Crowd & occupancy
Is tonight normal for each room?
Each room’s crowd right now against its own usual crowd at this time on this weekday. A room is only ever compared with itself — the only fair comparison when one holds 90 and another 400.
What it is
Is tonight normal for each room? takes each room's crowd right now and sets it against that same room's own usual crowd — at this half-hour, on this weekday. Tonight's observed band is compared with the median band the room has reached at the same half-hour across its recent same-weekday nights, and the result is read simply as busier than usual, about usual, or quieter than usual.
A room is only ever compared with itself. Two rooms both reading "busier than usual" are not doing equally well — they are each ahead of their own pattern, which is the only fair comparison when one holds 90 people and another 400. There is no group average and no cross-room benchmark here, by design.
How to use it at your venue
Operators tend to treat a "busier" or "quieter than usual" reading as a prompt to look, not a verdict — a reason to check the room before acting on it, and to pair it with the absolute crowd band before making a staffing or door call. The comparison tells you a room has stepped off its own pattern; it does not tell you the room is busy or quiet in absolute terms, so the two are worth reading together.
A grounded use is to trust the signal more as the baseline fills in: a room open only a few weeks has a thin, noisy median, while months of same-weekday, same-half-hour history make "usual" mean something. The self-only framing is deliberate and matches how operators already judge a single location against its own comparable history rather than against other units[1], and how any process is judged against its own past behaviour to tell ordinary variation from a real departure[2]. For the same reason, operators often avoid using this to rank one room against another — it is a self-referential read, meaningful for a room's own trajectory, never a leaderboard.
How to read it
Read it as a direction off a room's own baseline, at band resolution: tonight's observed band is above, level with, or below the median band that room usually reaches at this half-hour on this weekday. "Busier than usual" for a small room and "busier than usual" for a large one are the same statement about each room's own pattern — not a claim they are performing alike. Treat a short baseline as noise: a handful of measured nights can make an ordinary night look exceptional in either direction.
What it doesn't mean
⚠ What it can't tell you
It says a room is off its own pattern, not why. A one-off event, weather, or a nearby closure will move it, and none of that is visible here. It never compares one venue to another — a bigger room reading "usual" is not outperforming a smaller room reading "busier." Because it works in crowd bands rather than headcounts, it is not a precise occupancy figure and does not lean on the occupancy model being calibrated. And a room without a live reading, or without enough measured nights at this hour, is listed as not compared — never counted as normal. No reading is a state, not a quiet night.
How it's calculated
Computed client-side from data already loaded, with no extra reads. RollCall takes the room's current live crowd band and the current half-hour slot, then looks up the median band that room reached at the same half-hour across its recent same-weekday nights. The difference between the two is bucketed into busier / about / quieter than usual. If the room has no live reading, or too few measured nights at that half-hour to form a trustworthy median, it is set aside as not compared rather than assigned a normal reading. Nothing is averaged across rooms and nothing is scaled up to the group.
The exact method
The current slot is resolved at half-hour resolution from the room's local clock. band(observed_now) is the index of the room's current live crowd label on the ordered band scale (Empty → packed); median_band is the median of that room's observed bands at the same half-hour across its recent same-weekday nights, built from scan history. A cell only yields a comparison when its count of measured nights clears the baseline floor (cell.nights >= RC_BASE_FLOOR); otherwise the room is marked no baseline. With no current live band, the room is marked no read. The underlying typical same-weekday curves are themselves only drawn where a room clears the peak floor of four measured nights (RC_PEAK_FLOOR = 4). Inputs are the current live band and the same-weekday half-hourly crowd baseline, both from door scans — no POS. Computed by gdTonightVsTypical(rows) at, typical curves via rcTypicalPeak. The single-venue read-model analog is vsTypicalPoints.
The research behind it
Compare a room tonight only to how that same room usually behaves on this same day of the week at this same time — not to a different room, a different night, or an industry average. A quiet Tuesday and a packed Saturday are different businesses; mixing them (or mixing rooms) into one baseline hides real signal. Judge each room against its own history, at the matching moment, and nothing else.
-
Same-store sales (comparable-store sales) ✓ verified
Read the source ↗ -
Economic Control of Quality of Manufactured Product ✓ verified
Read the source ↗
Confidence & caveats
"Sure" here means the direction is honestly computed at band resolution against a room's own history — not that the underlying occupancy is measured to a person. This is not a named academic metric; it is an operational read built on two well-established foundations: the standard practice of comparing a location only to its own comparable history[1], and the foundational statistical-process-control principle of judging a process against its own baseline to separate normal variation from a real departure[2]. Confidence rises with the baseline: below the measured-nights floor a room is withheld as not compared rather than shown on a thin median. The comparison is never cross-venue, never a total, and a missing reading is a state, not a zero.