Crowd & occupancy
When is the whole group soft?
Half-hours where most of your rooms typically sit at Empty or Chill on this weekday. One quiet room is that room’s business; the same quiet hour across the group is when a group-wide offer has the least to cannibalise.
What it is
When is the whole group soft? flags the half-hours where most of your rooms typically sit at Empty or Chill on a given weekday. It is a count of states, not an average of bands: for each half-hour it asks how many of your baselined rooms are habitually quiet at that time, and surfaces the hour only when the quiet rooms are both numerous and a majority.
The distinction it draws is deliberate. One quiet room at 9pm is that room's own business. The same quiet half-hour repeating across most of the portfolio is a different kind of fact — it is a window where a group-wide offer has the least busy trade to pull from, and so the most room to add incremental traffic rather than discount people who were coming anyway.
How to use it at your venue
Operators tend to read the flagged half-hours as candidates — times to test a group-wide offer, a cross-venue push, or a staffing or comp decision — rather than as instructions. The logic comes straight from restaurant revenue management: measure demand across time periods, find the genuinely slack "shoulder" windows, and aim demand-shifting activity there, because that is where an offer captures new, price-sensitive traffic instead of cannibalising business a busy room would have earned at full price anyway.[1] The same literature frames the levers you'd reach for as pricing and duration of use, applied to the slack window rather than the night as a whole.[2]
A grounded refinement: the research on how guests perceive these moves suggests a time-of-day or daypart offer tends to read as fair, where a blanket or opaque discount can read as arbitrary — so a quiet-window offer is worth framing as belonging to that hour.[3] As with any revenue-management lever this is a starting hypothesis, not a guarantee: sanity-check a flagged window against what you already know locally — events, weather, staffing — and test it against real redemption and lift before committing budget.
How to read it
Read a flagged hour as "most of my rooms are reliably soft here on this weekday," nothing more. It is measured against each room's own same-weekday history, so it describes a habit, not tonight. When coverage is complete the summary speaks of the "whole group"; when only some rooms have a usable baseline it narrows its own wording to "these rooms," and you should read it as a statement about the rooms it could see, not the portfolio. The "N of M rooms" label tells you how many rooms actually had enough measured history to vote at that hour.
What it doesn't mean
⚠ What it can't tell you
It is a count of quiet states across rooms, never a mean of their bands, and never a comparison of one venue against another — every room is judged only against its own history. It describes what usually happens, not what will happen tonight, and it cannot tell you an offer would sell — only that the hour is habitually soft across the group. A room without enough measured nights at a half-hour does not count as quiet; it simply does not vote, so "N of M" is a coverage statement, not an adequacy one. Absence of a flag is not a promise the group is busy — only that no half-hour cleared the majority test.
How it's calculated
Computed on your device from crowd baselines already loaded for the group — each venue's typical half-hourly band for this weekday, built from door-scan history. No POS data is involved. For every half-hour it counts how many rooms have a usable baseline (the "measured" rooms) and, of those, how many are typically at Empty or Chill (the "quiet" rooms). A half-hour is surfaced only when the quiet rooms are at least two and make up a clear majority of the rooms that had a baseline there. Nothing is written back or added into a total — it is a reading, recomputed each time you open the card.
The exact method
Computed by gdQuietWindows(rows), client-side in. For each half-hour slot it walks every venue's same-weekday baseline cell, increments measured when the cell clears the baseline-nights floor (cell.nights >= RC_BASE_FLOOR) and quiet when that cell's median band is Empty or Chill (cell.median <= 1). The window is kept when measured >= 2 && quiet >= 2 && quiet/measured >= 0.6 — at least two rooms quiet, and quiet rooms a 60% majority of the rooms with any baseline there. The comparison basis for each room is its own same-weekday history; there is no shared baseline and no cross-venue normalisation. Under partial coverage the section title flips from "the whole group" to "these rooms" (honesty strings pinned in test-sites.js), driven by the coverage resolution loop.
The research behind it
Don't discount when you're already busy — find the times when most of your venues are genuinely quiet at the same time, and put your promotional energy there instead. That way an offer brings in new business rather than just giving a discount to people who would have shown up (and paid full price) anyway.
-
Restaurant Revenue Management: Applying Yield Management to the Restaurant Industry ✓ verified
Read the source ↗ -
The Strategic Levers of Yield Management ✓ verified
Read the source ↗ -
Perceived Fairness of Demand-Based Pricing for Restaurants ✓ verified
Read the source ↗
Confidence & caveats
The test itself is exact — it is arithmetic over stored baselines, so there is no band or estimate on whether a half-hour clears the threshold. What is not certain is the world the flag points at: it rests on each room having enough measured same-weekday nights (the RC_BASE_FLOOR floor), and rooms below that floor are silently absent from the vote rather than counted as busy. The 60% majority and two-room minimum are fixed thresholds, not tuned per group. And the grounding is a principle, not a proof: RollCall's specific multi-venue, half-hour, occupancy-state formulation is its own — it applies the established off-peak demand-management principle from hospitality revenue-management research,[1] it was not itself published there. Treat a flagged window as a well-evidenced place to look, not a verdict.