Occupancy analytics, honestly
All notes
Everything here, grouped by subject.
No product names. No accuracy percentages. Nothing here is legal advice.
For a public, independent reference related to “All notes”, consult the CIBSE Knowledge Portal. Its principles provide a useful check on scope, terminology, governance and the claims made during procurement or review.
What this covers
From the first question to the decision nobody can defend
Eight sections, in the order a programme actually runs: deciding whether to measure at all, choosing a method, deploying it, reading the output, acting on it, and the people and obligations around all of that.
What it measures
Everything above the raw detection is inference, and the inference is where the decisions get made.
6 notes →Counting methods
Six ways to count, each with different blind spots. Match the method to the question and deploy no more.
7 notes →Deploying it
Errors found later cannot be corrected in data already collected, which makes the first month decisive.
7 notes →Reading the data
Never publish average without peak. The two drive opposite decisions and both are usually true.
7 notes →Acting on it
Low utilisation has two readings: too much space, or the wrong space. The second is more often correct.
6 notes →The people in the building
The line is crossed gradually and by request. Each individual request is reasonable and each one granted changes what the system is.
7 notes →Obligations
Aggregating at the device means the detailed data never exists, which removes most of the obligations with it.
5 notes →Reference
The end state as a description, the twelve failures, and the order to do things in.
5 notes →If you are starting
Three things to do before anybody buys a sensor
Name the decision
What will this data inform, who makes that call, and what reading would change it. Write the threshold before the data arrives.
Write the limits
No individual data, no team-level reporting, no feed into attendance or performance. Stated at the start, limits are believed; added later, they are not.
Check what you already hold
Badge records and booking data answer more than people expect, and observation with a clipboard answers the rest. If they settle it, the programme is unnecessary.
All 50 notes
All fifty notes, by subject
- What Occupancy Analytics Actually Measures
- Presence Is Not Value
- The Questions Worth Asking First
- Utilisation, Occupancy, Capacity: Three Different Numbers
- What a Good Measurement Programme Looks Like
- When Not to Measure At All
- Badge and Access Data
- Desk and Room Booking Systems
- Passive Infrared Sensors
- Thermal and Depth Sensors
- Network and Device Counting
- Cameras and Vision Counting
- Which Method Answers Which Question
- Sensor Placement That Works
- Calibration and the First Month
- Coverage Gaps and What They Hide
- Power, Network and Maintenance
- Retrofit Versus New Build
- Pilots That Tell You Something
- Vendor Claims and How to Test Them
- Baselines Before Conclusions
- Peak, Average and the Gap Between Them
- Seasonality in Building Data
- Meeting Rooms: the Numbers Everybody Misreads
- Desk Utilisation and the Attendance Question
- Comparing Floors, Buildings and Sites
- When the Data Contradicts What People Say
- Reducing Space: the Decision and Its Evidence
- Redesigning Rather Than Shrinking
- Booking Policy Changes
- Cleaning, Catering and Facilities Scheduling
- Energy and HVAC Driven by Occupancy
- What the Data Cannot Justify
- Telling Staff What Is Measured
- The Line Between Space Data and Surveillance
- Anonymity, Aggregation and Small Numbers
- Works Councils, Unions and Consultation
- When Occupancy Data Becomes Attendance Monitoring
- Gaming the Sensors
- Accessibility and Who Gets Counted Wrong
The short version
Measure less than you are offered
Calibrate before anybody quotes a figure, report peak alongside average, and keep it about the space rather than the people in it.