Companies with marketing collateral in the last fortnight.

Rollout membership rides on every row rather than filtering the list: enrolment is not a precondition for sweeping — force_guideline_sets puts the trial runs on the requirement path either way — but the operator has to see whether the company's own analyses are already there.

The count is what gather_cohort would call the window total, not a looser one. See the filter below for why the two agreeing matters.