How to Measure Foot Traffic: Technologies and Methods Explained

If you run an outdoor advertising campaign and cannot answer the question "how many people actually saw it," you are flying blind. The same goes for retail stores, event venues, shopping malls, and urban planners. Foot traffic data is the backbone of physical-world measurement, and the technology behind it has changed dramatically in the last five years.

According to Grand View Research, the global people counting system market reached $1.3 billion in 2025 and is projected to grow at a 12.3% CAGR through 2030. That growth is driven by a simple truth: businesses that measure foot traffic outperform those that do not. Retailers with accurate foot traffic analytics see 15-20% improvements in staffing efficiency and a 10-15% lift in conversion rates, per a 2025 RetailNext study.

This guide breaks down the five primary methods for measuring foot traffic, compares them across accuracy, cost, privacy, and scalability, and explains why the industry is converging on a particular approach: on-device AI processing.

1. WiFi Sensing and Probe Request Tracking

WiFi sensing works by detecting the probe requests that smartphones continuously broadcast as they search for known networks. Every phone with WiFi enabled sends out these signals, and WiFi sensing hardware captures them to estimate how many devices (and by proxy, people) are in a given area.

How it works: Access points or dedicated sensors are placed in a venue. When a phone enters range, the sensor detects its probe request, logs a randomized MAC address, and records the timestamp and signal strength. By triangulating across multiple sensors, you can estimate not just count but also dwell time and movement patterns.

Accuracy: WiFi sensing used to be reasonably accurate (within 15-20% of actual foot traffic) before Apple and Google introduced MAC address randomization in 2014 and tightened it further in subsequent OS updates. Today, with iOS 18 and Android 15 fully randomizing MAC addresses on every probe request, raw WiFi counting overstates unique visitors by 30-60%. Most vendors apply correction algorithms, but accuracy still hovers around 70-80% in best-case scenarios.

Cost: Hardware ranges from $200-$500 per sensor, with software subscriptions of $50-$200 per location per month. Vendors include Meraki (Cisco), Purple, and RetailNext.

Privacy considerations: This is where WiFi sensing gets complicated. Even with MAC randomization, some regulators view passive device scanning as personal data collection. The European Data Protection Board issued guidance in 2023 stating that WiFi tracking in retail requires a legal basis under GDPR, which effectively means consent or legitimate interest with a balancing test. Several European retailers have faced fines for undisclosed WiFi tracking.

Best for: Indoor venue analytics where approximate counts and dwell times are sufficient. Shopping malls, airports, and large retail stores that have existing WiFi infrastructure.

2. Computer Vision and AI-Powered Counting

Computer vision uses cameras and AI models to detect, count, and classify people in a scene. This is the fastest-growing segment of the foot traffic measurement market and the approach that delivers the highest accuracy by a wide margin.

How it works: A camera captures video. An AI model (typically based on YOLO, SSD, or a custom architecture) processes each frame to detect human figures, draw bounding boxes, and assign tracking IDs. Advanced systems can estimate age range, gender, and even attention direction (whether someone is looking at a screen or walking past). The key distinction in 2026 is where the processing happens: cloud-based systems stream video to remote servers, while edge-based systems run the AI models directly on the device.

Accuracy: State-of-the-art computer vision systems achieve 95-98% counting accuracy in controlled conditions (single entry point, good lighting). In more challenging conditions like outdoor crowds, accuracy typically ranges from 88-94%. The technology has improved dramatically — five years ago, those numbers were 10-15 points lower.

Cost: Cloud-based solutions from vendors like Sightcorp and Quividi cost $100-$300 per camera per month. Edge-based solutions carry higher upfront hardware costs ($500-$2,000 per unit for the processing hardware) but lower or zero ongoing cloud fees.

Privacy considerations: This depends entirely on the implementation. Cloud-based systems that stream video to remote servers face significant privacy concerns and regulatory scrutiny. Edge-based systems that process video on-device and only transmit anonymized aggregate data (counts, demographic distributions, dwell times) are far more privacy-friendly. The distinction matters enormously under GDPR, CCPA, and emerging regulations. For more on this topic, see our deep dive on privacy-first analytics.

Best for: Any application where accurate counting, demographic insights, and attention measurement are critical. Outdoor advertising measurement, retail analytics, smart city planning, and event management.

3. Beacon and Bluetooth Technology

Bluetooth beacons are small, battery-powered transmitters that broadcast a signal to nearby smartphones. When a user has a compatible app installed (or has Bluetooth scanning enabled), the beacon interaction can be detected and logged.

How it works: Beacons using Bluetooth Low Energy (BLE) are placed in strategic locations. They broadcast a unique identifier. Smartphones within range (typically 1-70 meters, depending on signal strength configuration) can detect this signal through compatible apps or through the OS-level Bluetooth scanner. The interaction is logged, providing data on proximity, dwell time, and repeat visits.

Accuracy: Beacon technology is highly accurate for detecting app users — near 100% for devices with the relevant app installed and Bluetooth enabled. The problem is coverage. Only a fraction of the total foot traffic will have the right app installed. Industry estimates suggest beacons typically capture 5-15% of total foot traffic, which requires heavy extrapolation to estimate true counts.

Cost: Beacons themselves are cheap — $15-$50 per unit from vendors like Estimote, Kontakt.io, and Gimbal. The cost is in the software platform ($200-$1,000 per month) and the app development required to capture beacon signals. Battery replacement is an ongoing maintenance concern, though modern beacons last 3-5 years.

Privacy considerations: Beacon interactions require either an app opt-in or OS-level Bluetooth permissions, which makes this method more consent-friendly than WiFi sensing. However, the ability to track individual devices across locations has raised concerns. Apple's App Tracking Transparency (ATT) framework and Google's Privacy Sandbox have significantly limited what beacon data can be used for.

Best for: Loyalty program integration, indoor navigation, and proximity marketing where you already have an installed app base. Less useful for general foot traffic counting because of the low capture rate.

4. GPS and Mobile Location Data

GPS-based foot traffic measurement aggregates anonymized location data from mobile devices to estimate how many people visited a specific location during a given time period. This is the dominant method used by the advertising industry for footfall attribution.

How it works: Mobile apps that have location permissions collect GPS coordinates from users' phones. Data aggregators like Foursquare (Factual), SafeGraph, Placer.ai, and Unacast collect this data from hundreds of apps, normalize it, and sell foot traffic insights. The data is typically reported at a venue level: "Store X received an estimated 2,400 visits last week."

Accuracy: GPS accuracy varies from 3-15 meters outdoors, making it suitable for distinguishing between nearby stores in a strip mall but unreliable for indoor positioning. The bigger accuracy concern is sampling bias. Location data aggregators typically capture data from 10-20% of the population, then extrapolate. This extrapolation can introduce significant error, particularly in areas with demographics that skew away from the app user base. Studies have shown GPS-based foot traffic estimates can deviate from ground truth by 20-40%.

Cost: Aggregated foot traffic reports from vendors like Placer.ai start at $500-$1,000 per month per location. Enterprise plans covering hundreds of locations can run $10,000-$50,000 per month.

Privacy considerations: GPS data is inherently personal — it tracks where people go. Even when anonymized and aggregated, researchers have repeatedly demonstrated that location data can be re-identified. The FTC has taken enforcement action against several location data brokers, and both Apple and Google have tightened location permissions significantly. Users must now explicitly opt in to "always on" location sharing, and many choose not to. This is shrinking the available data pool.

Best for: Competitive benchmarking (comparing foot traffic across competing locations), trade area analysis, and footfall attribution for advertising campaigns. The OAAA and Geopath use GPS-based data as the foundation for OOH audience measurement.

5. Manual Counting and Clicker Surveys

The oldest method in the book, and still widely used. A person stands at a location and counts pedestrians passing by, either with a handheld clicker or a tally sheet.

How it works: An observer is stationed at a specific point. They count every person who passes within a defined area during a set time period. Some manual counts are done through video review — a person watches recorded footage and tallies pedestrians.

Accuracy: In low-traffic conditions (under 200 people per hour), manual counting is surprisingly accurate — within 2-5% of ground truth. In high-traffic conditions (1,000+ people per hour), accuracy drops to 80-90% due to counter fatigue and the difficulty of tracking multiple pedestrian streams. Manual counts are also limited to the specific times and locations observed, with no continuous coverage.

Cost: Labor-intensive. A professional counting service charges $25-$75 per hour per location. For a meaningful dataset covering multiple locations over multiple days, costs can reach $5,000-$20,000. This makes it impractical for ongoing measurement.

Privacy considerations: None. Manual counting does not collect any personal data. It produces aggregate numbers only.

Best for: Baseline calibration (validating other methods), short-term studies, and locations where technology deployment is impractical. Urban planners and transportation agencies still rely on periodic manual counts as a ground truth benchmark.

Comparison: All Five Methods at a Glance

Here is how each technology stacks up across the dimensions that matter most:

The pattern is clear: computer vision on edge devices offers the best combination of accuracy, privacy, and real-time capability. It is the only method that simultaneously delivers high counting accuracy, demographic insight, attention measurement, and strong privacy protection.

How AdLuxy Measures Foot Traffic: On-Device AI

When we designed the AdLuxy backpack, foot traffic measurement was not an afterthought — it was the core engineering challenge. Our advertisers need verified impression counts. Our ambassadors need proof that their routes are delivering value. And everyone involved needs confidence that the data is accurate and that no one's privacy is being compromised.

Here is how our system works:

Edge processing on NVIDIA Jetson: Every AdLuxy backpack contains an NVIDIA Jetson Orin Nano module running six computer vision models simultaneously. These models handle person detection, crowd density estimation, demographic classification (age range and gender), attention tracking (gaze direction), dwell time measurement, and scene classification (indoor vs. outdoor, day vs. night). All of this happens on the device itself. No video frames are ever transmitted to the cloud.

What gets transmitted: Only anonymized aggregate data leaves the device — numbers like "42 people detected in the last 5 minutes, 60% estimated female, 35% in 25-34 age range, average dwell time 3.2 seconds." These statistics are sent to the AdLuxy cloud dashboard over a compressed, encrypted connection. There is no way to reverse-engineer individual identities from this data.

Accuracy validation: We continuously validate our counting accuracy against manual ground truth counts. Across 1,200+ hours of real-world testing in seven cities, our system achieves 93.7% counting accuracy and 87.2% demographic classification accuracy. These numbers are published in our transparency reports and are available to all advertisers.

The result is that every AdLuxy campaign comes with verified, auditable impression data that meets the standards advertisers expect from digital campaigns — but without the privacy baggage of cookie tracking or mobile location surveillance. For a deeper look at our approach, read our piece on how AI audience detection works.

Choosing the Right Method for Your Use Case

There is no single best method for every situation. The right choice depends on what you are measuring, where, and why:

Retail store analytics: Start with computer vision for entrance counting and dwell time analysis. Supplement with WiFi sensing if you need movement flow data across a large venue. Beacon technology is a strong addition if you have a mobile app with a meaningful install base.

Outdoor advertising measurement: Edge-based computer vision is the gold standard. It provides the impression verification that advertisers demand while respecting privacy regulations. GPS mobile data is a useful supplement for understanding audience composition at a macro level. Our guide to measuring OOH ROI covers this in detail.

Event management: Computer vision for real-time crowd density monitoring (critical for safety). WiFi sensing for understanding attendee flow between zones. Manual counting as a ground truth calibration tool at key entry points.

Urban planning: GPS data for macro-level pedestrian flow analysis across city districts. Computer vision for specific intersection or corridor studies. Manual counts for validation and regulatory compliance.

Walking billboard campaigns: This is where on-device computer vision truly shines. Because the display is mobile, you cannot install fixed infrastructure. The measurement system must travel with the display. AdLuxy's integrated approach — where the same device displaying the ad also counts its viewers — eliminates the gap between display and measurement that plagues other DOOH formats.

The Privacy Imperative: Why Method Matters More Than Ever

The regulatory landscape around foot traffic measurement has tightened dramatically. GDPR enforcement in Europe has resulted in fines for undisclosed WiFi tracking. CCPA and its successor CPRA in California give consumers the right to opt out of location data collection. Brazil's LGPD, India's DPDP Act, and similar laws worldwide are creating a patchwork of requirements that make cloud-based, data-hungry measurement approaches increasingly risky.

The trend is unmistakable: the future of foot traffic measurement belongs to technologies that can deliver accurate, actionable data without collecting personal information. Edge-based computer vision — where raw data never leaves the device and only anonymized aggregates are transmitted — is purpose-built for this reality.

This is not just an ethical position. It is a competitive advantage. Advertisers are increasingly demanding privacy certifications from their measurement partners. Retailers face brand risk if their foot traffic analytics partner is caught in a privacy scandal. And regulations will only get stricter. Building on a privacy-first architecture is building on solid ground. Read more in our analysis of the privacy future of OOH measurement.

The Bottom Line

Foot traffic measurement in 2026 is not a solved problem — but it is a rapidly improving one. WiFi sensing still has a role for indoor flow analysis. GPS data remains valuable for macro-level insights. Beacons serve app-centric use cases well. And manual counting is an irreplaceable ground truth tool.

But the technology that checks all the boxes — accuracy, granularity, real-time capability, demographic insight, and privacy compliance — is on-device computer vision. It is what AdLuxy uses, it is what the leading smart city platforms are deploying, and it is where the industry is heading.

The brands and operators that invest in accurate, privacy-first foot traffic measurement today will have a compounding advantage over those still relying on estimates and extrapolations. The data does not lie, and the technology to collect it responsibly is here.