Chasing Timesheets Is a Symptom, Not a Job

Somewhere in your business, most likely on a Friday afternoon, a person is sending messages that say some version of "can you get your hours over".

Nobody hired that person to do that. It is not in anyone's job description. It emerged, gradually, because the alternative was payroll going out wrong. And because it emerged rather than being designed, it never gets examined — it is just what Friday is.

It is worth examining, because chasing hours is not a job. It is the visible symptom of three specific failures upstream, and none of them is fixed by chasing harder or by buying a timesheet app.

Failure one: hours are recorded away from the work

The fundamental problem is that in most contracting businesses, hours are written down somewhere other than where the work happened, at a time other than when it happened, sometimes by someone other than the person who did it.

An engineer finishes at four, drives home, and on Friday tries to reconstruct four days from memory, a van diary and their phone's location history. What they produce is an estimate. Usually a decent one. Usually rounded to the half hour, usually rounded in a consistent direction, and usually missing the twenty minutes spent waiting for a lift key or the hour lost to a site induction.

So two things are true at once: the office is chasing a document, and the document — when it finally arrives — is not accurate anyway. Chasing it faster does not make it more accurate. It just makes it arrive sooner.

This is one of the most consistently raised frustrations in the threads we reviewed. When contractors discuss what actually hurts about their systems, the answer that keeps coming back is the gap between the field and the office, and the manual re-entry that bridges it. Anything that closes that gap saves real time; nothing that decorates it does.

Failure two: the same hour is entered more than once

Follow one hour through. It gets recorded on a sheet or in a message. Someone types it into a spreadsheet to work out the week. It gets typed into payroll. And if it is contract work, it gets typed a third time into the valuation build-up for the month's application for payment.

Three entries, three opportunities for divergence, and — importantly — no authority when they disagree. If payroll says 38 hours and the valuation says 41, which is right? There is no answer, only an argument, and the argument is settled by whoever has more time to scroll back through WhatsApp.

That divergence is not only an internal annoyance. If a QS challenges your labour on a valuation and your own two records disagree, you have lost that conversation before it starts.

Failure three: it is nobody's actual job

The third failure is organisational. Chasing hours usually falls to whoever is least able to refuse it — often the person also doing invoices, answering the phone and pricing work. It is the definition of an interruption task: low-skill, high-frequency, impossible to batch, and it blocks the actual finance work behind it.

Ask contractors how much of the week goes on paperwork and the answers spread from about an hour a day up to more than half the week, depending on the role. The people carrying the most of it tend to be the ones covering both the work and the office — which is to say, exactly the people a small contracting business cannot afford to have doing data entry.

What actually fixes it

The fix is not a better chase and it is not a nicer form. It is moving the point of capture.

Capture at the moment, on site

If hours are captured when someone arrives and leaves — signing in on a phone at the site, rather than reconstructing on Friday — three problems disappear at once. There is nothing to chase, because it already happened. The figure is accurate, because it was recorded at the time rather than remembered. And it is timestamped, which matters enormously the first time someone disputes what your team was doing that week.

Make it the single source, then read from it

One record of hours. Payroll reads from it. The valuation reads from it. Job costing reads from it. This is unglamorous and it is the highest-value change available to most firms — it eliminates the re-keying, the divergence and the argument in one move, regardless of what software you use.

Accept six words and a photo

Any system that demands more input than an engineer will actually give it will be abandoned, and you will find out three weeks later when you need a record that is not there. Design for what people do: a photo, a few words, a tap. The job of turning that into something the office can read belongs to the software, not to the person standing in a plant room.

Then automate the exception, not the routine

Once capture is at the point of work, the only thing worth chasing is the exception — the person who did not sign in, the shift with no record against it. That is a much smaller list, it can be generated automatically, and it can be aimed at the right person rather than broadcast to a group.

Where GUVNR is on this

This is one of the areas where we can be reasonably concrete, so here is the honest split.

Running in production today, inside the contracting business that builds GUVNR: scheduling and shifts assigned through to a mobile portal, operatives completing shift reports on a phone from site, and timesheets. In practice that means the record is made where the work happened, which is the part that matters.

Currently being tested: AI turning those rough site notes into something the office can read without ringing back. The generator works; it is not yet part of the weekly routine, so we are not calling it live.

Under active development: the automation engine that spots the exceptions — a shift with no record, an operative who has not signed in — and chases the right person rather than putting a badge on a dashboard. Today that list exists; the thing that acts on it does not yet.

We are deliberately not claiming the second one is finished, because it is the part that would make the best headline. More on how we decide what we are allowed to claim in why we are building GUVNR inside a real UK contractor.

If you change one thing

Move the capture point. Everything else on this page follows from it, and it is achievable with almost any tooling — including a shared phone and a bit of discipline.

The test of whether it has worked is simple, and you will know within a fortnight: nobody sends a message on Friday asking for hours.