Data Literacy
Data literacy is three skills: read a number, question its source, know what it does not say. The reading errors literate people catch — averages that hide, survivorship, missing denominators — and how to build the skill into an organization.
What is data literacy?
Data literacy is the ability to work with evidence the way ordinary literacy works with text: read a number, question its source, and know what it does — and does not — say. It is the decisive skill of the Competence pillar, because every other investment this book describes — the single source of truth, the measurement, the dashboards — pays out only at the moment a person reads a number and decides something. If that reading is wrong, everything upstream of it was wasted.
The skill is timeless. The tools for producing numbers change constantly; the ways a number can mislead a reader have not changed in a century.
The reading errors literate people catch
These are the recurring failure modes — each one older than computing, each one still deciding budgets today:
| Error | What it looks like | The literate question |
|---|---|---|
| The hiding average | “Average response time is fine” while one customer segment burns | What does the distribution look like? |
| Survivorship | Studying churned customers is hard, so conclusions come from the ones who stayed | Who is missing from this data? |
| The missing denominator | “Errors doubled” — from one to two | Out of how many? |
| Correlation as cause | The metric moved after the campaign, therefore because of it | What else changed at the same time? |
| The convenient window | The chart starts exactly where the trend flatters | Who chose these dates, and why? |
A literate organization is one where these questions get asked out loud, by anyone, about any number — including the flattering ones.
What literacy needs from the other pillars
- A single source of truth to be literate about. Where numbers disagree between reports, literacy degenerates into arguing about whose number is real. Shared reading requires a shared text — that is Foundation’s job.
- Definitions that travel with the data. “Active user”, “churn”, “margin” mean different things in different mouths. Literate reading needs the definition recorded next to the number, not in someone’s memory.
- Measurement worth reading. Observation produces the numbers; literacy is the demand side. Each justifies the other.
Building it — behavior, not courses
Training has its place, but literacy is demonstrated in behavior, and behavior is shaped by norms:
- Make “where does this number come from?” a respected question. In a literate organization it is due diligence; in an illiterate one it is treated as an attack. Leadership sets which one you are — by asking it first, about its own numbers.
- Decide from pre-read numbers. Meetings that decide something start from the relevant measures, already distributed. Reading data becomes part of deciding, not a specialist’s errand.
- Let anyone trace a number. If only the data team can find out what a metric means or where it came from, literacy is capped at the size of the data team. The path from any dashboard back to definition and source must be walkable by a curious non-specialist.
- Measure the behavior. Not course completions — behavior: how many decisions cite their evidence, how often definitions get looked up, how quickly a wrong number gets challenged. What you measure here is what you actually believe about literacy.
Conclusion
Data literacy is reading, questioning, and knowing the limits of what a number says. The five reading errors above will still be misleading readers in fifty years, whatever produces the numbers by then — and the organizations that catch them will still be the ones where questioning a number is normal, definitions live next to the data, and decisions start from the evidence. Build that, and every other pillar of FORCE compounds; skip it, and the best measurement in the world is a book nobody can read.