We like to imagine data as clean and neutral, a mirror that reflects the world as it is. But most of the time, it reflects us – the people who built it in a hurry.
Data doesn’t misbehave on its own. It learns by our example. We love to say “the data doesn’t lie.” That’s true. But it also doesn’t tell the truth. It tells our truth, frozen at the moment of convenience.
Every table carries fingerprints. Every metric hides a shrug. Every “source of truth” is really a museum of yesterday’s intentions. When we call something a data problem, we’re often just seeing our own reflection in CSV form.
We chase purity through pipelines, dashboards, and frameworks. We call it data quality but what we really want is absolution – the comfort of believing our systems can clean up after us.
The myth of clean data is the myth of clean people. We tidy numbers to avoid tidying decisions. We ”fix” pipelines instead of conversations.
Every governance project begins as a moral wish: If only people did the right thing automatically.
Spoiler: They don’t and they never will.
Start smaller. Run a data retro the way you’d run a sprint retro. Ask the same questions you’d ask a team:
- What broke trust this cycle?
- Which definition felt dishonest?
- Where did we guess instead of agree?
You’ll see the blockers are rarely technical. They are about clarity, context, and courage. Data work is people work wearing a spreadsheet.
Another exercise:
Pick one dataset you rely on often. Now imagine it describes you personally. Would you feel seen? Or simplified?
If that question stings, it means the data carries human shortcuts. Bias isn’t an accident. It’s efficiency dressed as logic.
Empathy fixes data faster than validation scripts ever will.
Three small habits that quietly change everything:
- Run interpretation checks before schema checks. Ask if someone would understand your dataset without you in the room.
- Document intent, not just structure. A column without its reason already leaks meaning.
- Track decisions, not only values. The moment of choice is where truth bends.
These aren’t governance rules. They are acts of care. They build trust quietly, like maintenance done after hours. Over time, trust becomes the real framework.
Data behaves like us. If we rush, it rushes. If we care, it matures.
We keep trying to make data honest through control. But honesty cannot be enforced. It can only be practiced.
If we want our data to tell the truth, we have to start by doing the same.
