Request a consultation
Compliance & funding

People counting and the GDPR

Yes, you can count visitors under the GDPR, as long as the counter produces numbers and never a person. Here is which sensor types stay anonymous by design, where cameras and phone signals need extra care, and what to ask before you install a counter.

Yes, you can count visitors under the GDPR, as long as the counter produces numbers and never a person. Most sensors that count at a door or in a hall see heat, light pulses or 3D shapes, not faces. When no image and no device identifier is stored or sent, the result is anonymous data, and anonymous data is outside the scope of the regulation.

That short answer hides a few decisions worth making on purpose. This guide walks through them: what makes counting data personal, which sensor types process what, where cameras and phone signals need extra care, and what to ask a supplier before you mount anything above a door.

Why people counting raises a data protection question

Visitor numbers are useful well beyond a monthly report. A museum staffs its cloakroom by them, a supermarket opens a second checkout, a church decides whether the nave is too full for the next service. To keep that number current through the day, something has to register each person who walks past.

The GDPR applies to personal data: any information that relates to an identified or identifiable person. A camera image of a face is personal data. A phone's hardware address can be too, because it singles out one device and, with it, one person. A plain count of "47 people entered between 10:00 and 10:15" relates to nobody in particular. The question is therefore not whether you count, but how the sensor gets to the number and what it keeps on the way.

Which counting methods process personal data?

The people counting sensors baiond works with fall into a few families. They differ in what they physically capture, which is what matters for data protection.

Method What the sensor captures What it sends Personal data?
Infrared counter Body heat crossing a line In and out counts per door No, never records an image
Thermal imaging sensor Heat shapes, not faces Counts, density No identifiable image
Time-of-flight (ToF) Distance from light pulses Counts and direction No
LiDAR 3D point cloud, moving shapes Counts, flows across an area No faces in the data
Pressure mat Steps on a surface Step counts No
AI camera with on-device analysis A video image, evaluated on the device Numbers only, no footage Image is processed briefly on the device; check the setup
Wi-Fi / Bluetooth sensor Signals from phones Anonymous estimate of nearby devices Device identifiers are processed; check hashing and deletion

The first five never see a person as a person. They are anonymous by design, which is why they are the usual choice for entrances in museums, churches and shops.

Cameras: on-device analysis is the deciding detail

A camera is the right tool where a door counter is not enough: wide entrances, walkways, waiting zones, places where you also want group size or dwell time. It is also the method that needs the most care, because for a fraction of a second it does see faces.

The baiond AI camera evaluates the image on the device itself and sends numbers only. No footage is stored, and no video leaves the camera. That design keeps the image processing short and local, and the platform only ever receives counts.

Even so, a camera that looks at a publicly accessible area is something your data protection officer should assess. Depending on the setup, that can mean an entry in your record of processing activities, a sign at the entrance, and in larger settings a data protection impact assessment. Plan these steps before installation, not after.

Phone signals: a sample, and a second look at identifiers

Wi-Fi and Bluetooth sensors pick up the signals that phones send and turn them into an estimate of how many people are nearby and how they move between zones. They are useful for trends across a large site, but not every visitor carries an active device, so the result is a sample rather than an exact count.

From a data protection view, the signals carry device identifiers. Ask how the sensor handles them: whether they are hashed on the device, how long anything is kept, and whether only aggregated figures reach the platform. For most door and room counting, an infrared, ToF or LiDAR counter answers the same question without touching identifiers at all.

What anonymous counting looks like in practice

Take a historic venue that hosts congresses and balls. Organisers want to know how full each hall is, the team needs an alert before a room reaches its capacity limit, and the house wants to document how guests moved through the building afterwards.

None of that needs to know who someone is. Counters at each door report entries and exits by radio to the baiond Gateway. The Space System knows which door belongs to which hall, so it can turn door counts into the current occupancy per room. That is the setup behind our work at the Hofburg Congress Center Vienna: guests are counted per hall as they come and go, and the team is alerted at the capacity limit.

The same pattern carries over to other places. In a church, the count sits next to CO₂ and humidity readings. In a supermarket, it opens the second checkout before the queue grows. In a city, visitor flows show which streets carry the load on a market day.

Five questions to ask before you install a counter

  1. What does the sensor physically capture? Heat, distance and 3D shapes are anonymous. Images and phone signals need a closer look.
  2. Where is the data processed? On the device is better than in a cloud that receives raw video.
  3. What leaves the device? Ideally numbers only: counts, direction, occupancy.
  4. What is stored, and for how long? Aggregated counts can be kept for trends. Raw images or identifiers should not be kept at all.
  5. Who has signed off? Bring your data protection officer in early, and note the counting in your record of processing activities where required.

Frequently asked questions

Is people counting allowed under the GDPR?

Yes. Counting itself is not the issue. It becomes a data protection matter only when the sensor processes personal data, such as images or phone identifiers. Infrared, ToF, LiDAR and thermal counters, as a rule, do not.

Do I need consent from my visitors?

Not for anonymous counting, because no personal data is processed. Where a camera or phone signals are used, your data protection officer decides which legal basis and which information duties apply.

Do I need a sign at the entrance?

For anonymous counters, usually not. For cameras, a notice is the norm. Clarify it for your specific setup before the sensor goes live.

Can a camera count people without storing video?

Yes. The baiond AI camera analyses the image on the device and sends only the numbers. No footage is saved or transmitted.

Which counter is right for a normal door?

An infrared or ToF counter. Wide entrances and crowds call for thermal imaging or a camera, open squares and trails for LiDAR.

Want to count visitors without collecting personal data? Talk to us about your site.

All insights

Next step

Ready to give your data a voice?

Thirty minutes with the baiond team about your site and your questions. No slides.

Request a consultation
Ask baiond
baiond AI is typing