Name the evidence you don't have

The most useful sentence in any analysis I wrote this year was a caveat.

I was reviewing a year of content performance, and one dataset I needed didn’t exist in usable form: promotion records. I could see what every piece had earned; I could not see what had been spent promoting it. Which meant I could not distinguish “this article found its audience” from “this article was pushed to its audience.”

The standard move is to proceed as if the gap weren’t there — bury a limitation in a footnote and present the conclusions at full strength. Instead the caveat went on the second page, in bold, phrased as a rule:

Distribution is a hypothesis, not a conclusion.

Every finding that depended on knowing why something performed was explicitly downgraded. Findings that didn’t depend on it — what was published, how pieces connected, which formats converted — kept their full strength.

Why this works

It makes the strong claims stronger. When a reader can see you downgrading what the evidence can’t support, they stop discounting everything else. An analysis that admits its gaps is an analysis whose remaining claims have been checked. One that admits nothing forces the reader to do the discounting themselves — usually across the board.

It survives the meeting. Every serious analysis eventually meets a skeptic whose first move is to find the hole. If the hole is already named on page two, that conversation is over before it starts — and you chose the framing, not the skeptic.

It generates the next piece of work. A named gap is a commissioning brief. “We can’t see promotion data” converts directly into “track promotion alongside every piece from now on.” Buried limitations generate nothing; named ones become the to-do list.

The method, in three steps

1. List the evidence the analysis needed but didn’t get. Not general humility — the specific datasets, records or access that were missing.
2. Trace which conclusions depended on the missing evidence. Downgrade those, explicitly, to hypotheses. Use the word.
3. Say what would settle it. Name the data that would convert each hypothesis back into a conclusion, so the gap becomes a plan rather than an apology.

The whole discipline fits in a paragraph, costs nothing, and is rarer than it has any right to be. Evidence-led work is not work that has all the evidence. It is work that is honest about which evidence it has.