timesheet
Invoices go out on the fifth. On the third, four of the eleven contractors have submitted their hours. Two more send a message saying they will do it tonight. One reconstructs the month from a calendar and a memory of which days were busy, which produces a number that is close enough to invoice and not close enough to defend. Somebody spends the afternoon of the fourth sending the same reminder for the third time.
The tool is rarely what breaks here. Hours arrive late because entering them is a separate task in a separate place from the work, because nothing shows who is missing until someone counts by hand, and because the deadline is monthly, so forgetting is normal. A cloud based system fixes some of that by design and none of it automatically. Choosing well starts with being precise about which of four different jobs the hours are for.
Tools in this category look similar and are built for different purposes. Most abandoned rollouts are a mismatch between the tool's purpose and the team's actual question.
| The job | The question being answered | What the tool must do well |
|---|---|---|
| Billing | What goes on the invoice, and can it be defended | Rates, approval, a report a client will accept |
| Payroll and attendance | Who worked which hours, including overtime | Clock in and out, breaks, overtime rules, time off |
| Project cost | Did this project make money | Hours attached to specific work, cost against budget |
| Productivity monitoring | What are people doing during working hours | Activity capture, screenshots, location |
A team that needs the first and buys the fourth gets a rebellion. A team that needs the second and buys the first finds out during a labour inspection that break records are missing. The job that a small agency or studio almost always has is the first and third together: hours that become an invoice, and hours that reveal which projects are quietly unprofitable. That combination needs rates, approval and a link from hours to the work they were spent on, and it needs nothing at all from activity monitoring.
Four differences are real, and two of them cut the other way.
Entry happens where the person already is. A browser on a phone during a train journey is the difference between hours recorded on Friday and hours reconstructed on the third of the following month. This is the single largest effect, and it is worth more than any reporting feature.
There is one copy. Desktop applications with local files produce versions, and versions produce disputes. A shared system means the approved figure and the submitted figure are the same object.
Approval can be part of the flow rather than an email. Hours submitted, reviewed, locked. The lock is what makes the number usable later, because an editable timesheet is a draft no matter what it is called.
Against that, two costs. Data location becomes a question that has to be answered, particularly for client work under a contract that specifies where records are held, and some tools only allow a choice of region on their upper tiers. And availability becomes someone else's problem, which for time entry is usually tolerable and for a clock in terminal at a site entrance may not be.
Almost every complaint about time tracking traces back to this choice being made for the wrong reason.
A timer records a task as it happens. It produces accurate entries when it is actually used, and it fails in a specific way: the timer left running overnight, the hour of work nobody started a timer for, the constant small interruption of starting and stopping. Timers suit work done in long uninterrupted blocks by people who like the habit.
A weekly sheet is a grid, filled at the end of the day or the week, usually in steps rather than to the minute. It is less precise about when work happened, which for billing is normally irrelevant, and it is far more likely to be completed. For hours that exist to become an invoice, the sheet that gets filled in beats the timer that does not.
Granularity is worth deciding deliberately. Half hour steps produce a sheet somebody can complete from memory in two minutes. Minute level precision invites a kind of false accuracy, where a figure recorded as two hours and seventeen minutes was an estimate anyway. A lower granularity that is honest is better evidence than a high one that is invented.
The third pattern, the kiosk or clock in terminal, belongs to attendance rather than to project work and answers a different question entirely.
Pricing pages here need reading for two things: the price, and the definition of who counts as a person to be paid for. The second varies more than the first.
| Tool | Free tier | Paid entry point | Notable |
|---|---|---|---|
| Clockify | Yes | Standard at 5.49 US dollars per seat per month billed annually, or 6.99 billed monthly | Every active or invited user on the team page occupies a paid seat, including owner and admins |
| Toggl Track | Yes, for a limited number of users | Starter at 9 US dollars per user per month billed annually, or 12 billed monthly | Timesheet approvals appear on the Premium tier at 16 billed annually |
| A timesheet add-on to a project tool | Depends on the tool | Commonly priced per person entering hours | Managers who only approve may not be counted |
The figures above were taken from each vendor's own pricing page and change without notice, so they are worth rechecking before a decision. The pattern in them is the useful part. Seat counting is where the real cost hides. A team of twelve where only five people enter hours pays for twelve seats under a definition that counts every invited user, and for five under a definition that counts only people entering hours. At the prices above, that difference is larger than the difference between two vendors' headline rates.
The second thing to check is which tier holds approval. On at least one widely used tool, submitting and approving timesheets is not in the entry level paid plan, and on another it sits two tiers up. If the reason for buying anything at all is to produce a defensible figure, approval is not an optional extra, and finding that out after the trial is an expensive way to learn it.
Seven questions, in the order they usually turn out to matter.
Can hours attach to the work, not only to a project? Hours against a project tell you the total. Hours against the specific piece of work tell you which kind of work is unprofitable, which is the answer worth having.
Is there an approval step that locks the figures? Submitted, reviewed, locked, with a way to send something back for correction.
Do rates keep their history? When a contractor's rate changes in June, last quarter's reports must not change with it. Tools that store a single current rate silently rewrite the past.
Can more than one contract type be handled? Hourly, fixed price per project, and a fixed monthly fee are three different arrangements. A tool that only understands hours forces the other two into a spreadsheet, and then the totals live in two places.
Does the output match what gets handed over? A client or an auditor wants a page, not a screen. Totals by person, project and type of work, exportable as a file and printable as a document.
How does it show who is missing? A count of unsubmitted sheets is useful. A public list of names is a different thing socially, and it is worth knowing which one a tool produces before the team sees it.
Is entry possible on someone's behalf? There is always one person who will not use the screen. A lead entering their hours, with a record of who typed it, is more accurate than a monthly reconstruction.
Where hours are being recorded against work that already lives on a board, keeping both in one place removes the largest source of missing entries, which is having to open a second application to log the day. Whether a tool offers hours as part of the same system or only as a separate product is worth checking against its feature list and its pricing at the same time, since the answer to the first often changes the arithmetic on the second.
The chase is a process problem. Four changes remove most of it.
Make the deadline weekly, not monthly. A week is close enough to remember and small enough to reconstruct honestly if it is missed. Monthly deadlines guarantee that somebody is inventing a figure.
Put the entry screen next to the work. Every additional application between finishing work and recording it costs completion rate.
Report counts, not names. Three sheets outstanding is information. A list of who is late in a shared channel is a different message, and it buys short term compliance at a cost that shows up later.
Fix the process, not the person. When the same three people are always late, the usual cause is that their week does not contain an obvious moment to do it. A fixed slot, or a lead entering on their behalf, solves more than a reminder does.
Client contracts for hourly work sometimes specify more than a total. A record of which days were worked, a description of the work against each block of hours, and a named approver are all things that appear in agreements and are painful to reconstruct afterwards. Reading the contract before configuring the tool is quicker than rebuilding three months of entries because the description field was left switched off.
The same applies to retention. If hours are the evidence behind an invoice, they need to remain readable for as long as the invoice might be questioned, which is usually longer than the subscription is likely to last. A tool that exports totals and detail as a plain file, rather than only rendering them on a screen, is what makes that possible without keeping the subscription alive purely as an archive.
Some tools in this market include activity capture, periodic screenshots, photo capture at clock in, and location tracking. These exist because some contracts and some jurisdictions require them, and they belong to the fourth job in the table above.
The thing to be clear about is that these are not a stronger version of time tracking. They answer a different question, they usually sit on the more expensive tiers, and introducing them to a team that was asked for billing hours changes the relationship in a way that is hard to reverse. If a client contract requires them, that is a reason. Wanting more accurate invoices is not, because the accuracy problem is solved by a weekly deadline and an approval step.
Move the deadline from monthly to weekly and put the entry screen wherever the team already works, then add an approval step so the submitted figure becomes locked. Those two changes recover most of the lost time without any purchase. If the hours should sit against the actual cards rather than against a project name, so that the unprofitable kind of work becomes visible, Pinateca is free for up to five people and ten boards, with timesheets as an add-on for the people who enter hours.
Three things: entry from a phone or browser wherever the person is, one shared copy of the figures rather than versions in an inbox, and an approval step that locks a submitted sheet so it cannot be quietly edited afterwards. A spreadsheet can hold hours perfectly well, but it cannot show who has not filled theirs in, and it cannot make the approved number final.
A timer is more precise when it is actually used, and it fails through timers left running and work nobody started one for. A weekly sheet in half hour steps is less precise about when work happened and far more likely to be completed. For hours that exist to produce an invoice, completion matters more than precision, which favours the sheet.
Usually because the tool was built for a different job than the team's. Billing, payroll and attendance, project cost analysis, and activity monitoring are four separate purposes, and a tool strong in one is often weak in another. The second common cause is a monthly deadline, which is long enough that forgetting is normal and figures end up reconstructed from memory.
How a seat is counted, and which tier contains approval. Some tools charge for every invited user including administrators, while others count only the people who enter hours, and on a team where a minority record time that difference outweighs the headline rate. Approval also sits above the entry level plan on some tools, which matters if a defensible figure is the whole reason for buying.
Shorten the cycle to a week, put the entry screen where the work already is, and report the number of outstanding sheets rather than a list of names. Allowing a lead to enter hours on behalf of someone who will not use the screen is more accurate than waiting, provided the record keeps the name of whoever typed it.