There is a familiar pattern to the working day. Mornings are productive, and then somewhere after lunch a meeting room starts to feel heavy. People blame the sandwich, the time of day, or the meeting itself. Frequently the actual cause is measurable, invisible and cheap to fix: the carbon dioxide the occupants have breathed out has nowhere to go.
This is one of the few areas where a twenty-euro sensor tells you something genuinely useful about how a building affects the people in it.
What CO2 tells you
Carbon dioxide is not itself the problem at the concentrations found indoors. It matters because it is an excellent proxy: people exhale it constantly, so the CO2 level in a room reflects how many people are in it, how long they have been there, and crucially how much fresh air is arriving. A high reading means the air you are breathing has largely been breathed already.
Concentration is measured in parts per million, written ppm. Outdoor air currently sits at roughly 420 ppm, and that is the floor — no amount of ventilation gets a room below the air outside it.
- 400–600 ppm — effectively outdoor air. Well-ventilated or nearly empty.
- 600–800 ppm — a comfortable, adequately ventilated occupied room.
- 800–1,000 ppm — acceptable but heading the wrong way. Commonly used as a design target.
- 1,000–1,400 ppm — ventilation is not keeping up. This is where people begin reporting stuffiness and drowsiness.
- Above 1,400 ppm — poor ventilation. Common in small sealed meeting rooms within half an hour.
Research on cognitive effects exists and generally points the same way — decision-making and concentration measure worse as CO2 rises — but the size of the effect varies considerably between studies, and we are not going to quote a single dramatic figure as though it were settled. The safer and still useful claim is this: at above roughly 1,000 ppm you have evidence that a room is under-ventilated, and under-ventilation is worth fixing regardless of how you feel about the cognitive literature.
Why meeting rooms are the worst room in the building
Three things combine. Meeting rooms are small, so there is little air volume per person. They are frequently the most sealed spaces in an office, sometimes with no openable window at all. And they go from empty to crowded in seconds, which no ventilation system driven by a timer can anticipate.
An open-plan floor with thirty people has a large volume and usually some mechanical ventilation, so it rises slowly. A four-metre-square room with six people and a closed door can pass 1,200 ppm within twenty minutes. That is the room where the two o'clock meeting happens.
Measuring it properly
Get a real sensor
Insist on an NDIR sensor — non-dispersive infrared, which measures CO2 directly by how it absorbs infrared light. Cheaper devices sometimes advertise “eCO2” or “equivalent CO2”, which is estimated from a volatile organic compound reading rather than measured. Those estimates respond to perfume, cleaning spray and hand sanitiser, and will mislead you.
Put it in the right place
Mount at seated head height, roughly 1.1 to 1.5 metres, on an internal wall. Keep it away from the door, from windows, from ventilation grilles and from directly behind where someone sits — a sensor a foot from a person's face reads their breath, not the room. Avoid direct sunlight, which affects the temperature reading and can affect the CO2 electronics.
Let it calibrate
Most affordable NDIR sensors use automatic baseline correction: they assume that at some point over the past week the room reached outdoor levels, and recalibrate to that. This works well in a normal office that empties overnight. It fails in a room that is never empty, where the sensor slowly convinces itself that stale air is fresh. If you monitor a space with continuous occupancy, choose a sensor that allows manual calibration.
Turning readings into action
A dashboard nobody looks at changes nothing. The value comes from linking the reading to something that alters the air.
If you have mechanical ventilation
This is the ideal case. Increase the ventilation rate when CO2 passes a threshold and reduce it when the room recovers. Use two thresholds rather than one — boost above 900 ppm, return to normal below 700 — so the fan does not cycle noisily every few minutes around a single set point. That gap is called hysteresis, and leaving it out is the most common reason people find ventilation automation annoying.
If you have openable windows
Window actuators can be driven by the same logic, with two extra conditions: do not open when it is raining, and do not open when the outdoor temperature makes it unbearable. A simpler and often better version is to have the room panel or a light indicator turn amber above 1,000 ppm, prompting whoever is nearest to open the window. Low-tech, and effective, because the request arrives at the moment it can be acted on.
If you have neither
Then the honest answer is that the automation cannot fix the physics, and the data becomes an argument rather than a control input. That is still valuable. A month of graphs showing a room routinely above 1,400 ppm is a far better case for a ventilation upgrade than anyone's opinion about stuffiness, and it is the kind of evidence that gets budget approved.
What else is worth measuring
CO2 is the highest-value single measurement in an office, but three others earn their place.
Temperature, per room rather than per floor. Complaints about temperature are the most common facilities complaint in any building, and they are usually about a specific corner that the central sensor cannot see.
Relative humidity, which should sit roughly between 40 and 60 per cent. Below 40 dries out eyes and airways in air-conditioned offices; above 60 risks mould, particularly on cold external walls.
Particulates (PM2.5), if your building is on a busy road or near construction. This tells you whether opening the window is genuinely an improvement, which in some city-centre locations it is not.
What we would not bother with in a normal office is a cheap total-VOC sensor. The readings are hard to interpret, drift over time, and rarely lead to a decision anyone acts on.
Making the case internally
Air quality has an unusual advantage as a facilities project: the argument is about people rather than pennies, and it is easy to demonstrate. Put one sensor in the worst meeting room, log a fortnight, and show the graph at the next management meeting. The line that climbs to 1,500 ppm every afternoon and drops the moment the door opens tends to end the discussion quickly.
It also pairs neatly with work you may already be planning. If you are fitting presence sensors for meeting room release, several devices measure presence and CO2 in the same unit, so the marginal cost of the air data is close to nothing.
Where to start
- Buy one good NDIR sensor rather than six cheap eCO2 ones, and put it in the meeting room people complain about most.
- Log for two weeks without changing anything, so you have a baseline and can see the daily shape.
- Fix the worst room with whatever you have: ventilation boost, window actuator, or simply a visible indicator and a request to open the door between meetings.
- Re-measure and confirm the intervention worked, which is the step most often skipped.
- Then expand, one sensor per enclosed room, and drive ventilation from demand rather than the clock.
It is a small, cheap intervention with an unusually direct link to how people feel at work. Of everything in this category, it is the one we would do first.