Sizing the Boundary

The radius is the single most consequential setting, and it is usually left at whatever the product suggested.

The two errors

False rejection: the worker is at work and the punch is refused.

False acceptance: the worker is not at work and the punch succeeds.

They are not equally bad, and treating them as though they are produces the wrong radius.

Why the asymmetry matters

A false acceptance costs a few minutes of pay, occasionally.

A false rejection costs a worker standing at the door unable to start, then working unpaid or stopping work to resolve it.

It also costs the supervisor's time, every time.

And repeated, it costs the system's credibility — people stop trusting it and start asking for manual overrides, which defeats the point entirely.

So the radius should lean generous, and the residual acceptance risk should be handled by something other than geometry.

Working out the number

Start from your measured indoor error, not from a default.

Radius should exceed the error you observed at the worst indoor point, with margin.

Then look at what that circle contains. If it reaches a residential street, the system accepts punches from people at home in those houses.

If the honest answer is that no radius both accepts indoors and excludes the street, geofencing is the wrong method for that site.

That is a finding, not a failure, and it is better reached before deployment.

Shapes other than circles

Polygons fit an awkward site better than a circle, and most products support them.

Multiple small zones for a campus beat one large one.

Exclusion zones — accept inside the building outline but not in the adjacent car park — are offered by some products and are usually more trouble than they are worth given the error.

Per-site radii

A single organisation-wide radius is wrong for every site.

A city-centre shop and a rural depot have different error profiles and need different numbers.

Ask whether the product supports per-site configuration. Many do not, and that is a real limitation rather than a detail.

Reviewing it

Track refusal rates per site.

A site with a refusal rate far above the others has a boundary problem, not a workforce problem, and checking the geometry before speaking to anyone is the correct order.

Re-measure after any building work, which changes the wifi and mast picture more than people expect.

Watch the refusal rate by site

The number that tells you whether the radius was right.

Per site, per week, as a proportion of attempts.

A site far above the others has a geometry problem, not a workforce problem.

Check the radius and the coverage before speaking to anyone, which is the correct order and the one most often reversed.

Re-measure after building work, which changes the wifi and mast picture more than people expect.

A concrete product reference

When translating this principle into a buying test, this implementation prompt provides a concrete workflow reference. Verify current behaviour in a trial and judge it against the purpose and limits above.