Reading uptime bars

Every asset on the status page carries a 90-day bar: ninety thin vertical slats, one per calendar day, oldest on the left, today on the right. It’s the fastest way to answer “how has the elevator really been?” - a month of trouble is a visible red smear, not a number someone has to look up.

The colors

Each slat is painted by the worst thing that happened to the asset that day, in the building’s own timezone:

Color	Day status	Meaning
Green	operational	Nothing was wrong that day
Amber	degraded	A non-downtime incident was open or overlapped that day - the asset worked, but not fully (e.g. one treadmill out in the gym)
Red	down	A downtime incident overlapped that day - the asset was out of service for some or all of it
Blue	maintenance	A planned maintenance window touched that day, and nothing worse happened

Precedence when several things overlap the same day:

down  >  degraded  >  maintenance  >  operational

So a day with both a scheduled inspection and an actual outage paints red, and a planned-work day with no problems paints blue. Blue is the important nuance: planned maintenance is not an outage. The monthly fire-alarm test shows blue, not red - and it doesn’t dent the uptime percentage either (see below).

Hover over any slat (on the web) to see the exact date and that day’s status.

Downtime vs. degraded - who decides?

Each incident has an isDowntime flag, set by staff: was the asset actually out of service? An elevator stuck between floors is downtime (red). “Weeds in the north garden bed” is a real incident, but the garden isn’t down - that’s degraded (amber). Only downtime incidents on an asset count against its uptime percentage; degraded days affect the bar’s color but not the number.

The uptime percentages

Next to each bar you’ll see 30-day and 90-day uptime, always with two decimals - e.g. 99.74%. The math:

uptime over window W = 1 − (total downtime overlapping W) / (length of W)

In plain terms:

• Take every downtime incident on the asset. Each one contributes the interval from when it was created to when it was resolved (or to right now if it’s still open).
• Overlapping incidents don’t double-count. If two reports cover the same outage hours, those hours count once - the intervals are merged before summing.
• Maintenance windows are excluded. Any downtime that falls inside a planned maintenance window doesn’t count. A 4-hour scheduled elevator modernization costs 0% uptime; an unplanned 4-hour outage on a 30-day window costs about 0.56%.
• Divide the remaining downtime by the window length, subtract from 1, and report it to two decimal places.

A feel for the numbers on the 90-day window:

90-day uptime	Roughly how much unplanned downtime
100.00%	None
99.90%	~2 hours
99.50%	~11 hours
98.00%	~43 hours (almost two full days)
95.00%	~108 hours (four and a half days)

Reading the bar like a board member

• A solid red block = one long outage. Click into the asset to find the incident behind it and how long acknowledgement took.
• Scattered red singles = recurring short failures - often worse news than one long outage. The asset’s detail page shows the incident history (and, for staff, health trends like mean time between failures).
• Regular blue slats (e.g. one per month) = a recurring maintenance window, like the monthly fire-alarm test. Expected and healthy.
• Amber stretches = lingering partial problems. The thing “works”, but someone’s been living with a limp for weeks.
• Compare 30-day vs. 90-day - a 30-day number that’s much better than the 90-day one means things have improved recently; much worse means a recent incident is dragging it down.

Where the bars appear

• The home status page (/) - every asset, with the summary banner up top and scheduled-maintenance section when windows are upcoming.
• Asset detail pages - the bar plus the full incident history behind it.
• The mobile app - same bars, same colors, on Status and asset detail.
• The lobby kiosk (/kiosk) - oversized bars on a dark background, readable across a lobby, refreshed every 60 seconds.

One last point worth knowing: nobody paints these bars by hand. They’re computed from the incident record - the same record you can read, comment on, and verify. If the bar is green, it’s because no one reported otherwise; if something’s wrong and the bar doesn’t show it, report it.