Five or six numbers a partner can act on this month, each with its definition printed on the page. Most dashboards fail because they show what the software exports, not what the partnership manages.
Key takeaways
Five or six numbers a partner can act on this month, each with a definition printed on the page and a named owner. Most partner dashboards fail for the same reason: they show what the software exports rather than what the partnership has decided to manage, and a tile nobody would act on is a tile that trains people to stop looking.
The discipline is subtractive. Start from the decisions the partners actually make — hire, decline, chase, reprice, invest — and put on the page only what changes one of those.
Are we winning work? New engagements accepted this month and the value committed; proposals outstanding and their age. Not “leads” — a lead is not a decision.
Can we deliver it? Committed hours against available review capacity for the next eight weeks, and jobs by stage with days-in-stage. The second is the early-warning tile: work that has not moved is the cheapest problem you will ever find.
Are we getting paid? Unbilled work in progress and its age, debtor days, and write-offs taken this month. Lock-up — WIP plus debtors expressed in days — is the single most useful number a professional firm can put on a page, because it converts two arguments into one.
Are we compliant? This is the section most dashboards omit and the one with the clearest external consequences. It should be short, and every tile on it should be an obligation rather than an aspiration.
It writes the queries, joins the systems, and reconciles the versions — the plumbing that stops most dashboards being built at all. It drafts the commentary: a short paragraph on why lock-up moved, which a partner corrects before circulation. It flags anomalies against the firm’s own history, which is a far better use of a model than forecasting. And it can answer follow-up questions against the underlying data so a partner is not waiting a week for a variant of the same report.
Karbon’s reporting and analysis describes pre-built dashboards covering outstanding jobs, team capacity, realisation and client profitability, with export to a business intelligence tool; that is representative of what practice platforms now ship. The constraint is never the visualisation layer.
What it must not do is compute a number nobody can define, or make the decision. A dashboard that recommends releasing a client, declining a job or not hiring has crossed from reporting into deciding, and the partnership — not the model — carries that. Where a number touches a professional obligation, a licensed person confirms it before it is relied on.
A five-partner Ontario firm replaces a fourteen-tab monthly pack with one page. The page carries eight tiles: engagements accepted; proposals outstanding by age; committed versus available review hours for eight weeks; jobs with no movement in fourteen days; WIP days; debtor days; write-offs this month; and a four-item compliance strip showing the date of the last AML effectiveness review, open breach records, records due for disposal, and unproven marketing consents.
Two things happen in the first quarter. The jobs-with-no-movement tile turns out to be the most valuable on the page, because it surfaces files stalled on a client query that nobody had escalated. And the compliance strip catches an effectiveness review that had slipped past its date — found by a tile that took an afternoon to build.
The fourteen-tab pack is not deleted. It becomes the drill-down nobody opens unless a tile moves, which is the correct fate for a detailed report.
This is partnership reporting on a firm delivering to a book of clients. The inputs come from utilisation and realisation reporting, client profitability and the busy-season forecast. Automating the firm’s own close so the numbers arrive on time is a different job, covered on the accounting automation page.
Should staff see it?
Share the operational tiles widely and the partnership economics narrowly. A capacity tile everyone can see changes behaviour; a profitability tile everyone can see changes conversations you may not want to have that way.
How often should it refresh?
Monthly for the page, weekly for the delivery tiles. Daily refresh on a partner dashboard produces noise and a habit of ignoring it.
What if two systems disagree?
Name one as the source of truth per number, print which one on the tile, and treat the discrepancy as a work item rather than resolving it silently in a query.
Should it show partner-level profitability?
Only if the partnership has agreed what that means and how it is used. Publishing it into an undecided partnership is a reporting decision that becomes a governance argument.
A 30-minute call is enough to tell you whether AI pays for itself here.
