Building a Firm Dashboard That Someone Actually Opens

Almost every firm I have watched build a dashboard built a good one. The layout was clean, the numbers were correct, the charts were the right charts. Somebody spent a genuinely serious weekend on it. And then, three weeks later, nobody was opening it. Not because it was wrong, but because looking at it never changed what anyone did on Monday morning.

That is the actual failure mode, and it is worth naming precisely, because firms usually misdiagnose it. They assume the dashboard failed because it lacked something, so the next version has more panels, more filters, more breakdowns by attorney and practice area and referral source. The second version gets abandoned faster than the first. More data was never the missing ingredient. What was missing was a decision attached to each number and a person whose week changes when that number moves.

So this is not a post about which metrics to put on a screen. It is about the structure underneath: choosing few over many, separating the numbers that tell you what already happened from the numbers that tell you what is about to, setting a review cadence that survives a busy trial month, and being honest that a dashboard nobody acts on is decoration with a login screen in front of it.

Start from the decision, not from the data

The right first question is not "what can we measure." Every practice management system can measure a hundred things, and the temptation to display all of them is exactly what kills the dashboard. The right first question is "what decisions do we make repeatedly, and what would we need to see to make them faster or better." A firm makes a small number of recurring decisions: who gets the next intake, which matters need a partner nudge, whether to hire, whether to chase a receivable, whether a client is drifting. Five or six, honestly.

Once you have that list, each number on the dashboard has to earn its place by attaching to one of those decisions. If a metric is interesting but no decision hangs off it, it belongs in a quarterly review document, not on the screen someone opens every Monday. This single filter usually cuts a proposed dashboard by more than half, and the half that survives is the half people actually read. A number with no decision behind it is trivia, and people stop reading trivia quickly.

Why five numbers beat twenty-five

There is a practical reason small dashboards get used and large ones do not, and it has nothing to do with attention span. A screen with five numbers can be scanned in under a minute, which means it fits into the gap before a call. A screen with twenty-five numbers requires a decision about where to look first, and that decision costs enough effort that a busy attorney defers it. Deferred once is fine. Deferred every week is abandonment.

The other reason is signal. When you display twenty-five numbers, most of them are moving slightly at any given moment, and normal variation looks like meaning. People either chase noise or, more commonly, learn to ignore movement entirely because movement is constant. With five numbers, a change is genuinely conspicuous. You want the emotional response to a shifted number to be "something happened," and that response only exists when the baseline is quiet.

Leading and lagging indicators are not the same job

A lagging indicator reports the result of work already finished. Collected revenue, realization, matters closed, invoices paid. These numbers are trustworthy and completely uncontroversial, and they are also, by the time you read them, history. You cannot act on last quarter's collections in a way that changes last quarter. Their real job is scorekeeping and confirming whether a change you made actually worked.

A leading indicator moves before the result does, which means acting on it can still change the outcome. Unbilled work-in-progress that is aging past a threshold. Matters that have not moved a stage in a set number of days. Intakes sitting unconverted. Deadlines clustering in a single fortnight. These are messier and more arguable than collections, and that is precisely why they are useful. A dashboard made entirely of lagging indicators is a very accurate rear-view mirror, and firms build them constantly because they are easier to compute and harder to dispute.

What a leading indicator actually looks like inside a law firm

The best leading indicators in a firm are almost always about things sitting still. Legal work fails quietly, and it fails by stalling rather than by exploding. A matter that has not changed stage in three weeks is not necessarily in trouble, but the population of such matters is a genuinely reliable predictor of client complaints, write-offs, and the awkward call that starts with "I know it has been a while." This is why a matter stage tracker matters more than it first appears. Casely's stage tracker is a clickable stepper configured per firm and per practice area, which means "stalled" is defined by your own workflow rather than by a vendor's idea of one, and stage age becomes a number you can put on a dashboard honestly.

The same logic applies to money that has been earned but not yet billed. Unbilled hours are a leading indicator of a cash flow squeeze roughly a billing cycle before the squeeze arrives, and unlike collections, you can still do something about them today. One-click invoicing exists precisely so that acting on that number takes minutes instead of an afternoon, because a leading indicator you cannot act on cheaply gets ignored within a month. If seeing the number and fixing the number are separated by three hours of administrative work, the dashboard becomes a source of guilt rather than a tool.

Pair every lagging number with the leading number that moves it

This is the single structural change that turns a decorative dashboard into a working one. Do not list metrics as a flat set. List them in pairs, where each lagging number sits next to the leading number that drives it. Collections sits next to unbilled work-in-progress and invoice age. Matters closed sits next to stage aging. New client revenue sits next to intake conversion. Write-offs sit next to time entries logged more than a week after the work.

The pairing does something a flat list cannot. It converts an observation into a hypothesis, and a hypothesis into an action. When collections dip and unbilled work-in-progress has been climbing for six weeks, you have both the symptom and the cause on one screen, and the conversation in the meeting is about billing turnaround rather than about whether the number is real. Firms waste an extraordinary amount of meeting time arguing about whether a number is accurate, and the pairing structure short-circuits that argument by pointing at the mechanism.

The review cadence is the dashboard

A number without a scheduled moment where someone looks at it is not a metric, it is a database field. The cadence is not an administrative wrapper around the dashboard. It is the dashboard. Fix a standing time, keep it short, and make it survive a busy week, because the weeks you skip are exactly the weeks the numbers were trying to warn you about. Fifteen minutes weekly beats ninety minutes monthly, because the point is catching drift early rather than producing a thorough report.

Split the rhythm by what actually changes at that frequency. Weekly is for the leading indicators: stalled matters, unbilled work, unconverted intakes, deadline density in the coming fortnight. Monthly is for the lagging pairs and for trust reconciliation, though the required reconciliation frequency and format vary by jurisdiction across US states, the UK, Canada and Australia, so confirm your own regulator's rule rather than assuming a monthly cycle is universally sufficient. Quarterly is for the structural questions: practice area mix, referral concentration, capacity.

Every number needs an owner and a threshold

An unowned metric is admired and then ignored. Assign each number to one named person, not to a role and not to a committee, and make the expectation explicit: when this number crosses this line, you are the one who does something. The action does not have to be dramatic. Often it is a message to a fee earner asking what is blocking a matter, or a decision to bill early this cycle rather than waiting.

Thresholds matter as much as ownership, because most people are poor at reading trends on a chart and quite good at reacting to a line being crossed. Decide in advance what counts as a problem. More than a set number of matters stalled beyond your stage-age limit. Unbilled work above a figure your firm picks based on its own cost base. Any trust ledger where a correction was needed twice in a month. Numbers without thresholds generate discussion. Numbers with thresholds generate action, which is the entire point.

Where the numbers come from decides whether they stay honest

A dashboard assembled by hand each week is a dashboard with a countdown timer on it. Somebody has to export, paste, reconcile the mismatches, and explain why one figure disagrees with another. That work is done enthusiastically for a month and grudgingly for two more, and then it stops. The numbers should come out of the system where the work already happens, as a by-product of normal practice rather than as a separate reporting chore.

That is also what keeps the numbers trustworthy under pressure. Trust balances are only dashboard-worthy if they cannot silently drift, which is why enforcement at the database transaction level matters more than a warning dialog somebody can click through. In Casely, a disbursement exceeding a matter's actual trust balance is blocked outright rather than flagged, corrections are voided and stay visible rather than being deleted, and each matter carries its own isolated ledger. A number produced by a system that structurally cannot go wrong in that direction is a number you can put in front of partners without a caveat.

Metrics that look useful and quietly are not

Some numbers survive on dashboards for years without ever changing a decision, and they are worth naming so you can cut them. Raw hours logged firm-wide is the most common. It moves with headcount and season, it flatters a busy month that produced nothing collectible, and no one has ever changed a decision because it went up. Total matters open is another, since it conflates a genuinely active file with one that has been dormant for eight months.

Averages across the whole firm are the third trap. Average matter duration, average invoice value, average days to payment. The average is calm while the distribution is on fire, and the cases you needed to see are the outliers the average absorbs. Where you can, display the count above a threshold instead of the average: the number of invoices older than sixty days is a more useful figure than the mean age of invoices, because it names the specific pile someone can go and work through this afternoon.

Build the first version in an afternoon, then cull it in a quarter

Do not design the perfect dashboard. Pick three pairs, six numbers total, from decisions you already make. Give each an owner and a threshold. Put the weekly fifteen minutes in the calendar as a recurring commitment and hold it even in the week where nothing looks interesting, because the discipline is what you are building rather than the insight. Resist every request to add a panel for the first month.

Then, at the end of a quarter, do the part almost nobody does: go back and ask which numbers ever caused anyone to do anything. Cut the ones that did not. A dashboard that shrinks after its first quarter is a dashboard that is being used, because usage is the only thing that reveals which numbers were genuinely load-bearing and which merely looked serious.

What a working dashboard actually feels like

The test is not whether the dashboard is comprehensive, or attractive, or impressive to show a prospective lateral hire. The test is whether, in the last four weeks, somebody looked at it and then did something different because of what they saw. If the answer is no, the honest response is not to redesign it. It is to cut it down to the two or three numbers that would genuinely have changed a decision, and start again from there with a cadence somebody owns.

Firms tend to overestimate how much information they need and underestimate how much rhythm they need. Six numbers reviewed reliably every week will beat forty numbers reviewed enthusiastically in January and never again, and it is not close. The value was never in the reporting. It was in the fact that a stalled matter gets noticed in week three rather than week eleven, that unbilled work gets invoiced this cycle rather than next, and that somebody in the firm knows those things are their job.

The infrastructure question follows from that rather than leading it. Once you know your six numbers, the practical requirement is simply that they emerge from the system where the work is already recorded, without an export, a spreadsheet, or a Friday afternoon of reconciliation. If you want to see how that looks when matter stages, unbilled time, trust balances and deadlines all report from the same place, Casely's reporting and analytics is built for exactly that, and the Free plan is enough to test whether your six numbers hold up before you commit anything at all.