Evidence buying guide: A filter can change the number without bringing in fresh data
A fictional exhibit-label snapshot separates changing a filter from replacing the underlying data batch.
When Evidence introduced Universal SQL for its open-source framework, it described caching source data and recalculating filter results against that cache. In that architecture, changing a selection can produce a new result even though changes to the original records have not reached the page. This does not describe every Evidence Studio deployment.
Change the filter once and the snapshot separately
Make a fictional morning inventory: three labels at the blue table and two at the orange table, five in all. A paper report quickly changes from five under All to three under Blue, showing that filtering works on the same snapshot. Add a blue label to a separate piece of paper without replacing the report’s morning snapshot. Selecting Blue again should still return three: reliably changing filters is not evidence of reading the new inventory. Only after switching to the afternoon snapshot does Blue show four and All six, demonstrating that the two results came from different data versions.
Write two independent checks: Does changing the filter return 5 and 3? Does replacing the input batch return 6 and 4? Even correct numbers are not enough for handoff if the page does not identify its batch. If the numbers have not changed, distinguish a stale snapshot from an incorrect filter definition.
When should the report be interactive?
Teams that regularly slice the same data batch in several ways and can label it clearly have a reason to compare interactive reporting. If readers must see every source-record change, check refresh frequency, delays and what happens when data becomes stale; quick page responses do not answer those questions. If handoff only needs a fixed daily total, a labeled static table may be cheaper to maintain. This exercise only counts labels on paper; it connects to no database, writes no query and does not test performance.
Content note:This article is based on publicly available information and is intended as a reference for evaluating tools. Features, pricing, and related policies may change at any time; please refer to the latest official information.
Original source:evidence.dev