Each column is a time window
UptimeScout groups history into 36 intervals each for three hours, twelve hours and a day. The intervals are 5, 20 and 40 minutes long. Equal-width columns make time comparable; their color identifies a categorical signal, not complaint volume.
The most frequent saved signal determines the column color. A single disruption observation does not turn a mostly elevated interval red. When counts tie, the most recently observed of those tied states determines the color. Select the column to inspect the actual mix before describing the whole interval.
Samples are not users or duration
The inspector counts saved observations for each state. Ten observations do not mean ten affected users. A ratio calculated from those counts does not establish a measured uptime percentage because the product is not a continuous end-to-end measurement from every user network.
Likewise, the number of colored columns is not a verified outage duration. Grouping makes a chart readable but does not turn its time windows into confirmed incident boundaries.
Read neutral columns as unknown
An interval without a usable observation remains neutral. It may occur before collection began, between observations or when a usable result could not be obtained. It does not mean that the service failed or that it was working.
The latest observation can also become stale. Keep that separate from an interval’s historical content: a saved past observation remains a historical fact about the recorded signal, while the current situation may be uncertain.
Use the time range to answer a specific question
Use three hours to inspect recent changes, twelve hours for the surrounding context, and a day for a broader pattern. Keep the region fixed when switching views. Read the selected interval’s start, end and last observation time before comparing it with your own error log.
If you need data for your application, use the history endpoint and preserve region and timestamps. Explain unknown intervals in exports and reports. A useful report says what the observations support and what they do not establish, rather than hiding uncertainty behind a smooth line.
Content reviewed on 18 November 2026.
Markdown


