How to track your working hours accurately

Almost everyone paid by the hour starts a personal time record at some point, and most of those records die within two months. The ones that survive have a few things in common, and none of them is discipline.

Why keep your own record at all

If your employer runs a clock-in system, it is reasonable to ask why you should duplicate it. Three reasons, in rough order of how often they matter:

Clock systems record attendance, not work. They know when your badge touched the reader. They do not know that you spent twenty minutes on a handover after badging out, or that you were called at home on Sunday.
You cannot audit a system you cannot see. When a payslip looks wrong, "I think it was about 168 hours" is not an argument. A dated record, entered at the time, is.
Aggregates only exist if someone keeps them. How many hours did you work last year? How much of that was overtime? Nobody will tell you unless you have been counting.

The value of the record is entirely in its credibility, and credibility comes from being contemporaneous. A log written the same day is evidence. A log reconstructed in April for the whole of the first quarter is a guess with dates on it.

The minimum viable record

People abandon time tracking because they try to capture too much. For a normal hourly job, a day needs four facts:

The date. Obvious, but note that a night shift beginning on Tuesday and ending on Wednesday belongs to one of them by convention — pick one and never change it.
Start and end. Actual, not scheduled. If they are the same on most days, that is a reason to use a template, not a reason to record the scheduled time.
Unpaid break. The single most common source of a small persistent error, because it is the one number people carry in their heads rather than check.
Which job, if you have more than one.

That is enough to produce a defensible monthly total. Everything else — bonuses, travel, vouchers, notes about what you actually did — is useful, but it is a refinement. Add it once the basic habit holds.

When to record

At the end of the shift, before you leave. Not on the way home, not that evening. The window in which you remember whether you took the full break is about ten minutes wide, and it closes when you start thinking about dinner.

If entering a day takes more than about fifteen seconds, you will eventually stop doing it. That is the whole design constraint. It is why templates matter: define your usual shift once, and a normal day becomes one tap, leaving your attention for the days that were unusual — which are the days that actually need attention.

The unusual-day rule: if a day differed from your normal pattern in any way — you stayed late, you were sent home early, you covered someone else's shift, you worked through the break — write down why, not just the times. In three months the times will be meaningless without it, and the "why" is what you will need if the day is ever disputed.

Handling the days you forget

You will forget days. The question is what you do about them, and there is a right answer: fill them in as soon as you notice, and mark them as reconstructed.

A gap in a record is worse than an approximate entry, because a gap reads as "did not work" to anyone summing the column, including you in six months. An entry with a note saying "reconstructed from rota, break assumed" is honest, usable, and clearly distinguishable from the days you recorded properly. Do not silently guess — guessing without a note is how a record loses the credibility that was the entire point of keeping it.

Rounding

Decide on a rounding rule once and apply it always. The two defensible options are:

No rounding. Record 08:53 to 17:06 as it happened. Most accurate, slightly more effort, and occasionally awkward if your employer rounds and you do not.
Round to the nearest five or fifteen minutes, symmetrically. Round both the start and the end to the same granularity, in both directions. Symmetry is what makes it fair over time.

What you should not do is round in a way that always favours one side. Always rounding your start time up and your end time down will lose you real money over a year; always doing the opposite will make your record indefensible the first time it is checked. The asymmetric version is common in employer systems and worth knowing about — if your employer rounds clock-ins to the quarter hour in their favour, an unrounded personal record is precisely how you would demonstrate it.

What to do at the end of each month

Ten minutes, once a month, is what converts a log into something useful:

Scan the month for gaps. Any working day with no entry and no absence recorded is either a missed entry or a day off you forgot to mark. Both need fixing.
Check the total against your expectation. If your contract is 40 hours a week and the month shows 210, you either did 40 hours of overtime or made a data-entry error. Both are worth knowing about immediately rather than in December.
Compare against the payslip when it arrives. There is a separate guide for that.

Keeping it going

Three things predict whether a time record survives its first year. Entry has to be fast enough to do while tired. It has to live somewhere you already have in your hand at the end of a shift, which in practice means your phone. And it has to produce something you actually look at — a monthly total you check against a payslip — because a record nobody reads is a chore with no reward attached, and chores with no reward stop.

That last point is the one people underestimate. The habit is not sustained by wanting good data. It is sustained by the first time the data catches an error.