A Timesheet Filled In on Friday Is a Memory, Not a Record
· 7 min read · Written by Faceela Research & Editorial Team
Reviewed by Ahmed Hassan Algammal — Founder and Enterprise Systems Consultant
Every firm that sells time has the same complaint, and it is always phrased as a discipline problem. People do not fill in their timesheets. We have told them. We have linked it to appraisals. We have stopped the month-end close until they are in.
The complaint is real and the diagnosis is wrong, and you can prove it without changing anything. Ask your best manager — the diligent one, the one who never misses a deadline — when she filled in last week. If the answer is Friday afternoon, the problem is not that people are undisciplined. It is that the timesheet is a separate document from the work, and it always loses.
What is actually lost, and when
A timesheet entry is not one fact. It is four: which client, what was done, how long it took, and whether anybody can be charged for it. They decay at different speeds, and the last one goes first.
Within the hour, all four are present. You know which of a client's two open matters you were on, in the words the work itself used. You know the call ran forty minutes rather than the ten it was supposed to. You know the half hour before it was internal and belonged to nobody.
By the end of the day, the interruptions are gone as individual events. By Friday, everything under an hour has gone with them, and on some grades that is most of the week. By month end, the only surviving fact is that the period was busy, and the entries are written backwards from the fee that has to be justified.
The shape of it
The same hour, written down four times
Within the hour
A record
Still in the entry
- Which client, and which of their two open matters
- What was actually being done, in the words the work used
- The interruption, and who caused it
- The ten-minute call that turned into forty
- The half hour that was for nobody — internal, and worth knowing
No longer recoverable
Nothing yet.
Before the end of the day
A good reconstruction
Still in the entry
- Which client, usually
- The shape of the day and its big blocks
- Anything unusual enough to have annoyed you
No longer recoverable
- Short interruptions, individually
- Which of two similar matters an hour belonged to
- The order things happened in, where it explained the overrun
On Friday, for the week
A memory
Still in the entry
- Days that were one long meeting
- The engagement that dominated the week
- Anything with a deadline attached to it
No longer recoverable
- Every task under an hour, which on some grades is most of the week
- Non-chargeable work, almost entirely — it has no client to hang on
- Why a task took twice what it was quoted at
- Rework, as distinct from work
At month end, for billing
An estimate written backwards from the fee
Still in the entry
- That the period was busy
- The engagements that were invoiced
No longer recoverable
- The week as anything other than a total
- Any basis for pricing this engagement differently next year
- Any answer to which client is expensive to serve
- The distinction between a job that ran over and one that did not
And every one of the four adds up to a plausible week
That is what makes this undetectable rather than merely inaccurate. A reconstructed timesheet sums to forty hours, spreads sensibly across three clients, and passes every review a firm actually runs. Nothing in it is flagged, because there is nothing to flag it against.
Look at the fourth rung, because it is the one that quietly destroys the firm's ability to price anything. At that point the timesheet is no longer evidence for the invoice — the invoice is the evidence for the timesheet. Somebody knows the client is to be billed thirty hours, and produces thirty hours. The document is now a reconciliation of a number that was decided elsewhere, which is a legitimate thing to produce and a useless thing to analyse.
The reason nobody catches it
Here is the part that makes this different from every other data-quality problem in a business.
A reconstructed week adds up. It sums to forty hours. It distributes across three or four clients in proportions that look exactly like a real week, because the person writing it knows what a real week looks like. It contains no impossible entries, no overlapping hours, no time logged on a public holiday.
There is no exception report that fires. There is nothing to compare it against. A wrong invoice can be caught by a purchase order; a wrong stock figure can be caught by a count; a wrong timesheet has nothing behind it, because the timesheet is the record of what happened. It is the same structural blindness that lets a payment certificate be arithmetically perfect around a hole.
So the firm reviews its timesheets, finds them plausible, and concludes that the data is fine. The evidence that it is not arrives two years later, as a fixed fee that has been unprofitable since the second renewal and nobody could see it.
One entry cannot be two documents
Most firms are asking their timesheet to do two incompatible jobs at once.
As a billing document, it needs to be defensible to a client. It should be tidy, round, described in language the client would accept on an invoice narrative, and free of anything embarrassing.
As a management report, it needs to be honest about where the week went. That means the rework, the hour lost to a system that was down, the meeting that should have been an email, the client who telephones four times a day, and the proposal that did not land.
You cannot get both from one entry. If the entry has to survive being read by the client, the manager will not record the two hours of rework, because writing it down creates a document about his own team. If the entry is only ever read internally, he will — but then it cannot be the source of the invoice narrative.
The firms that solve this do not solve it with a policy. They solve it by capturing more than they bill: the entry carries what happened, and the billing layer sits on top of it deciding what the client sees. That is a structural choice, and it is one to make before a system is configured rather than after.
Enforcement fails. Reducing the cost of entry works
Enforcement has been tried in every firm in this country, and its results are consistent: compliance goes up and accuracy goes down, because what enforcement produces is a filled-in grid rather than a true one. A rule that punishes an empty cell is a rule that guarantees a full one.
What actually changes behaviour is making the entry cost less than skipping it.
| The blank weekly grid | Capture from the work |
|---|---|
| Opened once a week, deliberately, as its own task | Opened because you were already in the file |
| Asks you to remember what you did | Asks you to confirm what it can already see |
| Client, matter and task all typed from scratch | Client and matter already filled from the record you had open |
| Non-chargeable time has nowhere natural to go | Internal categories present in the same list, at the same cost |
| Rewards a plausible week | Rewards a fast one |
The distinction is not about the software's features. It is about where the entry starts. A timer on the task you already opened, a phone entry against the engagement, a suggestion built from your calendar that you confirm or correct — these are all the same idea, which is that the system should already know most of the answer and be asking you to check it.
The hardest half of this is non-chargeable time, and it is worth saying plainly. A firm that only records billable hours has no denominator. It cannot compute utilisation, it cannot see what its overheads actually consume, and it cannot tell whether a manager is unprofitable or simply doing three days a week of internal work nobody asked to be told about. Recording the unbillable honestly requires that recording it is safe, and that is a management decision rather than a configuration one. If the first thing that happens to an honest entry is a question about it, there will not be a second.
What to make a vendor show you
On a live system, with a real user, not on slides.
- Record time from inside a task or a document, without opening a timesheet.
- Record it from a phone, in under fifteen seconds, against the right client and matter.
- Show a non-chargeable entry taking the same number of steps as a chargeable one.
- Show what the client sees on the invoice narrative against what the entry actually says internally.
- Show a timesheet entered on the day beside one entered five days late, and show that the system knows which is which.
- Produce utilisation by person, including the non-chargeable, and reconcile the total to contracted hours.
- Correct an entry after it has been billed, and show what happens to the invoice and to the recovery figure.
Item 5 is the one that decides it. Every system captures a date for the work. Very few capture the date the entry was made, and without that second date a firm has no idea which of its data is a record and which is a memory — so it cannot tell a manager whose recovery is genuinely poor from one whose timesheet is simply late. Ask for the lag as a reportable field, and watch the demo carefully when you do.
Item 3 sounds trivial and is not. If logging internal time takes more clicks than logging billable time, the system has an opinion about which one matters, and the staff will read it correctly.
The short version
A timesheet written on Friday is not a bad record. It is a different kind of document — a plausible reconstruction that sums correctly, passes every review, and contains almost none of the facts a firm needs to price its next engagement.
Nothing in the business can detect it, because there is nothing behind a timesheet to check it against. That is what separates this from every other data problem: the wrong answer is indistinguishable from the right one at the point it is entered, and the cost surfaces years later in a fee that stopped making sense.
Enforcement produces full grids. Lower cost of entry produces true ones. The entry has to start from the thing being worked on, and non-chargeable time has to be as easy to record as chargeable time and as safe to record honestly.
Everything a firm wants to know about its economics is downstream of this. Unbilled work in progress, the realisation bridge and client profitability are all computed from entries that have to be true first — which is why a profitability report built on Friday timesheets is a well-formatted opinion.
Where this sits in the wider picture is the professional and business services page; the suite is ServX; and if the honest next step is fixing the flow of work before choosing software, that is process automation.