IoT in cleaning: what the sensors actually solve, and what you save
"IoT in cleaning" sounds like something that happens at a trade fair. In practice it is three fairly down-to-earth questions: Is the dispenser empty? Has anyone used the room? Is the air bad? If the sensors answer them, cleaning can be done where it is needed, when it is needed, instead of on a fixed route regardless of use.
This article is not about the technology, but about what it solves and what it saves. We use CleanFlow as the example, because we have connected three sensor vendors and an open API, and have seen which rules actually get used.
The problem with fixed frequency
An ordinary cleaning plan says "toilets three times a day, offices weekly". The frequency is set from a guess about use, and it applies every day, no matter what. That produces two errors at once:
- Too much cleaning where nobody has been. The meeting room that stood empty yesterday gets cleaned anyway. The office wing with everyone working from home on Friday too.
- Too little where the pressure is high. The toilets by the entrance on an event day run out of paper at eleven, and the next round is at two. The complaint comes at twelve.
The sensors solve both. They do not say "clean more often". They say "clean here, now", and "skip there, today".
Four data sources, one rule engine
In CleanFlow each data source is an integration on the building, and all of them end in the same thing: a task to the right team, in the same app as the rest of the plan.
1. Dispensers: Tork Vision
Tork dispensers for paper and soap report fill level and status: green, yellow, red, empty. Visitor counters at the entrance count how many have passed. In CleanFlow you see the dispensers on the floor plan, with status and fill level on hover.
The rules under "Task rules" say what should happen: if the paper dispenser goes red, "Refill paper" is created with high priority for the room team. If the visitor counter passes the threshold you set, cleaning of the toilets is created. The rules have a cut-off time: an alert after 3 pm becomes a task for the next working day, not tonight, and tasks that land on a weekend move to Monday.
2. Motion, occupancy and doors: Disruptive Technologies
Disruptive sensors are small and battery-powered, and attach to a desk, a door, a wall or a ceiling. They report whether a desk is in use, whether there is motion in a room, whether a door has been opened, whether someone has pressed a button, plus temperature and humidity. In CleanFlow each sensor is linked to a room on the floor plan.
This is where the two rules you will most often want live:
- Clean when the room is empty. The desk sensor reports "not in use", and the task is created right away. The cleaner comes when nobody is sitting there.
- Clean after x visits. The motion sensor or door sensor counts. After 50 movements, or 20 door openings, cleaning of the zone is created. The zone with little traffic gets no task that day.
On top of that: clean after extended use (a desk occupied for eight hours), alert when a door has been open too long, and a service request from a push button on the wall that the building's users can press themselves.
3. Air quality: Airthings
Airthings monitors give CO₂, particulate matter, volatile organic compounds, humidity, temperature and radon, and calculate person-hours, meaning how much the room has actually been used since last time. The rules in CleanFlow can trigger on "rating is poor", on a value above or below a threshold, sustained for a number of minutes, or on person-hours since the last cleaning. A meeting room with 40 person-hours since Monday gets cleaned on Wednesday. The one that stood empty waits.
4. Everything else: API and webhooks
Not all data comes from sensors. The cinema knows when the film ends. The conference centre knows when the meeting room is booked out. The gym knows when the class is over. CleanFlow has an open REST API where other systems can create tasks: when the booking system sees the auditorium is empty, it sends a task to the room team, and the auditorium is cleaned before the next screening. The same way, deviations and orders can be created, and sensor devices, events and rules read out.
Webhooks go the other way: CleanFlow notifies your system when a task is created, reassigned or completed, when a deviation is reported or changes status, and when a sensor has triggered an alert. Cleaning robots from Gausium check in their runs the same way, as completed tasks in the plan.
What it saves: a worked example
We do not publish a percentage, because it depends entirely on the building. But the calculation is easy to do yourself. Take an office building with 20 toilet groups cleaned three times a day, 15 minutes per round:
Item | Calculation | Hours per year |
|---|---|---|
Fixed frequency today | 20 groups × 3 rounds × 15 min × 260 days | 3,900 |
Assume 40 percent of days have low traffic, where one round can be dropped | 20 × 1 × 15 min × 104 days | 520 |
Hours left with needs-based cleaning |
| 3,380 |
With a true hourly cost of NOK 334, as in our cleaning cost estimate, 520 hours is roughly NOK 174,000 a year. Replace the assumptions with your own numbers. The share of low-traffic days is exactly what you do not know today, and exactly what the sensors measure in the first weeks.
Then there is what the calculation leaves out:
- Fewer complaints. The dispenser is refilled before it is empty, not after someone has called.
- Better quality where it counts. The hours saved on empty rooms can be spent on the entrance on a rainy day.
- Documentation. Every task has a reason: "paper empty at 10:42", "30 visits since last time". That is the answer when the client asks why.
Three things to know before you start
- The sensors do not replace the plan. The base frequency stays. The sensors add tasks where they are needed and let you lower the frequency where they are not. Start with the base plan, add the rules afterwards.
- Start with one zone. The toilet group with the most traffic, or the meeting room wing. Measure for four weeks before you change the frequency.
- The rules need a cut-off time. Without it, the evening shift gets a task at 10 pm. Set the time when today's alerts become tomorrow's tasks, and let weekend alerts wait until Monday.
Frequently asked questions
What is needs-based cleaning? Cleaning triggered by actual use, measured with sensors or taken from other systems, instead of only by a fixed frequency. The base plan stays, but the tasks are adjusted to need.
Do we have to buy new dispensers? Tork Vision requires Tork dispensers with sensors. Disruptive and Airthings sensors can be placed in existing rooms without changing equipment.
How does the cleaner see the sensor tasks? As ordinary tasks in the app, marked "Sensor alert", on the floor plan and in today's list. No separate app, no separate dashboard.
Can we connect our own system? Yes, via the API. A booking system, an access control system or a visitor counter from another vendor can create tasks directly, and receive notifications back via webhooks.
Want to see how sensor data becomes tasks in your building? Read more about IoT cleaning in CleanFlow, or try it free for 14 days.