Biometric Time & Attendance for 5,000 Workers with Automated Payroll Integration

Worker attendance management at the factory relied on a card punch system that had been in place since 2014. Each worker held a physical time card that was punched by a mechanical clock at gate entry. The punch cards were collected by supervisors at the end of each week and submitted to the payroll department, where a team of six clerks manually transcribed the punch times into the payroll software — an error-prone process that took the entire payroll team three full working days every fortnight.

Buddy punching — where a worker punches in for an absent colleague to credit them with attendance they did not fulfil — was widespread. The practice was an open secret on the factory floor. Supervisors estimated that between 3% and 5% of attendance records on any given day reflected buddy punching rather than actual presence. For a 5,000-worker factory with average wages of BDT 12,000 per month, a 3% buddy punching rate represented approximately BDT 18 lakh per month in fraudulent payroll. Beyond the direct financial cost, ghost attendance undermined production planning — daily production capacity models based on attendance records were systematically overstated, leading to delivery schedule commitments that could not be met with the actual workforce present.

Overtime calculation was a particular source of disputes. The Bangladesh Labour Act mandates specific overtime rates (two times the basic wage rate for hours beyond 8 per day), and workers were entitled to contest overtime calculations through the factory's grievance mechanism. Manual transcription errors in punch card reading regularly resulted in overtime underpayments and overpayments — both of which generated formal worker complaints. The factory's HR manager reported spending approximately 30% of working time each fortnight responding to payroll queries and disputes, diverting HR capacity from other responsibilities.

The evacuation muster situation was a safety concern flagged by the factory's external RSC (Remediation Coordination Cell) auditor. In the event of a fire or building emergency, the factory's evacuation procedure required department supervisors to take a manual head count at their muster stations and report to the Emergency Coordinator. This process took an average of 18 minutes in drill exercises to achieve a confirmed complete head count — far too slow to be useful in a real emergency. The RSC auditor's most recent report noted that the factory could not confirm a complete evacuation within the expected 10-minute post-evacuation window, which was a finding in the factory's RSC remediation action plan.

SSAL deployed 20 ZKTeco SpeedFace V5L facial recognition terminals across the factory's four entry gates, integrated with ZKBio Time 8.0 attendance management software and the factory's Orion ERP payroll system.

ZKTeco SpeedFace V5L Terminals — 20 Units: The SpeedFace V5L is a palm-vein and facial recognition terminal supporting up to 50,000 face templates, with infrared 3D face recognition that works accurately in outdoor light conditions — important for gate installations that are exposed to direct sunlight at various times of day. Recognition speed is under 0.5 seconds, supporting the high-throughput requirement of shift change periods when 1,200-1,500 workers may transit a single gate within a 15-minute window.

20 terminals were installed across the four gates: 6 at the main gate (3 entry, 3 exit), 4 at each of the two side gates (2 entry, 2 exit per gate), and 2 at the staff entrance (1 entry, 1 exit). Terminal placement was designed to support peak throughput without queue congestion during shift changeovers. Each terminal is wall-mounted at standard height with a canopy shade above to manage direct sunlight, and is connected to a UPS to maintain operation during the brief generator switchover periods.

ZKBio Time 8.0 Attendance Management Software: ZKBio Time is deployed on a dedicated server in the factory's IT room and serves as the central attendance management platform. The software receives real-time attendance transactions from all 20 terminals via TCP/IP, matches each transaction to the enrolled worker profile, and calculates attendance, tardiness, and overtime in real time according to the factory's shift schedules and Bangladesh Labour Act overtime rules.

Shift schedules are configured in the software for the factory's two regular shifts and peak-season night shift, with automatic shift assignment based on each worker's employment category. The software handles workers who regularly rotate between shifts through weekly schedule assignments managed by HR supervisors via the web interface.

Attendance exception reports — late arrivals, early departures, absences, and overtime exceeding thresholds — are generated daily and distributed automatically to department supervisors and HR. Supervisors can review and annotate exceptions via a browser-based supervisor portal from their workstation or mobile phone.

Payroll Integration — Orion ERP API: ZKBio Time integrates with the factory's Orion ERP payroll system via a REST API connection. At the end of each pay period, the approved attendance data — including regular hours, overtime hours, and leave deductions — is transmitted automatically to Orion ERP, which generates the payroll calculation without any manual data entry. The HR team reviews and approves the transmitted data in Orion before payroll processing, but the review is a verification step rather than a data entry exercise.

The API integration was configured by SSAL's software team in collaboration with the Orion ERP provider's technical team, requiring approximately two weeks of API documentation review and integration testing to ensure correct overtime rate mapping and leave type handling.

Real-Time Evacuation Head Count: ZKBio Time maintains a real-time "on-premises" count — the number of workers who have recorded entry without a corresponding exit for the current day. This count is accessible via the factory security and HR management dashboard and via a dedicated mobile app view configured for the Emergency Coordinator's phone. In the event of an evacuation, the Emergency Coordinator can access the real-time on-premises count within seconds of initiating the evacuation, providing the baseline against which muster station head counts are compared. When department supervisors complete their muster head counts, they submit counts via the mobile app to the Emergency Coordinator, who can compare against the real-time system count and identify any discrepancy — a process estimated to take 4-6 minutes in practice versus the 18 minutes required by the previous manual system.

Worker Enrollment: All 5,000 workers were enrolled in the ZKBio Time system over a 5-day enrollment campaign conducted at dedicated enrollment stations in the factory canteen and meeting areas. Each enrollment captured face templates (three positions), linked the biometric record to the worker's existing employee ID and national ID number, and issued the worker with a new employee RFID card (as backup authentication in the event of recognition failure due to injury, mask use, or environmental factors). Enrollment was conducted in batches by department during scheduled break periods to minimize production disruption.