The Trust Triangle

A data system is not trusted because it is fast, or pretty, or expensive. It is trusted when three things hold at once.

The three corners#

Accuracy — the numbers reflect reality, and when reality changes, the numbers change with it. Accuracy is not a one-time validation; it is a property you have to keep paying for.

Explainability — someone can ask "why is this number what it is?" and get a real answer that traces back to a source. A correct number nobody can explain is indistinguishable, in practice, from a wrong one.

Security — the data is only seen and changed by those who should. The moment people suspect a number could have been tampered with, or that their own data isn't safe, the other two corners stop mattering.

Why a triangle, not a checklist#

The corners are not additive; they are multiplicative. A system that is accurate and explainable but insecure isn't "two-thirds trusted" — it is untrusted, because no one will stake a decision on data they believe could leak or be altered. The same is true for any missing corner.

Trust is the product of the three, not the sum. Zero on any axis zeroes the whole.

How to use it#

When a stakeholder says "I don't trust this dashboard," the Trust Triangle turns a vague feeling into a diagnosis. Ask which corner is weak: Is the number wrong, unexplainable, or unsafe? The fix is almost never "add more charts." It is restoring whichever corner is missing — and then proving, visibly, that you did.