Define the two ratios once, in writing, and put the definition on the report. Nearly every utilisation problem is a definition problem wearing a technology costume.
Key takeaways
Define the two ratios once, in writing, and put the definition on the report. Almost every utilisation and realisation problem in a professional firm is a definition problem wearing a technology costume: two partners look at the same figure, hold different denominators in their heads, and reach opposite conclusions about the same person.
Utilisation is chargeable hours divided by available hours. Realisation is fees billed — or collected, and you must say which — divided by the standard value of the time recorded. Both are ratios, so both live or die on the denominator, and neither has a universal definition you can borrow.
Write the answers down as a one-page definition sheet, put a link to it on the report, and do not change the definitions and the reporting period in the same month.
Normalising time entries. The single largest obstacle to honest utilisation reporting is that half the time entries say “client work”. A model is genuinely good at mapping free-text descriptions onto jobs and service lines, and at flagging entries it cannot map rather than guessing. Have it flag; have a person resolve.
Finding the gaps. Days with no entries, weeks that total suspiciously exactly forty hours, entries posted three weeks late. These are data-quality signals and they matter more than the ratio itself, because a clean-looking number computed from reconstructed time is worse than an honest gap.
Reconciling WIP to the billing ledger. The classic month-end reconciliation between time recorded, work in progress carried and fees raised. Mechanical, tedious, and exactly the kind of joining work a model does well.
Drafting the commentary. Not the numbers — the paragraph explaining why the number moved, which a manager then corrects. Karbon’s reporting dashboards list realisation and client profitability among the metrics tracked, and offer export to a business intelligence tool; whatever the platform, the value is in the joining, not in the arithmetic.
A ratio you report is a description. A ratio you pay on is an instruction, and people follow instructions. Tie a bonus to utilisation and you will get utilisation — some of it by recording time that was not chargeable, some of it by declining the internal work that keeps a firm functioning, and some of it by not stopping to think about a file.
State the rule out loud in the definition sheet: where a target and professional standard conflict, the standard wins and the number takes the hit. A licensed person reviews and signs the file regardless of what the budget said, and the review is not optional because the job is over budget. Firms that never say this discover what their staff assumed at the worst possible moment.
Two retention rules sit under any hours-based report. In Ontario the ESA record-keeping requirements oblige an employer to record the dates and times an employee worked and to keep those records for three years, along with copies of any agreement to work excess hours or to average overtime. Federally, section 230 of the Income Tax Act requires records and books of account, and subsection 230(4) sets a general retention floor of six years from the end of the last taxation year to which they relate.
That matters for a practical reason: if your practice system is the only place hours exist, its retention settings are now a compliance control, and a migration that drops history is not merely inconvenient.
Three views, one page each. By engagement: estimate versus recorded versus billed, with variance ranked. By person: utilisation with the denominator stated, plus data-quality flags. By client: realisation and total recovered value, which is the bridge to identifying your least profitable clients.
Weekly for the engagement view, monthly for the other two. Anything more frequent produces noise; anything less means you find out about a job at the point it is unfixable.
A fourteen-person Ontario firm reports utilisation for two managers and concludes, from a wide gap between them, that the second is underperforming. The definition sheet exercise reveals the two figures were computed from different denominators — one excluded training and statutory holidays, one did not — and that the second manager carries the firm’s internal review work, all of it non-chargeable.
Recomputed on a single denominator, with internal review shown as a separate line rather than buried as non-chargeable, most of the gap disappears and the conversation changes from performance to allocation. Nothing about either person changed. The report changed.
This is reporting on delivery to a book of clients. The forecast that feeds it is in forecasting your busy season, and the partner-level summary is in building a partner dashboard. Your firm’s own bookkeeping and month-end close are a separate job on the bookkeeping automation page. For the employment-records background, see what records an Ontario employer must keep.
Should partners be in the utilisation number?
Either, consistently. If partners carry significant client delivery, excluding them hides your real capacity; if they mostly sell and review, including them makes the firm look idle. Decide and label it.
Billed or collected realisation?
Collected is the truer number and the harder one to move. Report billed for engagement management and collected for partnership reporting, and never quote one while calling it the other.
Is non-chargeable time bad?
No. It is how training, review, business development and internal improvement get done. What is bad is non-chargeable time that nobody chose. Break it into named categories and the argument becomes tractable.
Do I need per-person reports?
You need per-person data quality. Whether you publish per-person ratios depends on the culture you want, and that is a partnership decision rather than a reporting one.
A 30-minute call is enough to tell you whether AI pays for itself here.
