XXII | Blog

How can we improve the reliability of store foot traffic KPIs?

Written by Alicia | Jul 24, 2026 8:38:20 AM

Store foot traffic KPIs form the basis for all retail operational decisions: staffing levels, merchandising management, and measuring promotional effectiveness. However, in most retail chains, this data is skewed by a simple yet systematic problem: staff movement is counted as customer traffic.

Excluding staff from foot traffic data is a prerequisite for any reliable KPI. This guide explains why—and how to automate the process without using biometric data.

Why Your Foot Traffic KPIs Are Probably Incorrect

The Problem of Staff in Foot Traffic Data

In a standard-sized store, staff account for between 8% and 15% of the visits recorded by sensors. In a store during off-peak hours, this ratio can exceed 40%.

If these visits are included in your foot traffic data:

  • Your conversion rate (sales / visitors) is underestimated
  • Your hourly foot traffic is overestimated, which skews staffing decisions
  • Your cross-store benchmarks are incomparable if some stores have more staff than others

Manual exclusion methods do not work at scale

Some retailers attempt to correct their data retroactively using adjustment factors or exclusion time slots (before opening, after closing). These approaches are imprecise and ill-suited to operational realities: midday restocking, a cleaning crew, or a zone manager walking through the store multiple times—none of these situations are captured by static rules.

How non-biometric exclusion works

The approach using silhouette recognition and movement patterns

Modern AI video analytics solutions, such as XXII’s CORE , distinguish staff from customers without using facial recognition or any biometric data. The system relies on non-identifying characteristics:

  • Movement patterns: staff follow recurring routes (backroom, checkout, restocking) that are distinct from customer paths
  • Traffic source areas: back entrances, restricted areas
  • Temporal behavior: duration of presence, frequency of visits

This data enables real-time automated exclusion without any personally identifiable database, without the need for additional RFID badges, and without requiring individual consent.

GDPR Compliance and Data Privacy

The non-biometric approach complies with the GDPR because it does not process any data that could be used to identify individuals. Images are not stored. Only aggregated behavioral metadata (position, direction, duration) is analyzed.

Employee data privacy is thus guaranteed: no individual profiles are created, and no personally identifiable history is stored.

Reliable KPIs Through Automatic Exclusion

Actual Conversion Rate

Once staff are excluded, the number of customer visitors is accurate. The conversion rate (the ratio of transactions to actual customer entries) becomes a reliable indicator for measuring the effectiveness of merchandising, promotional placement, and customer service.

Net hourly customer traffic

The reliability of hourly foot traffic statistics allows staffing levels to be aligned with actual customer flow, rather than with a curve inflated by staff comings and goings. The result: a better match between staffing levels and actual demand.

Average customer visit duration

By separating customer visits from business-related movements, the measured visit duration accurately reflects purchasing behavior rather than the time a salesperson spends in the area.

Heatmap Coverage Rate

Foot traffic heatmaps by zone are only meaningful if they are not skewed by staff repeatedly passing through the same areas. Automatic exclusion makes these maps useful for merchandising decisions.

Results Observed After Implementation

Non-biometric exclusion of staff produces measurable effects within the first few weeks:

  • Improved conversion rate: generally +2 to +5 points compared to the value calculated before exclusion
  • Schedule revisions: certain shifts that appeared to be overstaffed actually revealed customer traffic well below projections
  • Standardization of cross-site benchmarks: comparisons become valid because data is measured using the same criteria

Implementing automatic staff exclusion: what you need to know

Technical Prerequisites

  • Existing camera infrastructure (standard CCTV) with access to the video feed
  • Access to floor plans for defining staff entry zones
  • Integration with the existing counting system is possible via API

Implementation timeline

At a pilot site, the configuration and calibration phase typically takes 2 to 4 weeks. The system learns site-specific movement patterns to refine the distinction between staff and customers.

Compatibility with BI tools

The reliable customer data generated by CORE can be exported to standard Business Intelligence tools (Power BI, Tableau, etc.) via a REST API, enabling integration into existing dashboards.

Frequently Asked Questions

Does staff exclusion require employees to wear a badge or specific equipment? No. The CORE by XXII solution identifies staff based on their movement patterns and access zones, without requiring any additional equipment for employees.

How does the system distinguish between a customer who stays for a long time and an employee? The distinction is based on a combination of criteria: the area where the person enters the camera’s field of view, their trajectory, frequency of visits, and temporal behavior. No biometric criteria (face, gait, body type) are used.

Can foot traffic data be shared with partners such as landlords? Yes. Aggregated net customer foot traffic data—excluding staff—is more relevant and justifiable in negotiations with landlords or partner brands, as it reflects actual shopper traffic.

How accurate is the automatic exclusion of staff? In XXII deployments, the detection model’s accuracy reaches 98.6% under real-world conditions, which limits false positives (customers incorrectly excluded) to a negligible level for aggregated KPIs.

Conclusion

The reliability of foot traffic statistics is not a technical detail—it is the foundation of all operational and strategic decisions in retail.The exclusion of staff from foot traffic data—automated via non-biometric AI video analytics—is now a reality, fully GDPR-compliant, and delivers a return on investment that can be measured in weeks.

Request a CORE demo to see your own data reprocessed in real time.