What a bookkeeper's week actually looks like when you measure it
· 6 min read
Bookkeeping has an awkward shape for anyone charging a fixed monthly fee. The work is not evenly distributed: it clusters at month end, at quarter end, and around whatever deadline is next. And a great deal of it arrives in pieces too small to log, a query here, a chase for a missing receipt there.
Both properties mean that the felt experience of the month and the actual distribution of hours are different, and fees set from the felt experience tend to be wrong.
The two things measurement reveals
Which clients cost what. With a fixed fee per client, profitability is entirely a question of hours. Two clients paying the same can easily differ by a factor of three, and the difference is rarely the transaction volume. It is the state the records arrive in, how many queries come back, and whether the client answers them.
Where the spike is. Most bookkeepers know month end is heavy. Fewer know how heavy, or how much of the rest of the month is spent on things that could have been batched into it. Measuring turns that from a feeling into a distribution you can plan against.
The measurement problem
Bookkeeping resists timers more than most work. The day is fragmentary by nature: fifteen minutes on one client’s bank feed, a message from another, back to the first. Starting and stopping a timer for each fragment is more disruptive than the fragment itself, so it does not happen, and the fragments go unrecorded.
They are also where the time goes. A day that felt like three clients was often seven, and the four brief ones are invisible in any reconstruction.
Automatic tracking addresses the denominator directly. It records which application was in front and for how long, without you starting anything, so the small pieces are counted along with the large ones. For bookkeeping the application split is informative on its own: the accounting package, the spreadsheet, mail, the browser for bank portals, and a PDF reader for whatever arrived as a scan.
That last one is worth watching. Time in a PDF reader is usually time spent reading documents a client could have sent as data, and it is a cost that can be quantified and then discussed.
The daily allocation step
The record does not know which client anything was for. That allocation is a step you do, and doing it daily is what makes it accurate.
Two or three minutes at the end of the day, with the record in front of you: this ninety minutes was Client A’s reconciliation, that twenty was Client B’s queries, this half hour was practice administration.
Sequence is what you remember; duration is what you estimate badly. The record supplies the durations, you supply the sequence, and together they produce something reliable. Doing it weekly instead produces a plausible fiction, because by Friday the small pieces have gone.
What to do with a month of data
Sort clients by hours. Compare against fees. The effective hourly rate per client is the number that matters, and it is usually more variable than anyone expects.
Look at the query time. For a client whose hours are high, check how much of it is correspondence rather than work. A client generating many queries is expensive, and that is a conversation about process rather than a reason to resent them: a shared checklist or a change in how records arrive often removes most of it.
Map the spike. Plot hours by day of month. If the last four days carry a disproportionate share, there is a scheduling question: what could move earlier, and which clients could be on a different cycle to flatten it.
Find the unbilled block. Practice administration, software, professional development, chasing payment. It is real work and nobody pays for it directly, so it has to be inside the fees. If you have never sized it, the fees were set without it.
Hours open versus hours billed covers that distinction, and the time audit to do before raising your rates covers acting on it.
Where Punchcard fits
Punchcard sits in the menu bar, notices which application is in front by name only, and prints a receipt of the day at the closing time you set: one line per application with a total.
It reads no window titles and no document names. For bookkeeping that is a meaningful constraint rather than a missing feature: window titles in this work are client names and file names are often company names and periods. A tool recording them would be assembling a list of who you act for, on a laptop, and syncing it somewhere. This one cannot, because it never looks. There is no account and no networking code, so the record stays on the Mac. Why most time trackers want your window titles covers what the alternative involves.
What it does not do is produce a client breakdown by itself, or connect to your practice software. It gives you an accurate total and an accurate shape; the client allocation is the daily step above.
A month worth running
Week one. Collect without changing anything. You want a normal month, including its worst days.
Daily. Two minutes allocating blocks to clients.
End of month. Total per client, divide fees by hours, sort. Plot hours by day of month.
Then act on exactly one thing. The single worst client by effective rate, or the single largest avoidable block. Changing one thing and measuring again beats changing five and not knowing which helped.
The Sunday roll covers keeping the habit.
Questions
Should I move to hourly billing? Usually not. Clients value the predictability of a fixed fee, and so do you. The point is to set that fee from measured hours and to review it when a client’s hours change.
A client’s records arrive in a mess every month. That is a pricing question and a process question. Measure it first, so the conversation is about a number rather than a complaint, and offer the alternative: cleaner input, or a fee that reflects the work.
What about time on the phone? Not captured by a desk-based record. For bookkeepers this is usually a modest share, and it is worth noting by hand for the clients where it is not.
Does this help with quoting new clients? Considerably. A few months of data gives you hours by transaction volume and record quality, which turns a new quote into arithmetic. How to estimate hours for a quote covers it.