Reporting surfaces

That question is asked before anything is built, and answering it honestly usually removes most of the requested content. What survives goes on the surface; everything else remains available on request, which is a different thing from being absent.

Freshness is displayed, not assumed. Each surface states when its data last synchronized, and one running on stale data says so on its face instead of presenting old figures in a layout that implies they are current. A surface that has quietly stopped updating is worse than one that is down, because the broken one stops being trusted and the stale one does not.

Every figure opens into the rows behind it. This is the property that determines whether a report survives its first disagreement, and the cost of building it in from the start is a fraction of the cost of retrofitting it during an argument about whether the number is right.

Numbers are shown against something. A figure with no prior period, no target and no comparable segment is a figure nobody can act on, because acting requires knowing whether it is good, and a bare total does not say. Where no meaningful comparison exists, the absence is worth noticing — it usually means the metric is not yet decision-bearing.

Access is scoped at the outset rather than retrofitted. Margin, payroll-adjacent figures and per-account revenue are frequently the most operationally valuable numbers a business has and the ones it least wants circulating, and deciding which roles see which is a design question with a real answer. Answering it late means answering it after somebody has already seen a column they should not have.

Reports are delivered where decisions are actually made. A scheduled summary arriving in a mailbox on Monday is read considerably more often than a URL requiring someone to remember it exists, and for many businesses the digest is the product while the interactive surface is where the digest gets investigated.

Chart treatments follow the data, not the argument. Axes start at zero where magnitude is the point, two series are not put on separate scales to make them appear to move together, and a change within ordinary variation is not drawn as a trend. A surface that overstates once is discounted permanently.

Surfaces are retired. A report nobody has opened in a quarter is removed rather than maintained, because every unused surface is still a thing that breaks, still a thing that must be understood during a migration, and still a place where a stale definition can survive after the live one has moved on.