Your live records — the invoice you just posted, the customer's balance — are immediate. The analytics are not, and the difference is deliberate.
Analytics are computed from published snapshots of the business data. On a rhythm — some sources refresh every fifteen minutes, others hourly, others daily, each source on the cadence its volume justifies — the system assembles a complete, consistent copy of everything the measures read, and publishes it. Every cockpit and dashboard then reads that published snapshot until the next one lands. The freshness chip on each cockpit names the snapshot you are looking at and when it was published, in plain form: a time, a date, and the snapshot's number.
Three consequences follow, all of them features rather than faults:
- Everyone sees the same numbers. Two people opening the Sales cockpit at the same moment read the same snapshot; a disagreement is about meaning, never about whose refresh happened first.
- Numbers are reproducible. Re-open the page, re-send the briefing: the figures are the snapshot's figures, stable until the next publication.
- A failed refresh serves nothing half-made. If a publication cannot complete, the previous snapshot keeps serving, untouched. Stale-but-whole beats fresh-but-torn, and the chip tells you exactly how stale.
So: post a large invoice and the customer's own page shows it at once, while the Sales cockpit picks it up at its source's next refresh — fifteen minutes, an hour, or overnight, depending on the source. When a figure matters to the minute, the chip tells you which snapshot you are holding it against.