Frequently asked question · Custom Reports
How fresh are the numbers?
Your prompts run once a day, and only completed runs inside the window count. A report's metrics are cached for six hours per filter combination, and the cache is cleared when a new run lands, so a report is at most a few hours behind the last daily run. It is daily data, not a live feed.
Where the daily rhythm comes from
Every number in a report is built from prompt tracking, and prompt tracking puts your workspace's questions to the engines once a day. There is no second collection to pull from between runs. A report can only be as current as the most recent completed run, and only completed runs inside the window count, so a run still in progress does not show up as a half-filled day.
What the six-hour cache does
Rendering 24 metrics over a window is real work, so the platform keeps the computed figures for six hours per filter combination. Open the same cut twice in an afternoon and the second load reads from cache. Change any filter and you get a fresh computation for that combination, which is then cached on its own.
The cache does not hold back new data. When a new run lands, the cached figures are cleared and recomputed, so at worst a report is a few hours behind the last daily run.
What this means for reading a report
- Opening the report twice in one day will usually show the same numbers, and that is expected.
- A change you see from one day to the next is a change in the answers, not a change in the cut.
- Trends plot daily up to 30 days, weekly up to 90, and monthly beyond, so a single day's run rarely moves a long window on its own.
- The Movers tab compares whole windows, so it is the place to read what changed rather than the day-by-day line.
Answered on
Custom Reports
A report is a saved filter set over your AI-answer data: 24 metrics, any window, engine, country, prompt, tag, or persona, compared with the previous period.
More questions about Custom Reports
See what AI is already saying about your brand.
A free audit shows you exactly where you stand across ChatGPT, Gemini, Perplexity, Copilot, Claude, Grok, and AI Overviews, and what it would take to close the gap.