Geofencing

Frontline work increasingly happens away from a fixed clock. Home health aides visit patients. Service techs move between customer sites. Retail associates work across a footprint of stores. Construction crews move between jobsites. For all of these, a phone-based time clock is the only practical option — and without geofencing, the phone-based punch is a compliance hole. Geofencing is how employers trust the punch without requiring the employee to carry a dedicated device or be physically at a kiosk. Done right, it reduces wage-theft risk in both directions — employees can't pad their hours, employers can't dispute that work happened.

Take a 500-tech home services company. Each technician has a route of 6–10 customer visits per day. The geofencing setup: every customer address has a 150-meter fence; the technician's mobile app auto-detects entry and prompts for a punch-in; exit triggers a punch-out; time-between-fences logs as drive time at a different pay rate. The data flows to the WFM system in real time. When GPS accuracy is poor (urban canyons, indoor locations, parking garages), the app falls back to a manual punch with a timestamped photo. Managers see a daily summary; exceptions (missed punches, manual overrides, GPS anomalies) get flagged for review.

Geofencing is the feature most often demo'd perfectly and implemented messily. GPS accuracy varies by device, by weather, by indoor/outdoor, by cellular signal strength. A 30-meter fence works fine outdoors in the suburbs and fails inside a downtown office tower. The mature implementations use a layered approach: GPS-primary with Bluetooth beacon or Wi-Fi fallback, manual override with photo/timestamp evidence, and tolerance for known-fuzzy locations. Implementations that assume ten-meter accuracy produce constant false-negative punches and employee frustration.

In home healthcare, geofencing is partially regulated — Medicare EVV (Electronic Visit Verification) mandates location verification for home-based services, and geofencing is the predominant way to satisfy it. In construction, the work site itself is often the fence, and the challenge is that the site boundary isn't always what the mapping data says. In field services, the fence is usually the customer's address — which requires clean, maintained address data (a surprisingly hard problem). In retail and distribution, fences around store or warehouse locations are easy; the challenge is preventing punches from the parking lot or adjacent properties.

By 2026, geofencing becomes smarter and less literal. Instead of a hardcoded 150-meter radius, the system learns the actual patterns of punches over time — if the parking lot for a store location is 300 meters away and all the legitimate punches happen there, the fence adjusts. Agents handle the anomaly triage: "this punch looks unusual; here's why, here's the similar cases, here's the suggested resolution." Managers spend less time on punch forensics and more time on actual work. Privacy-wise, the tension between location tracking and employee expectations intensifies; companies that over-collect location data invite regulatory and cultural backlash.

• Over-tight fences. A 10-meter fence around a building's lobby fails routinely for employees parked in the adjacent garage. Start wider and tighten based on real data.
• Under-tight fences. A 500-meter fence around a downtown store location can accept punches from the coffee shop across the street. Balance accuracy with practicality.
• Single-signal dependence. GPS alone fails in basements, parking structures, rural areas with bad cellular, and during weather events. Bluetooth beacons or Wi-Fi fingerprinting as fallback is often worth the cost.
• Ignoring battery drain. Continuous location monitoring drains phones. Event-based location detection (geofence entry triggers a check, not constant polling) is gentler and produces better adoption.
• Privacy posture. Employees want to know when they're tracked, why, and what happens to the data. A clear consent and policy story reduces resistance; the absence of one breeds suspicion.
• Treating geofencing as the whole answer. Location validates "employee was somewhere." It doesn't validate "employee was working." The WFM layer still needs to combine location with other signals to get the full picture.