The "?" button on every KPI tile

Every KPI tile on every report in the Report Library carries a small ? button, with a tooltip that reads "Explain this number." Clicking it opens a popover holding a single sentence of plain-language explanation for that tile — something like "Down mainly because one practitioner had noticeably fewer appointments this period" rather than a repeat of the number you're already looking at.

The point of this feature is narrow and specific: KPI tiles are compact by necessity, and a single number rarely tells you why it moved. "Explain this number" exists to answer that one follow-up question without making you leave the report, open a different view, or dig through raw data yourself.

What you'll learn
  • How to open an explanation for any KPI tile, and what a good explanation actually looks like
  • Why the explanation is generated from a fresh server-side check, not the number already on your screen
  • What "cached per button" means for repeated clicks during the same session
  • What you'll see while it loads, and if it fails to load at all
Explain this number popover open on a KPI tile
A report KPI tile with the "?" button clicked, showing the open popover with a one-sentence explanation and the sparkle icon marking it as AI-generated.

Opening an explanation

1

Open any report with KPI tiles

Most reports in the library show KPI tiles across the top.

2

Find the "?" button on the tile you're curious about

Hover over it to confirm the "Explain this number" tooltip before clicking.

3

Click it

The popover opens showing "Explaining..." while the server-side check and explanation run.

4

Read the explanation

A single plain-language sentence appears in the popover once it's ready.

Why it re-runs the report instead of trusting your screen

This is the detail that distinguishes "Explain this number" from a simple tooltip or a static help snippet: it doesn't describe whatever value happens to be sitting on your screen at the moment you click. Instead, clicking the ? button triggers a genuine server-side re-run of the report, scoped to your own access permissions, and the explanation is generated from that fresh result.

The reason for going to that trouble instead of just reading the displayed number back to you is trust: a number rendered in your browser could theoretically be stale, cached, or affected by a filter you forgot was applied. Re-running the report server-side, under your own permission scope, means the explanation is always grounded in a number ClinyPal just recalculated on the spot for you specifically — not a client-side echo of whatever's on screen.

Never trusts a number from your screen The explanation is built from a server-side re-run of the report, scoped to your own access permissions — not from whatever value happens to be rendered in your browser at the time you click.

Once an explanation loads for a given tile, it's cached per button for the rest of that session. Clicking the same ? again shows the cached explanation instantly rather than triggering a second re-run — useful if you're comparing a few tiles back and forth and don't want to wait for the same explanation twice. Reloading the report, or opening it again in a new session, clears that cache and the next click re-runs fresh.

AI-generated, fails silently The explanation text itself is written by AI, marked with a sparkle icon, working from the server's fresh re-run of the report. If it can't be generated, the popover shows "Couldn't load an explanation" — the KPI tile and the rest of the report are unaffected.

Reading the explanation critically

An explanation is meant as a starting point for your own judgment, not a final verdict. It's written to be plausible and grounded in the fresh re-run, but it's still one sentence summarizing what can be a genuinely complicated shift in the underlying data. If a tile's movement is surprising or consequential enough to act on — staffing, budget, or a conversation with a practitioner — treat the explanation as a pointer toward where to look next in the full report, rather than the whole story on its own.

FAQ

Was this article helpful?
Your feedback helps us improve our documentation.
Rate this article: