Citation

The five parts that make a figure checkable: the table, the row, the period cut, the definition version and the upload it was computed from. A link to a dashboard is not a citation, because the dashboard has moved on. A figure without one is a claim.

DefinitionThe five parts that make a figure checkable: the table, the row, the period cut, the definition version and the upload it was computed from. A link to a dashboard is not a citation, because the dashboard has moved on. A figure without one is a claim.

Thirty seconds

Open the table at the row, see the account rows beneath, compare the version to last month's. That is what turns a claim into a fact.

In the assistant's answer

Every number it reads carries the citation. A number it cannot cite, it did not read.

How it is used

Every figure shown carries five things: the table it came from, the row, the period cut, the version of the definition, and the upload it was computed from. With those, anyone can find the rows and get the same number.

Example

Coverage 78 percent. Citation: coverage table, row for the northern region, fiscal period 8, definition version 3, upload of 2 September. A reader opens that row and sees 312 of 400 accounts.

Where it goes wrong

A number with a source that says CRM. That is where it came from, not how to find it again.

Articles that use this term

Board and management reporting

A board narrative with a citation on every number: how to draft the commercial pack from reconciled figures

A method for writing the quarterly commercial narrative so every figure carries a citation to the calculation that produced it: the reconciled roll-up as the source, the narrative structure, the citation format that survives export, and why the model writes sentences but never numbers.

16 Sept 20263 min read AI and self-service analytics

A weekly sales digest worth reading: what a good one looks like, line by line

An annotated example of a weekly sales digest built from computed tables: the header that states the period and the checks, ten movements each with the measure, the level, the change, the threshold and the citation, the three lists that follow, untouched, slipped and dormant, the line that says what did not move, and the footer with the definition version. What each element is for, and the version of the same digest that nobody reads.

16 Sept 20263 min read How-to guides

Answer-first writing for commercial measures: the template behind every page on this site

The seven-part template every measure page on this site follows, and why: the answer in one paragraph, the formula in a line, the rows needed with identifiers only, one worked table, the roll-up and its identity, the four mistakes, and the three questions answered in full. What each part is for, the order and why it is fixed, the length each part gets, and how a company can use the same template for its own definitions document.

16 Sept 20264 min read AI and self-service analytics

Citations on every number: what a citation from an analytics tool must contain

A number in a report or an assistant's answer is a claim until it carries a citation, and a citation has five parts: the table, the row, the period cut, the definition version, and the upload it was computed from. Why each part is there, what a reader does with it, the difference between a citation and a link, how a citation survives a re-upload, and the three shortcuts that turn a citation into decoration.

16 Sept 20263 min read Board and management reporting

Customer concentration: how to measure it and what the number means

Three ways to measure customer concentration from a revenue ledger, top-n share, the Herfindahl index and the count to half, what each is good for, a worked example, why the measure is reported per segment and per rep as well as for the company, and how to tell healthy concentration from dependency.

16 Sept 20263 min read AI and self-service analytics

Deterministic first: the architecture of an analytics product that uses a model and stays right

How an analytics product uses a language model without letting it touch a number: deterministic code computes every measure from stated definitions, a tool registry names each computation, the model plans which tools to call and reads their outputs, every output carries its citation, and the model's own text is labelled as a reading. The four boundaries that make this hold, what crosses each, and what never does.

16 Sept 20263 min read