Biometric Attendance

A biometric system stores a representation of a physical characteristic for each employee and, at the point of attendance, matches a fresh reading against it. Fingerprint and face are the common ones in Indian workplaces; iris and palm vein appear in higher-security settings.

The specific problem it solves is proxy attendance: one employee marking another present. Cards, PINs and registers are all transferable, and in workplaces where attendance drives pay directly, they get transferred. A characteristic that cannot be handed over closes that.

It is worth being honest that the other benefits usually cited do not require biometrics. Accurate timestamps, integration with payroll, exception reporting and shift compliance are all available from card readers, mobile capture or web check-in. If proxy punching is not a real problem in a given workplace, the case for biometrics is weaker than it is usually presented, and the obligations that come with it are the same either way.

Collecting biometric data makes the employer the custodian of sensitive personal information about every employee, and that is a different posture from holding their address.

India's data protection regime places obligations on anyone processing personal data, and biometric information is treated as a sensitive category. The practical requirements an employer should be able to satisfy are these.

• A stated purpose. Attendance, and not repurposing the same data for access control, surveillance or performance monitoring without saying so.
• Notice and consent, obtained meaningfully rather than buried in a joining form, with a route for an employee who objects.
• Storage limitation. A template rather than a raw image where the technology allows, and confirmation from the vendor about which is stored and where.
• Retention limits, including deletion when an employee leaves. Biometric templates of former employees sitting on a device years later is the most common finding in any review of this.
• Security, including encryption at rest and in transit, and control over who can export the data.
• Vendor position. If the templates sit with a third party or on a device the vendor administers, the employer is still accountable for them.

Verify the current obligations under the applicable data protection law and rules before relying on any of this, since the regime and its rules have been evolving.

Employees do sometimes object, on privacy grounds or religious ones, and an employer that has not thought about it in advance handles it badly.

Two things make the objection easier to deal with.

• An alternative that is not punitive. A card, a supervisor-attested register or mobile capture achieves attendance for the small number who object. An alternative that requires the employee to seek permission daily is a refusal dressed as an accommodation.
• A clear position on whether consent is genuinely being sought. Consent obtained under a condition of employment is doing less work than it appears to, which is a reason to be able to justify the collection on its own terms rather than resting the whole thing on a signature.

Aadhaar-based biometric authentication is a separate matter from biometric attendance generally. Its use is subject to specific statutory restrictions on who may authenticate and for what purpose, and a private employer cannot assume it may use it for attendance simply because the employee has an Aadhaar number. That question should be taken on its own facts.

A fixed reader at a door assumes the employee passes that door twice a day. Where they do not, the system generates corrections rather than data.

• Field staff, who are at client sites, on the road, or at locations with no reader.
• Hybrid and remote employees, who have nothing to present a finger to.
• Multi-site staff who move between locations during the day.
• Contract and daily-wage workers at sites where enrolment is impractical and turnover is high.
• Anyone whose fingerprint reads poorly, which is a real and under-acknowledged problem for manual workers and older employees, and which produces failures the employee experiences as being accused of not turning up.

The diagnostic is regularization volume. A population that regularizes constantly is telling you the capture method does not match how they work, and no amount of process tightening around approvals will change that. Geotagged mobile capture, supervisor attestation or work-output-based recording are better answers for those groups than a stricter policy about the reader.

• Templates retained for former employees indefinitely, because deletion was never part of the exit process.
• Raw images stored rather than templates, usually because nobody asked the vendor which it was.
• Attendance data repurposed for monitoring or disciplinary use beyond the stated purpose.
• A device that fails for a subset of employees, whose absences then look like attendance problems rather than hardware ones.
• Enrolment treated as a one-off, so joiners work for weeks before being enrolled and their attendance is reconstructed by hand.
• The device clock drifting or the network dropping, so timestamps are wrong in ways that only surface in the payroll run.
• Biometrics deployed to solve a problem the workplace did not have, taking on sensitive data obligations for no operational gain.