Measuring Attention: Between Precision and Privacy
Public space is not a law-free zone for data analysis, and yet the measurability of advertising contacts remains the hard currency in the Digital-out-of-Home (DOOH) sector. Anyone investing in visual infrastructure today – whether it is a large-format LED wall with a NovaStar MX40 Pro controller or a network of Samsung The Wall displays – cannot ignore the question of performance measurement. However, the time for experimentation is over. What remains permitted in 2026 is no longer defined by technically feasible gimmicks, but by the strict separation of detection and identification. We are observing a shift away from cloud-based processing towards radical edge computing, where not a single byte of image material leaves the sensor.
The Technological Evolution of Anonymisation
For a long time, audience measurement was considered synonymous with video analytics. Sensors such as the Logi Circle or specialised industrial cameras delivered streams to central servers where algorithms estimated age, gender, and dwell time. According to the current interpretation of the GDPR and in view of upcoming guidelines for 2026, this model is obsolete. The requirement is "Privacy by Design".
Modern systems such as those from Quividi or BlueFox now work on the basis of real-time metadata extraction. This means: The sensor (the camera) captures a signal, the integrated processor immediately converts this into an anonymous vector – for example, "human, read as male, 30-40 years, eye contact 2.4 seconds" – and deletes the source image in the volatile memory (RAM) before the next frame is even processed. No storage takes place, no matching with databases, and no transmission of image data over the network. When we at Lumexo integrate such systems into an easescreen Crossfire CMS, compliance with this process chain is the decisive factor for approval by the data protection officer.
Sensor Technology Comparison: Optics vs. LiDAR vs. WiFi
For operators, the question arises as to which hardware best masters the balancing act between data quality and legal compliance. In 2026, a mix of different technologies will form the standard to increase redundancy and accuracy.
| Technology | Measurement Method | GDPR Relevance | Precision | Area of Application |
|---|---|---|---|---|
| Optical Sensors | Pattern matching at the edge | High (anonymisation critical) | Very high (demographics) | Retail, POS, Malls |
| LiDAR (Solid State) | Light pulses (point clouds) | Very low (anonymous per se) | High (tracking/paths) | Stations, Out-of-Home |
| WiFi/Bluetooth | Signal strength (MAC hashing) | Medium (hashing mandatory) | Medium (frequency) | Street furniture, events |
| Infrared (TOF) | Time-of-Flight measurement | Low | High (counting) | Entrance areas |
LiDAR: The New Gold Standard for Anonymous Tracking
Particularly in the area of large-scale tracking, LiDAR (Light Detection and Ranging) is gaining importance. A sensor, such as those found in modern automotive applications or high-end scanners, emits laser pulses and creates a 3D point cloud of the surroundings. Since no textures, colours, or facial features are captured, identification of individuals is technically impossible. Nevertheless, path analysis (heatmapping) can be realised with an accuracy of a few centimetres. For operators of LED infrastructures in transit areas, this is the safest method to measure pedestrian flows without ever coming into contact with biometric data.
Practical Example: Vienna-Schwechat Airport Terminal (Scenario)
Let's consider a concrete installation: A network of 15 double-sided 75-inch high-brightness kiosks (e.g., LG XE4F series), integrated into the departure gate areas.
The Objective: Measurement of gross contacts (OTS - Opportunity to See) and actual dwell time in front of the displays to validate advertising prices for agencies.
The Solution: Each kiosk contains a Quividi AMP (Audience Measurement Platform) sensor connected directly to a BrightSign XC4055 media player. Processing takes place locally.
- Step 1: The sensor detects a presence in the field of vision (up to 7 metres).
- Step 2: The algorithm classifies the person based on distance, head tilt, and dwell time.
- Step 3: Only aggregated figures (e.g. "14:00 - 14:05: 45 visual contacts, average 3.2s") are transmitted encrypted to the dashboard.
- Transparency: At the foot of the display, there is a small notice including a QR code, which informs about data processing in accordance with Art. 13 GDPR – an often underestimated but legally mandatory component of the infrastructure.
Legal Guardrails: What Will Be Phased Out in 2026
European supervisory authorities (EDPB) have toughened their stance. "Legitimate interest" (Art. 6 Para. 1 lit. f GDPR) as a basis for biometric analysis is viewed increasingly critically if no active consent is present.
- No retargeting without opt-in: Linking a person detected at the display with their smartphone (e.g. via Bluetooth beacons) to send personalised advertising later is inadmissible without explicit, prior consent (opt-in).
- No storage of hash values over 24h: Even if facial features are converted into a mathematical hash, this is often still considered personal data if it enables recognition over several days. In 2026, the "rolling delete" principle (deletion after a few hours) will become the standard for anonymous frequency measurements.
- Prohibition of emotion analysis: While age and gender (still) pass as statistical features, the analysis of emotions (happiness, frustration) is increasingly being criticised in ethical guidelines for AI. Lumexo recommends restraint here, as social consensus and regulatory severity are greatest in this area.
What We See in Practice
In the projects we implement at Lumexo for customers in the DACH region, clear trends are emerging:
- Hardware Consolidation: Customers demand sensors that are discreetly integrated into the housing (e.g. Alfalite Modularpix). No one wants additional boxes that look like "surveillance".
- Audited Software: Almost exclusively, solutions are used that can demonstrate an external data protection seal of quality (such as the ePrivacyseal). This massively relieves the operator's legal department.
- Hybrid Measurement: The combination of optical measurement for quality (advertising acceptance) and LiDAR/WiFi for quantity (total visitor flow) provides the most resilient data.
- Real-time Triggering: Instead of just collecting data, it is used to adjust content live. If the system recognises a group of young people, the playlist in the CMS automatically switches to corresponding content – entirely without personal reference.
- Focus on Connectivity: The API connection of measurement tools to programmatic booking platforms (SSP) is becoming a mandatory requirement to make DOOH spaces automatically tradable.
The Role of the CMS: easescreen and BrightSign as Enablers
A modern audience measurement system is only as valuable as its integration into the content workflow. When we at Lumexo install BrightSign Series 5 players, we use their ability to process serial data streams from the sensors without latency. In the easescreen Crossfire interface, the editor can then define threshold values.
An example from retail engineering: If the average dwell time in front of a promotional display (e.g. Samsung PM series) falls below 1.5 seconds, the system automatically generates a report for marketing. The problem is then often not the technology, but the content – measurement here serves as objective quality management for visual storytelling.
Conclusion: Trust Through Technology
Audience measurement in 2026 is not a necessary evil, but a precise instrument for efficient communication. Those who rely on transparency and radical edge anonymisation transform "surveillance" into "service optimisation". At Lumexo, we do not view the GDPR as an innovation brake, but as a guardrail for high-quality engineering in the field of visual infrastructure.
Lumexo Recommendation
- Rely on edge-native hardware: Avoid systems that send video streams to external servers for analysis. Anonymisation must take place in the sensor head or on the local player (e.g. BrightSign).
- Choose certified partners: Use established solutions like Quividi, which are regularly audited by independent experts. This reduces your liability risk during audits by data protection authorities.
- Integrate LiDAR for frequency analysis: For pure counting and path analysis in sensitive areas, LiDAR is the most legally secure and low-maintenance option.
- Transparent communication: Use the information obligation according to Art. 13 GDPR proactively. A clearly understandable notice on the display creates trust among passers-by and preventively fulfils legal requirements.