Unbilled work is not one problem. It is four, and only one of them is solved by nagging people about timesheets.
Key takeaways
You find unbilled work by comparing two lists that most firms never put side by side: everything that shows the firm did something for a client this month, and everything that appears on a draft bill. The first list is easier to build than partners expect — calendar entries against a matter, documents saved, emails sent, portal activity, job status changes, third-party invoices received. Anything on the first list with no counterpart on the second is a candidate.
That reconciliation is the whole method, and it is a far better instrument than asking people whether they have finished their timesheets. It finds work nobody remembers doing, which is precisely the work that gets lost.
Unrecorded time. Work done, never entered. The fix is capture at the point of work, not exhortation at the end of the month.
Recorded but not on a draft bill. Time exists, but the job was not included in the billing run — usually because its status was never moved. This is the largest category in most firms and the cheapest to fix, because it is a workflow gap rather than a behaviour problem. Practice platforms exist for exactly this visibility; Karbon describes knowing where every job stands and who is working on what as its core proposition.
Out-of-scope work. Real work, done willingly, never quoted and therefore never billable without a conversation. Flagging it is valuable even when you decide not to charge for it, because the pattern is the argument for next year’s fee.
Disbursements not on-charged. Search fees, filing fees, courier, third-party reports. Match supplier invoices to matters and flag those with no corresponding recharge.
There is a reason not to let unbilled work drift quietly across a year end, and it is not only cash flow. Under subsection 168(1) of the Excise Tax Act, tax on a taxable supply is payable by the recipient on the earlier of the day the consideration is paid and the day it becomes due. And subsection 152(1) of the Act deems consideration to become due on the earliest of three days: the earlier of the day the supplier first issues an invoice and the date of that invoice; the day the supplier would have issued an invoice but for an undue delay; and the day the recipient is required to pay under a written agreement.
Read the middle limb again, because it is the one professionals are surprised by. The timing of the tax is not entirely within the supplier’s control — a firm that finishes work and simply does not get round to invoicing cannot assume the liability sits wherever its billing habits leave it. What counts as undue delay in any particular case is a question for your own tax adviser on the facts; the structural point is that month-end unbilled work is a compliance question as well as a management one. On the invoice itself, the requirement to show your GST/HST registration number is worth checking against your own templates.
Schedule it four or five working days before the billing run, not on the day. The output is a single list per partner, grouped by client, with each candidate showing the evidence behind it and a proposed category. The partner does three things: confirm or reject each candidate, decide whether out-of-scope items are billed or absorbed, and identify anything that requires a client conversation before it appears on an invoice.
The reason for the four-day gap is that the third of those actions takes time. An out-of-scope item discovered on billing day gets absorbed, every time, because the alternative is an awkward call on a deadline. Four days earlier, the same item gets a conversation and often gets paid.
A consultancy running monthly billing turns on a sweep that reads matter calendars, documents saved, sent email and the supplier ledger. In its first month it surfaces three kinds of thing: a series of client calls attended by two people where only one recorded time; two engagements whose status never moved to “ready to bill”; and a set of third-party report fees paid but never recharged.
Only the first of the three is a timesheet problem. The other two are process defects that no amount of chasing people about time entry would have found, which is the general lesson. Measure it with numbers you already hold: value of work billed in the month it was performed, count of jobs missing the billing run, and disbursements recharged as a proportion of disbursements incurred.
Does this mean surveilling staff?
It should not. Read matter-level artefacts, not individual keystrokes or screens, and say so in writing to the team before you start. The purpose is finding work the firm did, not auditing people.
What about work in progress we have decided not to bill yet?
Mark it as a deliberate deferral with a reason and a review date, so it stops appearing as an exception. An exception list that shows the same twelve jobs every month is ignored within two months. That decision belongs in the monthly WIP review.
Can it write the invoice?
It can assemble a draft with narratives and disbursements attached. Whether the amount is right, and whether the client has been told, remains a decision a person makes and signs.
A 30-minute call is enough to tell you whether AI pays for itself here.
