, which makes it harder to address because there is nothing obvious to spot. The problem is usually drift: the data, the people collecting it, and the decisions it once supported have moved on, but the data remained.

It shows up in familiar ways:

  • You’ve not reviewed your program intake form since the program first launched, even though it made sense at the time.
  • You still use a grant-mandated assessment from four years ago, even though only three questions were ever useful, and the grant is over.
  • Your data system, created and implemented by a previous staff member, is being navigated largely by instinct by the current staff because they’ve left, and there is no (or not enough) documentation.

None of this requires starting over. It requires looking at what you have.

What made it messy, and what can you do about it?

Understanding why data loses usefulness usually points to one of the following:

  • Your data no longer reflects what you do. Programs evolve. Forms do not. Fields that made sense two years ago capture information nobody looks at, while what could actually inform decisions is missing.
    • First step: Make or revise a program logic model; it is the clearest tool for getting this mismatch back in alignment. When every data point connects to something in the model, collection becomes intentional rather than habitual.
  • The data exists, but the format requires constant cleanup. The most common offenders: open text fields that produce inconsistent entries, duplicate records, empty required fields, handwritten data that has to be transcribed, and dropdowns where “other” became the default because the right option was never added.
    • First step: Are there patterns in that open text that you can now convert to a dropdown? Workflows you can create to regularly clean, delete duplicates, or transcribe data? 
  • One term carries five different definitions, producing five different numbers. This surfaces when departments pull the same metric and arrive at different answers.
    • First step: Have a cross-team conversation: uncover the definitions in use, determine which one serves the mission, and retire the others or give them distinct names.
  • Your systems, processes, and people are out of alignment. Data lives in silos. Decisions get made without the information that would have made them better. Mapping where data is created, who touches it, and where it goes surfaces where things are breaking down and why.

Cleaning your data without addressing these root causes is like mopping the floor while the pipe is still leaking. You will be back with a bucket next week.

WWKD?

Before you reorganize anything, before you migrate, audit, or hire anyone, ask “What would Kondo do?” “Pick up” a dataset, tool, or process and ask whether it is still serving you. Bring in the people closest to it. They will tell you things no audit will surface.

This might produce the exact to-do list for which you have no time or budget. Have staff help design a low-lift interim solution that meets the need and serves your team. Capacity is part of the decision, not an obstacle to it.

That is your spring cleaning project. Not your entire database. Not a system overhaul.

About the Author: Dana Benjamin is a Principal Consultant at Back of the Napkin Consulting.