Guide

GPS Geofencing:Circle and Polygon Boundaries, Phone and Tracker Evaluation, Android Limits, Ping-Interval Misses and Job-Site Clock Records

GPS geofencing for field crews: circle and polygon boundaries, phone and tracker evaluation, Android limits, ping intervals that miss crossings, clock records.

By Published Updated

11 min read

GPS geofencing is the comparison of a device's satellite-derived position with a virtual boundary drawn around a place, so that crossing the boundary produces an entry, exit or dwell event. In field service the boundary surrounds a job site, yard or territory, and the event becomes a clock prompt, a fleet alert or a time record.

The mechanism fails in ways that show up on timesheets and alert logs. A van drives through a customer's car park and the fleet platform records no entry, because no position report landed inside the boundary. A technician's phone sits still in a plant room all afternoon, and the exit is reported several minutes after the van has left. A phone restarts overnight and stops watching the day's sites. FieldRanger publishes this guide among its time, GPS and fleet tracking guides. It explains how a position becomes an event, then the two boundary shapes, where the comparison is computed, Android's documented limits, accuracy and boundary size, and the arithmetic of missed crossings. A sample service day in Knoxville, a setup procedure, a checklist and the software record follow.

Four terms carry a narrow meaning here. A position fix is one latitude and longitude estimate, with an error radius, produced from GPS, Wi-Fi or cell signals. A transition is the event a geofence produces: entry, exit or dwell. The update interval is the time between position reports a tracker sends to its platform. The evaluation point is where the fix is compared with the boundary: on the device or on a server.

Position fixes, boundaries and transition events

Every geofence event comes from the same chain. A receiver estimates a position. Software compares the estimate with a stored boundary. When the result changes between outside and inside, the software records a transition and passes it on: to a time clock, an alert rule or a log.

Each link in the chain has its own error. The position estimate has an error radius. The comparison happens only when a fix is available, so a device that reports once a minute is compared once a minute. The transition is reported after a delay set by the operating system or the platform. A geofence event is therefore the time the system became sure of a crossing, not the time of the crossing. Geofencing as a term, and its use on thermostats, is defined in the glossary entry on geofencing; the advertising use of the same mechanism is covered in geofencing marketing.

Circle and polygon boundaries

Two shapes are in use. A circle is a centre point and a radius. Android's geofencing interface defines every geofence that way, as a "circular area, or fence, around the location of interest" (Android Developers, Create and monitor geofences, last updated October 28, 2026). Fleet platforms add polygons. One Step GPS describes drawing "a virtual zone (circle or polygon)" for the platform to compare with the vehicle's reported location (One Step GPS, GPS tracker with geofencing, checked November 13, 2026). ThingsBoard's tracker module reads each device's coordinates "against its own perimeter server-scope attribute (a polygon or circle)" (ThingsBoard, GPS tracker geofencing, checked November 13, 2026).

The two shapes are tested differently. A circle test is a distance: the great-circle distance from the fix to the centre, compared with the radius. A polygon test is a point-in-polygon check. One standard method casts a ray from the point and counts boundary crossings: an odd count means inside (Wikipedia, Point in polygon, read November 13, 2026).

The choice matters on real sites. A circle around a house on a residential street takes in the neighbours' driveways. A polygon drawn to a commercial lot's boundary follows the property line and excludes the road. One Step GPS states that polygon geofences, "tighter than circles", reduce false alerts in dense areas.

Phone-side and tracker-side evaluation

Where the comparison runs decides what limits detection.

Geofence evaluation points and what limits each one
Evaluation pointWho computes the crossingWhat limits detectionTypical field use
Phone operating systemThe phone's location service, on behalf of the appOperating-system latency, battery management, permissionsClock prompts and time records
Fleet platform serverThe platform, from positions the tracker reportsThe tracker's update intervalVehicle entry and exit alerts
Tracker firmwareThe tracker itself, with geofences downloaded to itThe device's own fix rate and memoryAlerts that must fire without waiting for the server

On a phone, the app registers boundaries with the operating system and receives transitions; it does not compute them. On a vehicle tracker, the usual pattern is server-side: the tracker reports positions and the platform does the comparison. Digital Matter describes a third option. Geofences are downloaded "directly to supported devices", so that "the tracker itself can detect boundary changes and act immediately without relying solely on a software platform" (Digital Matter, Advanced geofencing trackers, checked November 13, 2026).

Android geofencing limits and behaviour

Android's developer guide documents limits that a time-tracking app inherits (Android Developers, Create and monitor geofences, last updated October 28, 2026):

  • An app can have "a limit of 100 per app, per device user" active geofences.
  • An app can request entry and exit events, or a dwell event after a set time inside, configured as a "loitering delay".
  • Alerts can be late. The guide states that "usually the latency is less than 2 minutes", about 2 to 3 minutes on average under background location limits, and "up to 6 minutes" for a device that "has been stationary for a significant period of time".
  • The system restores geofences after a Google Play services update or crash, but the app must re-register them after a reboot, a reinstall or cleared data.
  • An app targeting Android 10 or higher must request background location permission, ACCESS_BACKGROUND_LOCATION, in addition to fine location.
  • The guide recommends "a minimum radius of 100 meters" to allow for Wi-Fi location accuracy, and the dwell transition to reduce alerts from driving past a fence.

Background location is rationed as well: on Android 8.0 and higher, an app in the background "can receive location updates only a few times each hour" (Android Developers, Access location in the background, read November 13, 2026). iOS documents its own region-monitoring rules, including a minimum time beyond the boundary before an event is reported; those are summarised in geofencing for employee time tracking.

Two consequences follow for a field app. A technician with more than 100 open jobs cannot have a geofence on each, so the app watches the day's jobs. A phone that restarts without the app reopening has no geofences until the app re-registers them, which is a reason to check the first clock-in after a reboot.

Position accuracy and boundary size

The boundary cannot be smaller than the error of the fix that tests it. GPS.gov states that GPS-enabled smartphones "are typically accurate to within a 4.9 m (16 ft.) radius under open sky", and that accuracy "worsens near buildings, bridges, and trees" (GPS.gov, GPS accuracy, read November 13, 2026). One Step GPS puts fleet-grade trackers at "roughly 6–15 feet" under open sky.

Indoors the phone falls back on Wi-Fi and cell positions, which is the reason for Android's 100-metre minimum. A boundary sized for a GPS fix outdoors, say 30 metres around a house, works at the van and fails in the basement. Boundary size is therefore set by the worst place the device will be when the event matters: the plant room for a clock-out, the street for a vehicle entry.

Update intervals and missed crossings

A server-side geofence sees only the positions the tracker sends. Between two reports the vehicle moves a distance equal to its speed multiplied by the update interval, and a crossing that falls entirely between two reports leaves no point inside the boundary. One Step GPS makes the same point: "a tracker pinging once a minute can miss a fast pull-out", and its own devices "update every 1-30 seconds".

The arithmetic is short. At 25 miles per hour a vehicle travels 11.18 metres a second. A circle of 150 metres radius is 300 metres across through the centre. With reports every 60 seconds, the vehicle covers 670.6 metres between reports, more than twice the diameter.

Chance that a straight drive through a 150 m geofence at 25 mph produces no report inside it (geometric computation; sample speed and radius)
Update intervalDistance between reportsThrough the centre (300 m chord)100 m off centre (223.6 m chord)
60 seconds670.6 m55% no report inside67% no report inside
30 seconds335.3 m11% no report inside33% no report inside
10 seconds111.8 m0%: at least two reports inside0%: at least two reports inside

Each cell is one minus the chord divided by the distance between reports. That is the probability of no report inside for a random timing of reports along a straight path. A van that parks inside the boundary is detected at any of these intervals, because it stays inside long enough for a report. The miss affects drive-through events, such as a technician stopping briefly at a supply house or a vehicle passing a territory edge.

Sample service day: a Knoxville commercial job site

The sample is a hypothetical commercial HVAC service call in Knoxville, Tennessee. The times, radius and settings are sample inputs; the latencies are Android's documented figures. The technician's phone runs a time app with a dwell geofence on the site. The van's tracker reports every 30 seconds to a server-side geofence of the same site.

Sample day at one Knoxville job site, actual times against recorded events (sample times and settings; Android documented latencies)
EventActual timeSettingEarliest recordLatest documented record
Van enters the site geofence7:58:00Tracker every 30 s, server-side7:58:007:58:30
Phone dwell event, clock-in prompt7:58:00 arrivalLoitering delay 5 min8:03:008:05:00 (latency under 2 min)
Phone exit after a stationary afternoon15:30:00Exit transition15:30:0015:36:00 (stationary, up to 6 min)
Van leaves the site geofence15:31:00Tracker every 30 s, server-side15:31:0015:31:30

The day shows why the clock record and the vehicle record differ. The van record lags by no more than the update interval. The phone's dwell event is late by the loitering delay plus latency, by design, and the exit after a still afternoon can lag by six minutes. An automatic clock-out at the exit event would add up to 0.1 hours to that day, or half an hour across five such days. A prompt that the technician confirms, with the punch time taken from the confirmation, removes the lag from the pay record and leaves the geofence as evidence of presence.

Geofence setup and verification for job sites

The procedure applies to a job-site geofence used for clock prompts or vehicle alerts:

  1. Draw the boundary at the property line for a commercial lot, or as a circle of at least 100 metres where the platform supports circles only.
  2. Set the transition: dwell, with a loitering delay of a few minutes, for clock prompts; entry and exit for vehicle alerts.
  3. Confirm the update interval of the vehicle trackers; at 30 seconds or longer, treat drive-through events as unreliable.
  4. On each technician's phone, confirm background location permission is granted to the time app.
  5. Restart one phone, open nothing, and check whether the app shows the day's geofences before the first job.
  6. Verify on site: walk from the van to the work area and back, and compare the recorded transition times with the watch. A transition more than the documented latency late, or missing, points to permissions, battery optimisation or a boundary too small for indoor fixes.

For construction sites that stay open for weeks, the same boundary covers the closing walk-through, and the punch list template records the items found on that visit against the site.

GPS geofencing checklist

  • Boundary shape matched to the site: polygon at the property line where the platform allows it, otherwise a circle of 100 metres or more
  • Dwell transitions for clock prompts; entry and exit transitions for vehicles
  • Tracker update interval known, and drive-through alerts discounted at 30 seconds or longer
  • No more than 100 active geofences per app on an Android phone, so only the day's jobs are watched
  • Background location permission granted, and the app reopened after every phone restart
  • Punch times taken from the technician's confirmation, with the geofence event kept as evidence
  • A walk test on site after any change to boundaries, phones or app versions
  • Written notice to employees where state law requires it before location is collected, as set out for New York and Delaware in geofencing for employee time tracking

Trackers, phone location and telematics are compared as fleet methods in GPS fleet tracking.

Clock locations and fleet geofence alerts in FieldRanger

FieldRanger records location at the clock event and raises geofence alerts on the fleet side. With Geo-location time tracking, clock-ins and clock-outs record where the technician was, so hours match job sites, and Job-level time tracking logs each clock-in and clock-out against its job. Fleet alerts cover speeding, geofence exits and idle time. Vehicle tracking shows real-time GPS location for every vehicle on a live map, and Mileage tracking logs odometer readings for expense reports and IRS mileage records.

Territory management defines service zones and assigns technicians to them, and service area mapping shows coverage and gaps on a map. The GPS time tracking app page sets out what each clock event records. Plans are on the pricing page; Start free trial runs for 14 days with no card.

Frequently asked questions

What is GPS geofencing?

GPS geofencing compares a device's position with a virtual boundary around a place and produces an entry, exit or dwell event when the device crosses it. Field service uses it for clock prompts at job sites and alerts when vehicles enter or leave yards and territories.

How accurate is GPS geofencing?

Accuracy is set by the position fix: GPS.gov puts smartphones within 4.9 metres under open sky, worse near buildings and trees. Indoors a phone falls back on Wi-Fi, and Android recommends a minimum geofence radius of 100 metres for that reason.

How fast does a geofence alert arrive on Android?

Android's guide puts the latency under 2 minutes in the normal case, about 2 to 3 minutes under background location limits, and up to 6 minutes for a device that has been stationary for a long time. A dwell event adds the loitering delay the app sets.

Can a vehicle pass through a geofence without an alert?

Yes, when the evaluation runs on a server and the tracker reports infrequently. At 25 mph with 60-second reports, a vehicle crossing a 300-metre-wide circle has about a 55% chance of producing no report inside it.

What is the difference between circle and polygon geofences?

A circle is a centre and a radius, tested by distance. A polygon follows a drawn outline, such as a property line, and is tested by a point-in-polygon check, which keeps roads and neighbouring lots outside the boundary.

How many geofences can a phone app monitor?

Android allows 100 active geofences per app, per device user. A field app therefore watches the day's job sites rather than every customer address.

A geofence reports when the system became sure of a crossing. For a van that is the update interval; for a phone it is the loitering delay plus up to six minutes of latency. The time record is built on a confirmed punch, with the geofence event kept beside it as evidence.

Run field service on HubSpot with FieldRanger

Estimates with e-signature, scheduling and dispatch, a technician mobile app and invoicing, synced to the HubSpot deals and contacts the business already keeps.

Book a demo