duo-proprietary-data-reports

If your product has scale, you are sitting on a dataset no journalist or academic can obtain. This skill is about converting it into a recurring public report with a fixed window and published exclusions — the format that gets cited rather than discounted.

Two things separate a citable series from a press release with a statistic in it. The first is repetition against an unchanged window, which is what lets any edition report a change instead of a level. The second is disclosure, which is what lets an outsider attribute your number without hedging it.

This skill owns the recurring program. It does not own one-shot publishing triggered by a news moment, a cultural event or a commissioned survey — that belongs to the sibling skill duo-timely-data-publishing. It also does not own how the motivation or segment fields inside the report were collected and interpreted internally, which belongs to duo-motivation-segmentation.

Decide what you are measuring

Get the shape of the series right before the first edition, because the first edition sets the contract.

• [[references/fixed-window-repeated-forever]] — freeze the window, the unit, the floor and the metric definitions; the sixth edition is worth more than the first because every prior one became a baseline.
• [[references/cut-data-to-a-unit-people-belong-to]] — aggregate numbers are about you and nobody forwards them; the same rows cut to a state, city, borough or cohort are about the reader.
• [[references/two-engagement-axes-not-one-composite]] — volume per period and frequency of return can anticorrelate, so rank them separately or you hide the cohort that is cramming and about to churn.

Decide how you rank it

The same dataset produces a boring table or a story depending on what you divide by and what you lead with.