Most finance dashboards are abandoned within two months of being built.

The reason is almost always the same. The dashboard was built to be comprehensive, so it shows forty metrics. Forty metrics means no metric stands out, which means looking at it produces no decision, which means people stop looking at it. It is not that the data was wrong. It is that a dashboard which does not change what anyone does is just a slower version of a report.

I built KPI dashboards covering AP, AR, and inventory in my current role. The design constraint that made them useful was restrictive: every metric on the dashboard has to point to a specific action when it moves. If a number changes and nobody would do anything differently, it does not belong there. It may belong in a report, but not on the thing people are supposed to check.

Why dashboards and not the close

The close tells you what happened. It tells you accurately, in conformity with GAAP, and about three weeks after the period it describes began.

That lag is fine for reporting and useless for operating. An AR problem that starts in the first week of a month is visible in the close about five weeks later, by which time it has had five weeks to compound. The data needed to see it in week one already exists in the ERP. It is simply not surfaced anywhere anyone looks.

That is the gap dashboards fill. Not better reporting, but earlier visibility on the operational side of accounting where the difference between noticing now and noticing at close is measured in real money.

Accounts payable

The question AP dashboards should answer is whether obligations are being processed and paid on the terms you agreed to, without surprises.

Aging by bucket. The current shape of what is owed and for how long. The value is in the trend rather than the snapshot: a bucket growing month over month means something upstream changed.

Invoices in process, by stage and age. Where invoices are sitting right now, and how long they have been there. This is the metric that most reliably produces action, because an invoice stuck in approval for eleven days is a specific problem with a specific owner. Aging tells you that you are late. This tells you why.

Exception volume and age. Matching exceptions outstanding, and how long they have been outstanding. A rising exception count usually means a receiving or purchasing problem rather than an AP problem, which is exactly the kind of thing that is invisible until you count it.

Discounts captured versus available. Early payment discounts you were entitled to and did not take. This is one of the few finance metrics that translates directly into money left on the table, and it tends to get attention for that reason.

Accounts receivable

The question is whether what you have earned is turning into cash on schedule.

DSO, trended. The standard measure, and worth showing as a trend rather than a number. A single DSO figure is not meaningful without the shape of the months around it.

Aging by bucket, with concentration. The aging profile plus how much of the overdue balance sits with the largest few accounts. These call for completely different responses. Broad aging across many accounts is a process problem in invoicing or collections. Concentration in two accounts is a specific conversation with two customers, and possibly a credit decision.

Invoices disputed or on hold. Receivables that are not collectible because something is contested. These are frequently buried in the aging as though they were ordinary slow payments, which makes collections effort go to the wrong place. Separating them changes what the collections team does today.

Cash application backlog. Payments received but not yet applied. A growing backlog makes the aging wrong, which makes everything built on the aging wrong. It is unglamorous and it corrupts the rest of the picture when it slips.

Inventory

The question is whether the inventory you think you have is the inventory you actually have, and whether it is worth what the books say.

Cycle count accuracy. Count results against system quantities, tracked over time and broken out by location or category. This is the single most important inventory metric, because everything downstream, valuation, margin, availability, depends on the system and the shelf agreeing. When accuracy declines in one area, that is where to look.

Inventory value by category and age. Where the money is sitting and how long it has been sitting there. Aging inventory is a valuation question before it is an operations question, and noticing it early gives you more options than noticing it at year end.

Adjustment volume and reason. Inventory adjustments by cause. A rising adjustment volume in a category is almost always a process problem: receiving errors, unrecorded movement, or a systematic issue with how something is counted. The reason codes are what make this actionable, which means they have to be maintained honestly rather than defaulted.

Standard cost variance. Where actual costs diverge from standard. This matters directly for margin accuracy, and in a business selling through multiple channels it matters more, because the same variance lands differently depending on the channel mix.

Design principles that held up

Keep the metric count small. Six per area is a reasonable ceiling. The temptation to add is constant and should be resisted, because every addition dilutes the attention available for the rest.

Show ranges, not just values. A number without context requires the reader to know what normal looks like. Showing the expected range moves the reader's attention straight to the exceptions, which is the entire point.

Make it drillable. A metric that shows a problem but not which transactions constitute it forces the reader to go build a query. Most will not, and the dashboard becomes a place where problems are announced rather than solved.

Refresh it often enough to be current, and say when it refreshed. Stale data that looks live is worse than data clearly labeled as of yesterday, because people act on it without discounting it.

Use the same definitions as the close. If the dashboard's DSO does not match the reported DSO, you will spend the rest of the dashboard's life explaining the difference, and its credibility will never recover.

The part that matters most

A dashboard is only useful if someone looks at it on a schedule and something happens as a result.

Attach it to an existing routine. A weekly review where the exceptions are discussed and assigned is enough. Without that, even a well designed dashboard becomes a thing that exists, and the operational problems it was built to surface go right back to being discovered during the close.

The dashboards themselves are the easy part. The habit around them is the part that determines whether they were worth building.