Skip to main content

Returning Customers

The Returning Customers metric is designed to estimate how many visitors have been seen before, based on the presence of persistent wireless device identifiers detected within the store environment.

How it works

Modern smartphones and wearable devices periodically broadcast signals via Wi-Fi and Bluetooth. These signals include a hardware identifier (commonly referred to as a MAC address) that can, under certain conditions, remain consistent across visits.

When a device with a non-randomised identifier is detected again on a later date, the system can infer that the same individual has returned to the location. By aggregating these repeated detections over time, the platform estimates the proportion of visitors who are new vs returning.

This approach is:

  • Anonymous (no personal identity is known)
  • Probabilistic (based on device presence, not guaranteed person-matching)
  • Statistical (accurate at population level, not individual level)

Real-world coverage and accuracy

Due to operating system privacy controls, most modern smartphones randomise their MAC addresses. However, in real-world deployments, a meaningful subset of devices still broadcast stable identifiers, including:

  • Certain Android configurations
  • Older phones
  • Smart watches and fitness trackers
  • Wireless earbuds and accessories

In practice, this represents approximately 15-30% of visitors, depending on:

  • Region
  • Device mix
  • Retail environment (e.g. malls vs high-street stores)

While this is not a full census, it is sufficient to produce statistically valid trends when measured over time - particularly for comparing returning behaviour across days, weeks, or campaigns.

Privacy and system limitations

  • No facial recognition is used.
  • No images, biometric data, or personal identifiers are stored.
  • The metric is not designed for individual-level tracking, loyalty identification, or re-identification of a specific person.
  • Results should always be interpreted as directional insights, not absolute counts.

Important Notice - Feature Deprecation

Due to evolving privacy regulations and extended GDPR interpretation, Returning Customers based on Wi-Fi / Bluetooth identifiers has been deprecated as a standard feature since January 2026.

While the underlying methodology remains statistically valid, it is no longer enabled by default and is not included in new deployments.

This change reflects a precautionary privacy stance and does not imply a technical limitation of the system.