Metrics reference

This article explains what each figure in FlowAnalytics measures. The dashboard articles describe where a figure appears and what to do with it, and link here for the definition.

Every key figure card and most charts also carry an information icon in the portal, which gives a short definition next to the figure itself. That is the quickest way to check a number while you are looking at it. This article covers the same definitions with the context around them: how the figures relate to each other, what they depend on, and what commonly makes two of them appear to disagree.

How to read a key figure card

The cards at the top of a dashboard follow the same pattern:

• The value for the period and selection set in the filter bar.
• The target, shown underneath as a single value or a range, such as Target 60%-80%. A recommended target is pre-selected by Flowscape, and can be changed per dashboard under Settings.
• The comparison, showing the change against the preceding period of the same length, for example +8.7% vs last 30 days. A green upward arrow means the figure has risen, which is not automatically an improvement: a rising no-show rate is bad news.
• Setup required, in place of a value, where the configuration a figure depends on is missing.

A figure only covers what the filters include. Two cards that appear to disagree are usually answering questions about different selections, different periods or different asset types.

Where each metric appears

Utilization

An asset counts as utilized when it is either booked or used. A desk somebody is sitting at counts, and so does a desk that is booked even if nobody ever arrives.

This is the single most important thing to understand about the utilization figures, because it means a high figure is not by itself proof that the space is being used. An office where half the desks are booked and never occupied reports the same utilization as an office where half the desks are genuinely in use. What separates the two is the booking versus presence split described in the next chapter.

The three variants below differ only in how they treat time. Confusing them is the most common reason two people quote different numbers for the same office.

Average utilization

The average share of your space in use during the selected period, days and hours. An asset counts as in use whenever it is booked or a sensor detects presence, so bookings nobody attended are included. The asset utilization chart on the dashboard breaks that down.

This is the figure to use when the question is whether the office is the right size.

It is always lower than people expect, because it averages the quiet hours together with the busy ones. An office at 43% average utilization is not half empty all day; it is busy in the middle of the day and near empty at the edges.

Average daily peak utilization

The average of the highest utilization level reached each day. Each day in the period has a peak, and this figure is the average of those peaks, so it describes a normal day at its busiest.

This is the figure that decides whether you have enough desks. If 64 desks reach a typical daily peak of 58%, about 37 desks are in use at the busiest moment of an ordinary day, and the remaining 27 are what carries a bad day.

Peak utilization

The highest utilization reached at any point during the selected period. One busy hour on one day sets this figure, so treat it as the worst case rather than as a description of normal operation.

Booking versus presence

An asset can be booked, used, both or neither. Since utilization counts an asset that is either booked or used, these categories are what the utilization figure is made of, and the difference between them is where most of the recoverable space in an office sits. The asset utilization chart on Office Utilization separates the four cases.

• Booked and used. A reservation that was honoured. This is what you want to see.
• Used without booking. Someone used the space without reserving it. A high share suggests booking is not necessary in that area, or that people have stopped trusting it.
• Booked but unused. A reservation with no measured presence. This is the no-show volume, and it is space that other people could not book.
• Booked (no sensor data). A reservation on an asset that has no sensor, so presence could not be measured either way. This share is not a finding about behaviour; it is a limit on what the data can tell you. Where it is large, the utilization figure is largely a booking figure.

The last category matters when you read every other figure on the page. Where sensor coverage is partial, utilization based on bookings alone counts a booking as usage, which overstates how busy the office really is.

Room occupancy and room fit

Occupancy is a different measurement from utilization. Utilization asks whether a room was booked or used at all; occupancy asks how much of its capacity was filled while it was in use. A twelve-seat room with four people in it is fully utilized and poorly occupied.

Occupancy therefore needs people counting in the room, either from a people counting sensor or from a video bar. Rooms with presence detection only can report utilization but not occupancy.

Average room occupancy

The average occupancy per room while the meeting rooms are in use. Idle time is not counted, so this figure describes how full the rooms are when somebody is in them rather than across the whole working day.

Average seat occupancy

How many of the seats across all the selected rooms were used: the total number of occupied seats divided by the total number of seats in those rooms.

The difference from average room occupancy is what each figure treats as one unit. Room occupancy averages the rooms, so a two-seat room counts as much as a twenty-seat one. Seat occupancy counts seats, so the large rooms carry most of the figure. Where the two diverge, the small and the large rooms are being used differently, and reading them together tells you which of the two is the problem.

Average room fit

The percentage of meetings where occupancy reached at least 50% of the room’s capacity.

Room fit is evaluated in 15-minute intervals. If occupancy reached the 50% mark at any point within an interval, that interval counts as a good fit, so a meeting where people briefly filled half the room is credited with it. The measurement is deliberately forgiving: people arrive late, step out and come back, and a room that was genuinely the right size should not be marked down for the minutes when half the attendees were in the corridor.

Room fit is what tells you whether a shortage of large rooms is real. Four people meeting in a twelve-seat room register that room as fully used while wasting eight seats, and the fix is more small rooms rather than more space.

Attendance

Attendance counts people rather than assets, so it answers how many came in rather than how much space they used.

Average in office, average remote, average off work

The average number of people per day in each state across the period. Remote and off work depend on that information being available to the platform, so they read zero where it is not.

Average days in office

How many days per week a person is in the office on average. This is the figure to hold against an attendance policy, because a policy is written in days per week.

Employees outside the in-office target

The share of employees whose average days in office falls outside the target range set for the dashboard.

No-show

A no-show is a booking that was never used. The dashboard separates how the no-show was detected and how late a cancellation came, because the two lead to different actions.

No-show

The percentage of bookings that resulted in a no-show during the selected period. It includes both missed check-ins and bookings where no presence was detected, and the two are reported separately:

• No-show (missed check-in). The booking required a check-in and none was made.
• No-show (no presence). The sensors measured no presence during the booking.

Late cancellation (0-1h) and late cancellation (1-4h)

The percentage of bookings cancelled less than 1 hour before the scheduled start time, and the percentage cancelled 1 to 4 hours before it. A cancellation is better than a no-show because the space is released, but a late one releases it too late for anyone else to plan around.

Recovery rate

The percentage of released assets that were booked or used by someone else. An asset is released when nobody checks in for a booking, and the recovery rate is the share of those releases that somebody else then took up.

Recovery rate requires check-in to be activated. Without check-in there is nothing to release, so an office that does not use check-in has no recovery rate to report, however high its no-show rate is.

Where check-in is in use, this is the figure that shows automatic release doing its job. A no-show blocks an asset for the length of the booking, while a released no-show puts it back into circulation. A high no-show rate alongside a high recovery rate is a booking discipline problem that the system is already containing. A high no-show rate with a low recovery rate means the capacity is being lost outright.

Detection source

Which signal identified the no-show, either a missed check-in or a sensor. This is a data quality reading as much as a behavioral one: an office with little sensor coverage detects no-shows almost entirely through check-in, and therefore misses the bookings where somebody checked in and left.

Climate

Climate figures come from climate sensors and are reported against comfort and air quality thresholds rather than against a utilization target. All of them are shares of monitored time or of rooms, so an office with climate sensors in three rooms reports on those three rooms and nothing else.

CO₂

• Time with high CO₂. The percentage of monitored time when CO₂ was above the acceptable target of 800 ppm. Lower is better.
• Rooms with high CO₂. The share of rooms that went above the target during the period.

The CO₂ conditions chart shows how monitored time is distributed across four bands:

Temperature

• Time outside range. The percentage of monitored time when the temperature was outside the 21 to 23 °C comfort range. Lower is better.
• Rooms outside range. The share of rooms where the temperature fell outside the comfort range during the period.

The temperature conditions chart uses the same four bands:

Time in poor indoor climate

How often poor CO₂ or poor temperature conditions occurred over the period. Each point on the chart is the percentage of monitored time with poor conditions on that day, reported as two series:

• Poor CO₂. Time when CO₂ was above the acceptable target of 800 ppm.
• Poor temperature. Time when the temperature was outside the 21 to 23 °C comfort range.

Poor therefore means anything worse than the Good band, so time in the OK band counts as poor here even though the conditions chart lists it separately. Reading this chart day by day is what separates causes: a room that is fine in the morning and poor by mid-afternoon has a ventilation problem, while one that is poor from the moment the building opens has a heating or cooling problem.

Climate targets

All four climate cards carry a target, set on the Climate Overview dashboard under Settings, in a dialog called Climate Settings. Each target is a single percentage marked with a less-than sign, meaning stay below it, and each is recommended at 5%. Select Reset on one target to return it to the recommended value, or Reset All for every target.

These targets work the opposite way to utilization: a low figure is the good outcome, and 5% is a ceiling rather than something to aim for. A room above the CO₂ target for 4% of monitored time is within target; the same room at 20% is not.

What the targets change is how much poor climate you are willing to accept, not what counts as poor. The 800 ppm CO₂ threshold and the 21 to 23 °C comfort range are what decide whether a given measurement is high or outside range. They are fixed and cannot be changed, so the only thing you set here is the share of monitored time you are prepared to tolerate above them.

Cost, area and savings

These figures translate utilization into money and floor area. Unlike every other figure in FlowAnalytics they are not measured; they are calculated from values recorded on the office itself, so they are only ever as good as what has been entered.

Where the values come from

Select Company, then Offices, and open the office. On the General tab, four fields feed the analytics:

• Total area (RBA/GLA) [sqm]. The floor area of the office.
• Space cost [Eur/sqm]. The annual cost per square metre.
• Total headcount. The number of employees based at the office.
• Capacity. The number of people the office is intended to hold.

The area, space cost and headcount are also shown, read-only, as Company Information at the foot of the dashboard settings dialog, so you can check what a figure is based on without leaving the dashboard. To change them, go back to Company → Office settings.

The figures

• Total area. The floor area of the offices in the selection.
• Total rent cost. The annual cost of that area, from the area and the space cost per square metre.
• Potential savings per year. The annual cost of the space the utilization data suggests is not needed.
• Potential space savings per year. The same conclusion expressed in square meters.
• Potential CO₂ reduction per year. The emissions associated with that space.

Where a summary card covers only part of the selection, the card says so, for example based on 4 out of 5 offices. Read that note before quoting the number, because a total that silently excludes an office is worse than no total at all.

A card with nothing entered behind it shows Setup required with a link to the office settings. A card can also read zero where the office data is present but incomplete, which is not the same thing and is worth checking against the office card before concluding that there is nothing to save.

How the savings figures are calculated

The three figures build on each other, so an error in the first one carries through to all three. They are also calculated separately for desks and for meeting rooms, from the number of assets of each type and their own average utilization, rather than from one combined figure.

Potential space savings start from how much floor area the assets occupy. Because that is not measured, it is estimated per asset:

• Desks. 8 m² per desk.
• Meeting rooms. 4 m² per unit of room capacity, so a six-seat room represents 24 m².

For each asset type, average utilization is compared with the lower bound of the utilization target. Capacity that is surplus to reaching that target is then converted into square metres using the figures above.

Potential savings are the resulting space savings multiplied by the office’s cost per square metre.

Potential CO₂ reduction is the space savings multiplied by 75 kg CO₂ per square metre per year. That is the CIBSE TM46 benchmark for a general office, the most established figure in Europe and the one most used in sustainability reporting, which makes it the safer choice for a number that may end up in a published report.

Three things therefore move all three figures: the utilization target, the asset inventory and the utilization figures themselves. Changing the average utilization target changes what counts as surplus; an asset that no longer exists still contributes its square metres to the calculation; and anything that distorts utilization, such as thin sensor coverage, distorts the savings by the same margin.

Because desks and rooms are calculated separately, the asset type filter decides what the savings cover. A selection limited to desks reports the savings available from desks alone.

Why two dashboards can show different numbers

Before concluding that a figure is wrong, check these four things. In practice one of them explains almost every discrepancy.

1. The selection. One page may cover a whole office while the other is narrowed to a floor or a zone.
2. The period. A 30-day and a 90-day figure will differ, and so will their comparisons against the preceding period.
3. The asset type. A figure covering desks and rooms together is an average of two different things.
4. The days and hours. Including weekends, or measuring across the full day rather than working hours, lowers every utilization figure.

The selection summary under the filter bar shows all four at once. See Filters, targets and exports.