Case study
Leaders couldn't see realization, utilization or matter profitability in real time
The books closed monthly and the numbers arrived nine days later. We rebuilt a law firm's reporting on Microsoft Fabric so partners see hours, realization and matter margin while the work is still live.
- Overall realization
- +4.8 ptsOverall realization82.1% → 86.9%
- Write-offs
- ↓23%Write-offsCaught before billing
- Lockup (WIP + AR)
- ↓19 daysLockup (WIP + AR)118 → 99 days
- Reporting cycle
- 9d → dailyReporting cycleFrom a nine-day month-end wait
Nine days behind, every month
A Boston firm, 210 fee earners, six practice groups. Every month the finance pack landed nine working days after close. By then the work was done, the bill was out, and the discount had already been given. The engagement ran around five months, phased, for the CFO and the practice group leaders.
Nine days behind, always. The pack described a month that had already ended. Good for the record. Useless for changing anything.
Realization was a single number. One firm-wide percentage, with no way to tell whether value leaked at billing, at negotiation, or at collection.
Nobody knew which matters lost money. Profit was measured by practice group. A big-name matter could bleed for months and nobody would see it.
Utilization was anecdotal. Partners knew who felt busy. Nobody could show hours against target by person, level or group.
Value leaking between the timesheet and the bank
Every recorded hour loses value at three points before it becomes cash. The firm could see the total loss. It could not see where the loss happened.
- 82.1% realization: almost a fifth of recorded value never reached the bank.
- $6.4M of recorded time written down or written off last year.
- 118 days of cash locked in unbilled work and unpaid invoices.
Where recorded value disappeared, before:
| Stage | Value | Lost at this stage |
|---|---|---|
| Standard value (hours × rate card) | $100.0 | — |
| Billed, after partner edits | $91.4 | −$8.6 write-down |
| Invoiced, after client agreement | $85.3 | −$6.1 discount |
| Collected, cash in the bank | $82.1 | −$3.2 write-off |
Three separate decisions, made by three different people, none of whom could see what the others had already given away. The biggest leak was the write-down at billing, decided one matter at a time, with the total invisible until month end.
"We knew our realization number. What we couldn't tell you was which matters, which partners, or which stage of the billing cycle it was leaking from, and by the time we could, the bill had gone out."
— Chief Financial Officer, Boston-based law firm
What partners would actually open
We analysed three years of time, billing and collections data, and sat with partners as they wrote bills. The hard constraint was trust, not technology: a number a partner doubts is a number they ignore. What they needed:
- Realization split into write-down, discount and write-off.
- Profitability at matter level, not just practice group.
- True cost per timekeeper, including salary and overhead.
- Utilization against target by level, group and individual.
- Alerts while a matter is running, not after it closes.
- Partners see their own book; management sees the firm.
One model from timesheet to collected cash
Five systems joined into one model in Microsoft Fabric (time, billing, collections, payroll and overhead), with margin calculated on what a timekeeper actually costs, not the rate card.
One model, five sources. Joined at matter and timekeeper level, so every hour traces to cash.
Margin on true cost. Salary, benefits and overhead per timekeeper: what the hour really costs to deliver.
Alerts mid-matter. Budget burn, ageing WIP and margin drift flagged while there is still time to act.
Row-level security. Partners see their book. Group leaders see their group. Management sees the firm.
The view that changed the conversation plotted every matter by fee value and margin against the 25% target. Most sat at or above it, but a cluster of large, below-target matters, the blind spot nobody knew existed, included a top-10 client running at a 9% margin: high value, heavily written down at billing.
Two practice groups first, then the firm
A number that contradicts a partner gets argued with, not acted on. So the cost model was agreed and signed off by practice leaders before a single dashboard was built.
- Data foundation: time, billing, collections, payroll and overhead unified in a Fabric lakehouse at matter level.
- Agree the cost model: definitions of true cost, overhead allocation and matter margin locked with the practice leaders.
- Pilot two groups: run alongside the month-end pack and reconciled to the penny. Trust first, adoption second.
- Add alerts: budget burn, WIP ageing and margin drift pushed to matter partners while work is live.
- Roll out firm-wide: all six groups, partner-level access, and the manual month-end pack retired.
Visibility alone moved the numbers. Associates sat well under the 1,700-hour target while senior associates ran hot, and work was rebalanced inside one quarter:
| Level | Before | After |
|---|---|---|
| Partners | 1,420 hrs | 1,535 hrs |
| Senior associates | 1,905 hrs | 1,760 hrs |
| Associates | 1,330 hrs | 1,655 hrs |
Billable hours, annualized.
Decisions made while the matter is still open
The firm stopped explaining last month and started managing this one. Partners see budget burn early enough to have the scope conversation. Group leaders know which work is worth taking again. Finance spends month-end analysing the numbers instead of assembling them.
| Before | After | |
|---|---|---|
| Realization | 82.1% | 86.9% |
| Write-offs | — | ↓ 23% |
| Lockup | 118 days | 99 days |
| Reporting cycle | 9 days | Daily |
How these are defined here. Realization is collected cash as a share of standard value (hours × rate card). Write-offs are the value lost after invoicing. Lockup is the days of fees held in unbilled work in progress plus unpaid invoices (WIP + AR). Utilization is billable hours against the 1,700-hour annual target.
Built on Microsoft Fabric and Power BI.
Recognise any of this in your own estate?
Start with the problem rather than the technology, and we will tell you honestly whether it is ours to solve.
A professional-services firm, a mid-sized legal and advisory practice
They rolled out Copilot to save time. It kept guessing wrong.
A law firm handling 2,000+ cases a year
Billable work waited on unbillable work
A tier-2 automotive-components supplier feeding just-in-time under IATF 16949
The parts passed inspection. The defects shipped anyway.
