Field note · May 28, 2026
How do I explain a confidence interval to a stakeholder?
Do not explain it. Use it.
A question that came in more than once, so the answer goes here.
How do I explain a confidence interval to a stakeholder?
Do not explain it. Use it.
"Conversion is 3.2%, and with this much data we can pin it to somewhere between 2.9% and 3.5%." Nobody needs the word "confidence" or a paragraph about repeated sampling. They need to know the number is fuzzy and by how much.
Where it earns its place is in the next sentence. "Last month it was 3.0%, which is inside that range, so this is not a change we can distinguish from noise." That sentence prevents a meeting.
The failure mode is presenting the interval and then discussing the point estimate as if the interval were decoration. If the range covers both "good" and "bad", the honest answer is that you do not know yet — and saying so early is much cheaper than saying it after a decision has been made on the point estimate.
The follow-up question is always the same and worth pre-empting: how do we make it narrower. The answer is more data, and the exchange rate is bad — halving the width costs four times the sample. Knowing that number turns "can you be more precise" from a request into a trade-off with a price on it.