Near One of Them Is Enough
An employee can be assigned to several sites, and clocking in requires being inside the radius of one of them — not all. The rule is simple; the three numbers that decide it are worth setting deliberately, because two of them are yours and one of them belongs to the phone.
A driver assigned to four depots is inside one of them and nowhere near the other three. That is the normal case, and the check is written for it.
What a site is
A site is a name, a pair of coordinates and a radius in metres. The radius can be set anywhere from ten metres to five kilometres, which is a wide range on purpose: ten metres is a specific doorway and five kilometres is a plantation, a construction site or a district a field team works across. Names are unique within your organisation, so two sites cannot quietly become one in a report.
Employees are assigned to sites, and the relationship is many-to-many. One employee can hold four depots; one depot can hold two hundred employees.
An employee assigned to no site has no geofence
The entire location check sits behind the question of whether this employee is assigned to any site at all. Assign nobody and the device check still applies, but the clock-in will be accepted from anywhere. This is the right default for a workforce that genuinely works everywhere, and it is the single most common reason an organisation believes geofencing is off when it is simply unassigned.
The three numbers
When an employee does hold sites, three numbers decide whether the punch is accepted, and it is worth being clear about who owns each of them.
| Number | Whose it is | What it does |
|---|---|---|
| The radius | Yours, per site | Ten metres to five kilometres. The distance inside which a punch counts as being at that site. |
| The distance | Computed on the server | Great-circle distance from the submitted coordinates to the site, worked out where the client cannot influence it. |
| The accuracy | The phone's | How confident the device is in its own reading. Anything worse than a hundred metres is refused before distance is even considered. |
The accuracy rule is the one people meet first and understand last. A phone that reports its position give or take a hundred and fifty metres has not made an error — it is being honest about a weak fix, usually indoors, usually in a stairwell or a basement. Accepting that reading would mean accepting a punch that could be anywhere within a three-hundred-metre spread, which makes the radius meaningless. So the attempt is refused with the measured accuracy in the message and an instruction to move outdoors or near a window.
A weak reading is not a wrong reading. It is a reading that cannot answer the question.
Nearest is not the same as matched
When an employee holds several sites, all of them are measured and sorted by distance. The system then looks for the first site, in that order, whose own radius contains the reading. Those are two different things, and the difference shows up in a specific and slightly surprising way.
Suppose an employee is assigned to a head office with a fifty-metre radius and a field area with a three-kilometre radius. Standing eighty metres from head office and two kilometres into the field area, the nearest site is head office — and the site they are matched to is the field area, because that is the first one in distance order that actually contains them. The record will say they clocked in at the field area, correctly.
The same logic produces the message when nothing matches: it names the nearest site, the measured distance to it, and that site's radius. "You are 214m from Mombasa Depot (your nearest assigned site) — you must be within 100m of one of your assigned sites." That is a message somebody can act on, which is the whole point of naming the specific site rather than saying the attempt failed.
The case where a code is scanned and the punch is still refused
There is one path worth understanding because it looks like a contradiction. If the location check fails but the employee scanned a valid code for one of their assigned sites, the system knows two things that disagree: the code says they are at the door, and the phone says they are two hundred metres away.
It resolves that honestly rather than conveniently. The message reports that the device's location signal looks unreliable rather than telling the employee they are somewhere they are not. The attempt is recorded with the code marked as verified and the distance the device claimed, so whoever reviews it can distinguish a bad fix from somebody in the wrong town. And the clock-in stays refused, because location is a mandatory factor and a code is not a substitute for it.
Staff work at one building
One site, a radius that covers the car park
Fifty to a hundred metres is usually right. Ten metres sounds precise and produces refusals from the far side of the reception desk.
Staff rotate across branches
Assign every branch they might work at
Being near one is enough, so over-assigning costs nothing operationally. Under-assigning costs a refused punch on the morning somebody covers a shift.
Field teams cover an area
One large site, not twenty small ones
The radius goes to five kilometres. A single area site is easier to reason about than a constellation of points, and the match rule means the large radius will win where it should.
Staff work at client premises you do not control
Consider no site assignment plus a code
A geofence around somebody else's building is a maintenance burden that grows with every new client. Where the work genuinely moves, the device check plus a corrections process is more honest than a fence nobody keeps current.
What AWRA OpsHub does today
- Sites carrying a name unique to your organisation, coordinates stored to seven decimal places and a radius settable from ten metres to five kilometres.
- A many-to-many assignment between employees and sites, so one person can hold several and one site can hold a whole shift.
- Great-circle distance computed on the server from submitted coordinates, so no client-supplied verdict about being in range is ever trusted.
- A match rule that accepts an employee inside the radius of any one of their assigned sites rather than requiring the nearest.
- A minimum accuracy requirement of a hundred metres, refusing a reading too vague to answer the question and saying so with the measured figure in the message.
- A refusal message naming the nearest assigned site, the distance to it and the radius that would have been needed.
- Honest reporting when a valid site code is scanned but the location disagrees — the device's signal is named as the likely problem, the attempt is recorded with the code verified, and the punch stays refused.
- A record of every attempt carrying the coordinates, the accuracy, the distance and the site, whether or not it succeeded.
More we can add to your workspace
- Sites drawn as shapes rather than circles, for a building, a yard or a stretch of road that a radius describes badly.
- A per-site accuracy threshold, so a rural depot can accept a vaguer reading than a city office where a good fix is expected.
- Time-bounded site assignment, so a two-week cover shift at another branch expires on its own rather than leaving a standing assignment behind.
- A map view of a proposed radius before it is saved, showing what the circle actually covers.
- Automatic site suggestion from where punches actually happen, for organisations setting up a site list they inherited on paper.
- A distance recorded on successful punches surfaced as a report, so a site whose radius is routinely being used to its edge can be found before it starts refusing people.
Where we point you to a specialist
- We will not accept a client-calculated distance, however convenient it would make an offline case. A device that reports it is in range is reporting an opinion it controls, and the whole value of a geofence is that the calculation happens somewhere the person being measured cannot reach.
- We will not silently accept a weak location reading in order to reduce refusals. A punch recorded from a fix accurate to plus or minus three hundred metres is a punch that cannot support the decision it will later be used for, and letting it through would make every other record less trustworthy by association.
- Whether you may geofence attendance, and what you may record about where staff are, is governed by employment and data protection law where you operate. Your advisers settle that and we configure what they permit.
Time-bounded site assignment and a map preview of a radius are the two contained pieces, and they address the two things administrators most often get wrong — assignments that outlive the cover shift, and a radius chosen without seeing what it covers.
Fences that fit the ground
A circle and a fixed accuracy rule cover most sites well. The work worth scoping is for the sites they cover badly — long buildings, rural depots, and assignments that were meant to be temporary.
Shaped sites
A drawn boundary rather than a radius, for premises that a circle either under-covers or spills across a road.
Assignments with an end date
Cover shifts that expire, so the site list reflects where people work now rather than everywhere they have ever worked.
A radius preview
See the circle on a map before saving it, which is the fastest way to stop a fifty-metre radius being set on a two-hundred-metre yard.
We publish scope, not dates.
Scope attendance sitesFive questions to ask about a geofence
Where is the distance calculated?
A good answer sounds like
On the server.
What ours actually is
On the server, from the coordinates submitted, using great-circle distance against the stored site position.
If somebody works at three sites, must they be near all of them?
A good answer sounds like
No — one is enough.
What ours actually is
One is enough. Every assigned site is measured, and the first in distance order whose own radius contains the reading is the match.
What stops a deliberately vague location?
A good answer sounds like
An accuracy threshold.
What ours actually is
Readings worse than a hundred metres are refused before distance is considered, with the measured accuracy shown to the employee.
What does a refused employee see?
A good answer sounds like
Something actionable.
What ours actually is
The nearest assigned site by name, the measured distance to it and the radius required.
Can a site code override a failed location check?
A good answer sounds like
No.
What ours actually is
No. The punch stays refused. What changes is the explanation and the record, both of which name the device signal rather than the employee.
Our take
Geofencing is easy to specify and hard to set up well, and almost every complaint about it traces back to a radius chosen without walking the site. The rules here are deliberately few: be inside one of your sites, with a reading good enough to mean something, measured somewhere you cannot influence. Everything else — how tight the circle is, who is assigned where, whether a code is also required — is configuration you own, and the refusal log will tell you within a week which of those decisions was wrong. What we would resist is the instinct to loosen the accuracy rule when refusals appear. That number is the one keeping the other two honest, and a fence measured from a reading nobody trusts is scenery.
Set the radius by walking it
The best half hour anybody spends on attendance is standing at the far corner of the car park with the site record open. If you would rather we did it with you, that is a conversation we have often.
Talk through site setupFrequently asked questions
What radius should I use?
For a single building, fifty to a hundred metres usually works — enough to cover the car park and the far entrance without reaching the road. Ten metres is available and is almost always a mistake outside a controlled doorway, because a phone with a good fix still moves a few metres between readings. For field areas, the radius goes to five kilometres and one large site beats a scattering of small ones.
Why was a clock-in refused when the employee was clearly on site?
Almost always the accuracy rule rather than the distance rule. Indoors, in a stairwell or against a metal roof, a phone can report a fix accurate only to a couple of hundred metres, and a reading that vague is refused before the distance is even measured. The message includes the measured accuracy, so the refusal log will confirm this in seconds.
Can an employee be assigned to too many sites?
Operationally, no — being near one is enough, so extra assignments do not weaken the check for the sites they are actually at. What they do cost is clarity: a person assigned to everywhere produces a record that cannot distinguish a normal morning from an unusual one. Assign where they might genuinely work, and prune the list when a cover arrangement ends.
What happens if an employee is assigned to no sites at all?
The location check does not run, and the clock-in is accepted from anywhere once the device check passes. That is the correct behaviour for a genuinely mobile workforce and the usual explanation when somebody believes geofencing is switched off. It is an assignment question rather than a settings one.
Does the system record where somebody clocked in from when it succeeds?
Yes. Every attempt, successful or not, stores the coordinates, the accuracy, the computed distance and which site it matched. That is what makes it possible to review a disputed day later, and also what makes this a matter to be clear with staff about, since it is a record of where people were at the start and end of their shifts.
Is there a way to allow one-off clock-ins away from any site?
Not as a punch. The route for a shift worked somewhere the geofence does not cover is a correction: the employee states the times and an approver decides on them, which leaves a named person on the record. Where that becomes routine rather than occasional, the answer is usually a new site or a wider radius rather than a workaround.
