timesheet
On most small construction jobs the hours exist in three incompatible places. The crew knows roughly what they worked. The foreman has a notebook, or a run of text messages. Payroll has a deadline on Monday morning. The work of turning the first two into something the third can use falls on one person, usually in the evening, usually from memory that is already a few days old.
That reconstruction is the paperwork. It is not caused by a missing app. It is caused by a chain with gaps in it, where each handoff loses information that then has to be guessed. Closing the gaps is a design problem, and most of it can be settled before any tool is chosen.
Hours travel through four steps before they are worth anything. Capture, coding, review, approval. Each step has a known failure, and each failure has a different fix.
Capture is the record of who was on site and for how long. It breaks when it happens after the fact. A person writing down Tuesday on Friday will round, will forget the second site they visited, and will not remember which task the afternoon went to.
Coding attaches the hours to a job and to a kind of work. It breaks when the list of codes does not match the way the work is actually planned. Given a list of thirty codes and no obvious match, a crew picks the nearest one, and the totals stop meaning anything.
Review is someone who was there checking the entries before they become money. It breaks when it happens once a month, because nobody can recall a Tuesday five weeks back well enough to dispute it.
Approval is the moment the hours become the basis for payroll or for an invoice. It breaks when it is implicit. If hours flow to payroll without anyone saying yes, then every correction afterwards is a clawback conversation instead of an edit.
Most of the time lost to construction time tracking is spent repairing the first two steps at the last minute. Shifting effort earlier in the chain is the whole game.
The temptation is to collect everything. Resist it, because every extra field is a field that gets filled in wrong or left blank.
Four items cover payroll and job costing for a small operation. The date. The person. The hours. The job and the kind of work, which can be a single combined code. Anything beyond that should have to justify itself.
Two optional additions are usually worth their cost. A free text note, because the explanation for an unusual day is worth more later than any dropdown. And a quantity, where the work is measured in units: linear feet of trim, squares of roofing, cubic yards placed. Hours without a quantity tell you what was spent. Hours with a quantity tell you the unit rate, which is the only number that makes the next bid better.
What to avoid is a code list that grew by accretion. A practical rule is that the codes on the timesheet should be the same phases that appear on the schedule. If the schedule has demolition, rough-in, inspection, insulation, and finishes, then those are the codes. When the two lists diverge, nobody can answer whether a phase is over budget, because the hours and the plan are described in different vocabularies.
There are four common shapes, and the trade-offs are stable across products.
| Approach | What it gets right | What it gives up |
|---|---|---|
| Paper timesheet on site | Works with no signal, no devices, no training | Someone retypes every line, and errors enter at that step |
| Spreadsheet filled weekly | Free, flexible, familiar to office staff | Written from memory, no audit trail, breaks when two people edit |
| Clock in and out app with location | Timestamps at the moment, verifiable arrival | Needs devices and signal, raises privacy questions, records presence rather than work |
| Weekly sheet inside the tool that holds the schedule | Hours land next to the tasks they belong to, coding matches the plan | Still self reported, so review matters more |
None of these is strictly better. A crew of four on one site with reliable phones is well served by clock in and out. A mix of employees and subcontractors across five sites, where the subcontractors invoice rather than get paid hourly, is usually better served by a weekly sheet that the person fills in themselves, with half hour steps so nobody is debating seven minutes.
The choice that matters more than the shape is whether entry happens on the day or after the week. Same day entry removes the reconstruction, and reconstruction is where both the hours and the evenings go.
These two streams look alike on a timesheet and serve different purposes, which is why one tool often fits one and not the other.
For employees, the record exists to compute pay and to satisfy recordkeeping obligations. Which hours must be recorded, how long the records must be kept, and how overtime is calculated are set by law and vary by jurisdiction. Those rules are worth confirming with the relevant labour authority rather than inferred from a software feature list, because a tool's defaults are not a legal position.
For subcontractors and independent contractors, the record exists to support an invoice. The hours are the basis for a payment, not a payroll input, and the person supplying them is not an employee. In practice this changes three things. The contractor should enter their own hours, because someone else entering them starts to look like direction of work. The approval step becomes a commercial agreement rather than a check. And the output needed is a document that can sit behind an invoice, listing hours by project and work type for the period.
Mixing both streams into one screen without distinguishing them is where reconciliation goes wrong at month end. Keeping them separate, even in the same tool, means the payroll export and the invoice backup are produced from the same entries without being confused for one another.
Most of the arguments about site hours are not about whole days. They are about the edges, and the edges need a written rule before they show up rather than a ruling afterwards.
Travel. Time spent driving between two sites in the same day is treated differently from the commute to the first site, and different jurisdictions treat both differently again. Whatever the rule is locally, the timesheet needs somewhere to put it. A separate work type called travel keeps it out of the job's installed hours, so a phase does not look over budget because of a long drive.
Breaks. A clock in and out record produces gross presence. If breaks are unpaid, either the break gets recorded or the gross figure is wrong. The cleanest arrangement for small crews is a stated deduction agreed in advance for a full day, with an exception recorded on the days it did not happen. Deciding this once beats deciding it per person per week.
Two sites in one day. This is where reconstruction does the most damage. A person who visited two jobs will remember the day as a day, and the second job quietly absorbs nothing. Entry by line rather than by day is what fixes it: one row per job, hours in each, totalling to the day. Any capture method that offers a single daily figure will lose the split.
Standby and weather. Crews sent home after two hours, or held on site waiting for another trade, are hours that happened and that nobody wants to code. Give them a work type. Coding them to the phase makes the phase look inefficient. Leaving them out makes the payroll total disagree with the job totals, which is worse, because the discrepancy then has to be explained every month.
Write these four rules on one page and give it to whoever enters hours. Most timesheet disputes are not disagreements about facts. They are two people applying different unwritten rules to the same day.
Approval only works if it is small and frequent. Three decisions turn it from a chore into a five minute habit.
Pick a cutoff and hold it. Entries for the week close on a stated day. After the cutoff, changes are corrections, and corrections carry a note. A deadline that moves teaches everyone that the deadline is optional.
Name one approver per person, not per site. Approval by whoever is nearest produces inconsistent standards and gaps in the record. One named approver can answer questions about the entries months later.
Say what approval means. A reasonable definition: the approver believes the person was on site for those hours, and that the job and work type are right. It does not mean the work was good, and it does not mean the invoice is paid. Written down once, this stops approval from silently becoming a quality review that nobody has time for.
The reason to keep this cadence weekly rather than monthly is memory. A disputed entry from four days ago gets resolved in one message. The same entry from four weeks ago gets settled by whoever is more insistent.
A total of hours is an expense report. The useful reading comes from comparing spent hours against planned hours for the same phase, while the phase is still running.
That comparison needs three things to line up. The phase names on the schedule and the codes on the timesheet must match. The planned hours must have been written down before the work started, even roughly, because a plan reconstructed afterwards always agrees with the actuals. And the comparison has to be visible during the phase, not in a month end report, since a phase that is already finished cannot be steered.
This is the practical reason the timesheet and the schedule belong close together. When the hours land beside the tasks, the phase that is at eighty percent of its budget with half the quantity installed is obvious, and there is still time to add a person or reduce scope. When they live in separate systems, the reconciliation happens once a month, and by then the answer is history rather than a decision. A Gantt or resource board and a weekly timesheet in the same place remove the export step that usually kills this comparison.
Move entry from the week to the day, for one crew, for two weeks. Same day entry is the single change that removes most of the evening paperwork, because nothing has to be reconstructed. Then make the timesheet codes identical to the phases on the schedule, so spent hours can be read against planned hours while the phase is still open. Pinateca is free for a team of up to five, and the timesheet is an add-on that counts only the people who enter hours.
No. Location verifies arrival, which matters when crews are dispersed and attendance is genuinely disputed. It does not verify what was worked on, and it introduces privacy questions and device requirements. For a small crew with subcontractors who invoice, a self entered weekly sheet with a named approver usually produces a better record.
Weekly, against a fixed cutoff day. Memory is the limiting factor. A disputed entry from a few days ago is resolved in one message, while the same entry a month later is settled by whoever argues harder. Monthly approval also means a month of hours can be wrong before anyone notices.
Half an hour is enough for most construction work and keeps disputes away from single minutes. Smaller increments only pay off when the hours are billed to a client at an hourly rate and the client audits them. Whatever is chosen, keep it the same across every person and project so totals stay comparable.
Yes, provided the two streams stay distinguishable. The entries can share a screen, but the employee stream feeds payroll and legal recordkeeping while the contractor stream supports an invoice. Confirm the recordkeeping requirements for employees with the relevant labour authority rather than relying on a tool's defaults.
For four or five people on one site, often yes, as long as one person owns it and entry happens on the day. It stops being enough when several people need to edit at once, when the same hours are needed for both payroll and an invoice, or when spent hours need to be read against the schedule while a phase is still running.