People counting: WiFi vs camera vs door sensor

To choose between WiFi, camera, and door sensor people counting, assess your goals. WiFi analytics, integrated via Cisco Meraki in 80,000+ venues, uses the IEEE 802.1X standard for venue-wide tracking, while cameras suit single zones and door sensors measure entry threshold totals.

Which people counting technology should you choose?

Start with the question you need answered, not the sensor. Most venue operations teams ask one of three questions.

• How many people came in? Door sensors answer this best. Infrared (IR) beam or 3D stereo sensors mounted over each entrance count people as they cross the threshold.
• How many people are in this zone right now? Cameras answer this best. Computer vision software counts people inside a camera's field of view, such as a queue, a gate or a seating block.
• Who came back, and how long did they stay? WiFi answers this best. Your access points already see devices across the whole venue. WiFi logins give you a stable identity for each returning visitor.

No single technology answers all three questions well. That is why many estates combine two of them. A retail chain might pair door counters for conversion with WiFi for dwell time and repeat visits.

If WiFi is in the mix, it can sit on the network you already run. Purple works as a cloud overlay on Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme and Fortinet access points. Purple runs in 80,000+ live venues and recorded 440 million logins in 2024 (Purple's own data).

Where do WiFi, camera and door sensor counting genuinely differ?

This people counter comparison sets the three technologies against the axes procurement teams usually score.

Criterion	WiFi sensing	Computer vision camera	IR or 3D door sensor
What it counts	Devices seen by access points, plus people who log in	People inside the camera's field of view	People crossing an entrance line
Coverage area	Whole venue, wherever access points reach	One zone per camera	One doorway per sensor
Headcount precision	Lowest: counts devices, not people	Highest inside a defined, well-lit zone	Highest at a single entrance
Repeat visits	Yes, through authenticated logins	Only with facial recognition, which is biometric data	No
Dwell time	Yes, across the whole venue	Inside each camera's zone only	No
New hardware	None if your access points are supported	Cameras, mounts, PoE cabling, analytics compute	One sensor per entrance, plus mounts and power
Personal data involved	Device identifiers and opt-in profile data	Images of identifiable people	None for beam sensors; images processed for counting on 3D models
Main compliance task	Privacy notice and consent at login	DPIA and signage under ICO video surveillance guidance	Privacy check for 3D models; little for beam sensors
Best for	Dwell, repeat visits, busiest times, venue-wide trends	Queues, live occupancy, capacity limits	Entrance totals and conversion rates

PoE (Power over Ethernet) supplies power and data over a single network cable. A DPIA (Data Protection Impact Assessment) is the risk assessment GDPR requires for high-risk processing.

How accurate is WiFi people counting?

WiFi counts devices, not people. Some visitors carry no phone or keep WiFi switched off. Others carry a phone and a laptop. A raw device count is therefore an index of activity, not an exact headcount.

Two further effects distort device counts. First, staff devices appear alongside visitor devices unless you separate them. Second, phones now hide their real hardware address. Apple introduced private addresses in iOS 14, and Google made randomised MAC addresses the default in Android 10. A returning phone can look like a new device to a system that relies on detection alone.

Authenticated logins fix the identity problem. When a guest logs in to your Guest WiFi, the visit is tied to a profile rather than a hardware address. That gives you a firmer basis for new and returning visitor figures.

To turn device data into a headcount estimate, calibrate it. Run a manual tally or a door count over a sample period. Calculate the ratio of WiFi visitors to real entries, then apply that ratio to later periods.

How accurate is camera-based footfall?

Computer vision footfall accuracy depends mostly on where and how you mount the camera. A camera looking straight down separates people well. An angled camera loses people standing behind others, an effect called occlusion.

Lighting, glare, crowd density and children also affect results. Vendors publish accuracy figures, but they measure them under their own test conditions. Ask for figures from conditions like yours, then validate with a manual count before you sign.

Existing CCTV is rarely a shortcut. Security cameras are angled to identify faces, not to count heads from above. Expect to add dedicated counting cameras even where CCTV is already installed.

How accurate are door counters?

A single IR beam registers one count each time the beam breaks. Two people walking side by side can register as one. Someone pausing in the doorway can register twice. Dual-beam sensors add direction, so you can separate entries from exits.

3D stereo sensors look down from above the door. They separate individuals in a group and read direction more reliably than beams. Many models also filter by height, so a pushchair or small child is not counted as an adult.

Every door counter has the same blind spot. It counts staff walking in and out unless you filter them, and it knows nothing about what happens past the threshold.

When is WiFi the right call for people counting?

Choose WiFi when coverage and behaviour matter more than an exact headcount. It suits four situations.

You need the whole venue, not just the door

Multi-floor hotels, campuses, stations and shopping centres have too many entrances and zones to sensor economically. Access points already cover those spaces. Passenger hubs make the point clearly: see how Purple works for Trains and stations.

You need repeat visits and frequency

Purple's behavioural analytics report the number of unique visitors in a period and the share who are new or returning. A returning visitor is one who has attended your venue before.

New and returning percentages can add up to more than 100%. A visitor whose first visit and return both fall inside the report period counts in both groups. Purple's support article walks through a five-visitor example that produces 80% new and 60% returning.

You need engagement and busiest times

The reports also show dwell time, engagement rates, marketing opt-in and the most popular days and times. Engagement bands sort visitors into not engaged, partially engaged and fully engaged by dwell time. You can set those thresholds for each venue, and system defaults apply if you do not.

With a Presence licence, reports also include unauthenticated visits. These are devices detected nearby that did not log in, shown alongside authenticated visitors who did. See the behavioural analytics support article for the report definitions and settings.

You want no new hardware

Purple is hardware-agnostic. WiFi counting runs on the access points you own, so there is no cabling, drilling or downtime.

To keep staff devices out of visitor counts, put staff on a separate network. IEEE 802.1X is the standard for authenticating each device before it joins a network. See How to Configure WPA2-Enterprise on Common Access Point Platforms (Cisco, Aruba, Ubiquiti). A VLAN (virtual LAN) then keeps staff and guest traffic apart, as described in How to Configure NAC Policies for VLAN Steering in Cisco Meraki.

Worked scenario: a hotel restaurant open to non-residents

Figures in this scenario are illustrative.

Situation. A 200-room hotel runs a restaurant and bar open to the public. The general manager wants to know whether local diners come back, and when. A door counter on the street entrance could not separate diners from residents passing through.

What was done. The hotel added a Guest WiFi login with conscious-choice opt-ins in the restaurant and bar. The team set engagement thresholds to fit a typical meal. They then reviewed the behavioural reports monthly.

Outcome. In one month, the reports showed 1,200 unique visitors, of whom 360 were returning, or 30%. Fully engaged visitors clustered on Thursday and Friday evenings. The hotel moved its local loyalty offer to Tuesday and Wednesday, then measured the change in returning visitors on those nights. More on hotel deployments at Hotels.

When is computer vision footfall the right call?

Choose cameras when you need a precise live count inside a small, defined area.

• Queues. Registration desks, security lanes and ticket counters, where wait time drives staffing.
• Live occupancy. Halls, galleries or terraces with a fire capacity limit.
• Gates and turnstiles. Event entry points where you need direction and a real-time total.

The trade-off is privacy and cost. Cameras record images of identifiable people, which is personal data under UK GDPR. The Information Commissioner's Office (ICO) guidance on video surveillance expects you to assess the need, usually through a DPIA, and to tell people with clear signage.

Facial recognition raises the bar further. Biometric data used to identify a person is special category data under Article 9 of UK GDPR. You need an Article 9 condition before you process it. That is why few venues use cameras to measure repeat visits.

In Healthcare settings, waiting areas are often a good fit for counting cameras. Occupancy matters there, while identity does not. Choose models that count without storing faces.

Worked scenario: a conference centre main hall

Figures in this scenario are illustrative.

Situation. A conference centre has a 3,000-capacity main hall and a single registration area. Organisers need a live occupancy figure for the fire capacity limit. They also want to know how long attendees wait to register.

What was done. The venue fitted overhead counting cameras above each of the four hall doors and over the registration queue. It configured the analytics to count only, with no images retained. WiFi analytics stayed in place across breakout rooms and catering areas for dwell time.

Outcome. Stewards saw live occupancy on a screen and closed the hall doors at 2,900 attendees, inside the limit. Queue data showed waits peaked in the 20 minutes before opening keynotes. The organiser added two registration desks for that window at the next event.

When do door counters beat WiFi?

The door counter vs WiFi question comes down to precision at the threshold. Door counters win in four situations.

• Conversion rate. Retail conversion is transactions divided by entries. You need a precise entry figure, and a door counter gives you one.
• Reported visitor numbers. Libraries, museums and public buildings often report visitor numbers to funders or councils. A threshold count is simple to explain and audit.
• Few entrances. A shop with one door needs one sensor. Cost stays low.
• Minimal privacy work. Beam sensors capture no personal data at all.

Door counters cannot tell you dwell time, which zones people visit, or whether they come back. If you need those answers, pair the door counter with WiFi rather than choosing between them.

Worked scenario: a 40-store retail chain

Figures in this scenario are illustrative.

Situation. A fashion retailer with 40 stores wants to know why sales vary between stores of similar size. It already runs managed WiFi in every store.

What was done. The retailer fitted 3D door counters at each store entrance to count entries. It added Purple WiFi Analytics on its existing access points to measure dwell time and returning shoppers. Store managers received a weekly report combining both data sets.

Outcome. Store A recorded 12,000 entries and 2,400 transactions, a 20% conversion rate. Store B recorded 9,000 entries and 2,700 transactions, a 30% conversion rate. WiFi data showed shoppers in Store A had shorter dwell times and fewer returning visits. The retailer reviewed Store A's layout and staffing at peak times first. More on retail deployments at Retail.

Which people counting method is cheapest to run?

WiFi is cheapest where you already run enterprise WiFi on supported access points. Door sensors are cheapest for a small venue with one entrance and no managed WiFi. Cameras carry the most cost lines, because each one needs installation, processing and privacy work.

Cost line	WiFi sensing	Computer vision camera	IR or 3D door sensor
Hardware	None on supported access points	One or more cameras per zone	One sensor per entrance
Installation	Configuration only, no site works	Mounting, PoE cabling, network ports	Mounting above each door, power
Software	Analytics licence per venue	Video analytics licence, often per camera	Counting platform licence, often per sensor
Processing and storage	Cloud-hosted by Purple	On-camera, on-site server or cloud compute	On-sensor or vendor cloud
Upkeep	Access point firmware you already manage	Lens cleaning, re-aiming after refits	Recalibration after door or layout changes
Compliance effort	Privacy notice and consent at login	DPIA, signage, retention policy	Little for beam; privacy check for 3D
Cost grows with	Number of venues	Number of zones	Number of entrances

Two hidden costs catch procurement teams out. The first is refits: every new door or shop-floor layout means re-mounting and recalibrating door sensors and cameras. The second is integration: each system has its own dashboard unless you feed data into one reporting tool.

WiFi avoids both. Access points already cover the space and survive a refit. Purple integrates with CRM and marketing platforms, so visitor data can sit alongside your other reporting.

How do you decide for your estate?

Use this decision matrix to map your situation to a starting recommendation.

Your situation	Recommended approach
You need dwell time and repeat visits across a large venue	WiFi analytics on your existing access points
You need a precise entrance count for conversion	Door counters, paired with WiFi for behaviour
You need live occupancy in a hall, queue or gate	Overhead counting cameras in those zones only
You run many small sites with managed WiFi	WiFi analytics, adding door counters at flagship sites
You run a single small shop with no managed WiFi	One door counter
You report visitor numbers to a funder or council	Door counters at every public entrance
You must avoid any personal data in counting	IR beam door counters

Five steps to a decision

1. Write down the decisions the data will drive. Staffing, layout, marketing and capacity each need a different measure.
2. Audit what you already own. List your access point vendor, models and coverage. Check for existing door counters or CCTV.
3. Pilot at two or three sites. Run a manual count alongside each technology for at least one busy week.
4. Calibrate. Compare each technology against the manual count and record the ratio for each site.
5. Complete the privacy work before rollout. Update privacy notices and complete a DPIA where cameras are involved.

Questions to ask every vendor

• What accuracy did you measure, under what conditions, and will you prove it on our site?
• What hardware does each venue need, and who installs it?
• Where is data processed and stored, and for how long?
• Can the data feed our existing reporting and CRM tools?
• What happens to counts after a refit or a new entrance?

Purple's answers to the data questions rest on its certifications. Purple is ISO 27001, GDPR, CCPA, Cyber Essentials and B Corp certified.

Frequently asked questions

Is WiFi or camera better for people counting?

Cameras are better for a precise count inside one defined zone, such as a queue or a hall entrance. WiFi is better for covering a whole venue and measuring dwell time and repeat visits. Cameras need new hardware and a DPIA because they capture images of identifiable people. WiFi runs on access points you already own. Many venues use cameras for a few high-value zones and WiFi everywhere else.

Do door counters beat WiFi for counting visitors?

Yes, for a precise entrance total, door counters beat WiFi. They count people crossing a threshold, while WiFi counts devices, and not every visitor carries one. Door counters cannot measure dwell time, zones visited or repeat visits, which WiFi can. If you need conversion rates and behaviour, combine both. The door counter supplies entries and WiFi supplies how long shoppers stayed and whether they returned.

Which people counting method is cheapest?

WiFi is cheapest if you already run enterprise WiFi on supported access points, because it needs no new hardware or site works. For a single small shop with no managed WiFi, one door counter is usually the lowest cost. Cameras carry the most cost lines: hardware, cabling, analytics licences, processing and privacy work. Cost also scales differently: WiFi with venues, cameras with zones, and door counters with entrances.

Will WiFi people counting work with the access points we already own?

Yes, Purple runs as a cloud overlay on your existing access points from Cisco Meraki, HPE Aruba, Ruckus, Juniper Mist, Ubiquiti UniFi, Cambium, Extreme and Fortinet. There is no rip and replace. You configure your network to work with Purple, then reports populate as guests log in. Check that your specific models and firmware are supported before you plan a rollout across the estate.

Is WiFi people counting GDPR compliant?

Yes, WiFi counting can be GDPR compliant when you collect data transparently and with a lawful basis. Purple uses conscious-choice opt-ins at login, so guests see what they agree to. You still need an accurate privacy notice covering device data and analytics. Purple is ISO 27001, GDPR, CCPA, Cyber Essentials and B Corp certified. Unlike cameras, WiFi counting involves no images or biometric data.

How does MAC address randomisation affect WiFi counts?

It makes device detection alone less reliable for repeat visits. Since iOS 14 and Android 10, phones present a randomised MAC address to each network by default. A returning phone can therefore look new to a system that only detects devices. Authenticated logins solve this, because visits are tied to the guest's profile rather than a hardware address. Use authenticated data for repeat-visit reporting.

Can we combine door counters and WiFi analytics?

Yes, and for retail it is often the strongest setup. Door counters give you a precise entry figure for conversion rates. WiFi analytics adds dwell time, busiest days and times, and new versus returning shoppers. Comparing the two also lets you calibrate WiFi data at each site. Fit door counters at flagship stores and rely on WiFi across smaller sites to keep costs proportionate.

How long does each option take to deploy?

WiFi is fastest to deploy where your access points are supported, because it involves configuration rather than site works. Door counters need a sensor fitted and powered above each entrance, then calibrated. Cameras take longest: mounting, PoE cabling, analytics setup, signage and a DPIA before go-live. Whichever you choose, allow a pilot week with a manual count to confirm accuracy before rolling out across the estate.