How to prove your crew's hours were actually worked on site

Why paper timesheets fail

A paper timesheet is filled in at the end of the week, from memory, by someone who spent forty hours in the sun. It is not dishonest. It is just wrong.

Memory rounds. Nobody writes 7:12 to 3:48. They write 7 to 4, because that is roughly how it felt. A day split between two sites gets assigned to whichever job comes to mind first, moving cost off one job's margin and onto another's.

A paper sheet also cannot be checked. If a worker says he was there and you say he was not, there is nothing to appeal to. In most jurisdictions the duty to keep accurate time records sits with the employer, so an unverifiable record gets read against you.

Why an honour-system app is only half a fix

A phone app fixes the memory problem. Punches happen in real time and land in one place instead of six clipboards. That alone is worth doing.

What it does not fix is location. A clock-in button with nothing attached records the moment somebody pressed a button, not where they pressed it — a truck, a diner, or bed.

It also does nothing about buddy punching, one worker clocking in a mate who is running late. On paper that is a forged signature. On a plain app it is a text and two taps, which leaves no trace at all.

What the gap costs on a small crew

Work it out for yourself rather than trusting a headline number. Here is the shape of it, with made up figures. Take a crew of six, each averaging fifteen minutes a day of drift — a late start rounded down, a long lunch not deducted, a Friday sheet padded to the nearest half hour. That is ninety minutes a day across the crew, seven and a half hours a week, roughly one extra worker-day. At a $30 burdened rate, about $225 a week, or a bit under $12,000 across a working year.

None of that requires dishonesty. Fifteen minutes sits inside the noise of a normal day, which is why it never gets caught and why the fix has to be structural rather than a conversation about trust.

What counts as real evidence

Evidence is what survives someone disagreeing with it months later, in front of a third party.

The bottom row is the target. Not because it cannot be worked around, but because it produces a record with a date, a coordinate, a job number, and a named approver — the kind you can hand to an auditor or an unhappy worker and have the conversation end there.

How geofencing works, and where it does not

A geofence is a circle drawn on a map. You give the app a site address and a radius — a hundred metres covers a typical residential lot, a commercial site needs more. When a worker taps Clock In, the phone reports its position, the app checks whether it falls inside the circle, and the answer is stored with the entry. That is the whole mechanism: one yes-or-no check at the punch, no tracking involved.

Then you choose what happens on a no. Blocking the punch is cleanest but will strand a legitimate worker eventually. Flagging works better on a small crew: the punch goes through, arrives marked out of range, and you decide.

Be honest about the limits, because overselling this to your crew is how you lose them in the first month.

• A dead phone records nothing. If a battery goes at 6:50 the worker punches in manually later, and you are back to trusting him.
• GPS in a basement, a steel-framed core, or a dense downtown block can be far off or unavailable. Set the radius generously and expect some false flags.
• A phone left in the truck satisfies the geofence perfectly while its owner is at a hardware store. It proves the phone was on site, not the person, and never that they were working.
• Someone determined to defeat it can. Faked-location tools exist. The value is removing casual drift and making deliberate abuse take real effort.

So treat a geofence as a strong record rather than a lock. It makes an inaccurate hour visible, which is all you needed it to do.

The privacy line that matters

There is a real difference between checking a location at the moment someone clocks in and following them all day, and your crew will grasp it immediately if you explain it.

Location at the punch means the app asks for a position twice a shift, while the worker is holding the phone. Nothing in between, nothing after clock-out, nothing at the weekend. On iOS that fits the when-in-use permission, which a worker can confirm in his own phone settings.

Continuous background tracking is a different product and a different relationship. It runs while the app is closed, records routes, and tells you where somebody had lunch. Some fleet businesses have a reason for that. A framing crew does not, and asking for it turns a reasonable payroll control into surveillance your best people will resent.

What to tell your crew, because rollout is where this fails

The technology is easy. Introducing it badly kills it — announce it as theft prevention and every honest worker hears that you think they are stealing.

1. 1Tell them in person, before it appears on their phones. A push notification about a new tracking app, landing with no warning, is the worst introduction there is.
2. 2Lead with what it does for them: hours recorded as worked rather than reconstructed on Friday, overtime that stops getting lost, and a record that backs them as often as it backs you.
3. 3Say exactly what is captured and what is not. Then show one person the permission screen on his own phone, so the crew hears it from a colleague rather than from you.
4. 4Explain what happens when it fails, before it fails. Dead battery, no signal, forgot to punch: manual entry, flagged, approved, nobody in trouble. Crews fear the punishment, not the app.
5. 5Run it alongside paper for a week or two. The gaps that show up are your training material, and rolling back costs nothing.
6. 6Follow through on the flags. If out-of-range entries pile up unreviewed, the crew learns it is theatre.

How it changes payroll disputes

Most wage disputes on small crews are not fraud. They are two people remembering the same fortnight differently, with nothing between them but a smudged sheet. Those arguments are unwinnable, and they poison a crew, because someone walks away feeling called a liar.

With time, place, job, and an approval on every entry, the argument changes shape. You open the week and look together, and it either shows an eleven-hour Thursday or it does not. Disagreements shrink to single entries, which are quick to settle and often settle in the worker's favour. Saying that out loud early is what gets a crew on side.

It matters more if you are ever audited or sued over unpaid wages. Records made at the time of the work carry more weight than a sheet filled in later from memory. Keep them as long as your jurisdiction requires, with the approval history — an edited entry with a visible trail beats a clean one with no history.

One way to set this up

Whatever you use, the requirements are the same: a punch that captures position when it happens, a job on every entry, an approval step, and an export your payroll process can consume.

SiteLedger does this with a per-job geofence — you set the radius, clock-ins are checked against it, out-of-range punches are flagged, and hours export as CSV for whoever runs payroll. It is one option among several. What decides whether this works is not the app; it is whether you explain it properly and review the flags every week.

Common questions

Is it legal to track employee location at clock-in?

In most places, tracking work-related location during working hours is permitted when employees are told about it, and several jurisdictions require written notice or consent. The safer and simpler practice is to capture location only at clock-in and clock-out, tell the crew plainly what is recorded, and avoid background tracking outside working hours. Check the rules where you operate before you switch it on.

What happens if a worker's phone dies or has no signal?

The punch cannot be verified, so it becomes a manual entry that is flagged for your review. Agree the process before rollout: the worker records the time, notes the reason, and you approve it. Treat it as normal rather than as an accusation, or people will stop reporting the real times.

Does geofencing prove a worker was actually working?

No. It proves a phone was inside the site boundary at the moment of the punch. Someone can leave a phone in the truck and walk away. Geofencing removes casual drift and makes deliberate abuse take real effort, which is a different and more modest claim than proving productivity.

How large should a job site geofence be?

Around a hundred metres suits a typical residential lot. Large commercial sites, long linear work, or anywhere with poor GPS need more, because a tight radius on a bad-signal site produces false flags that erode trust fast. Start generous and tighten only if you see a real problem.

How do I introduce time tracking without upsetting my crew?

Tell them in person before it lands on their phones, lead with the benefit to them rather than theft prevention, and state exactly what is captured and what is not. Run it alongside paper for a week or two, and be clear that a dead battery is a flagged manual entry rather than trouble. Rollout is where this fails far more often than the technology is.