What a Good KPI Actually Looks Like
Every business I work with has metrics. They track revenue. They track expenses. They have some version of a monthly report that lists numbers and, in many cases, a chart or two. What most of them do not have is a set of KPIs that actually function as key performance indicators in the way that phrase is supposed to mean.
The gap between having metrics and having useful KPIs is significant, and it is almost entirely a design problem. The numbers exist. The issue is that they are not structured, presented, or maintained in a way that connects them to the decisions the owner makes on a regular basis.
Tied to a decision
The first test of a useful KPI is whether it connects to a specific action the owner takes regularly. Revenue is a number. Revenue by product line compared to last month is a KPI, because it tells the owner whether to adjust pricing, shift marketing spend, or investigate a drop in a specific category.
The distinction matters because many metrics look useful in theory but do not connect to anything the owner actually decides. Website traffic is a number. If the owner does not control marketing spend or content strategy, it is not actionable for them. Employee count is a number. If the owner is not actively making hiring decisions this quarter, it is context, not a KPI.
When I build a dashboard for a client, the first conversation is about decisions, not data. What do you decide weekly? What do you decide monthly? What information would change how you make those decisions? The answers to those questions define the KPI set. Everything else is reference data that can live in a supporting tab.
Updated at a maintainable cadence
A KPI that is three months stale is worse than no KPI at all, because it creates a false sense of visibility. The owner glances at the dashboard, sees numbers, and assumes they reflect reality when they do not.
The cadence has to match two things: the decision rhythm and the effort required to update. If the owner makes pricing decisions monthly, the margin KPI needs to be current at least monthly. If the data entry required to update it takes two hours and the person responsible for it is already overloaded, the KPI will fall behind within a quarter.
This is why I design inputs to be as lightweight as possible. If a KPI requires the owner to manually compile data from four sources, it will not survive contact with a busy month. If it pulls from data that is already being entered for other reasons, like sales transactions or payroll, the maintenance cost is close to zero and the KPI stays current.
Compared to something
A number in isolation has limited value. Revenue of $42,000 last month means nothing without context. Is that up or down? Compared to what? Against budget? Against the same month last year? Against a trailing average?
The comparison is what turns a data point into information. A KPI should always display alongside its reference point: budget, prior period, target, or trend. Without that context, the owner has to remember what normal looks like, and memory is unreliable when you are making decisions under time pressure.
This is one of the simplest improvements I make to existing reporting tools. The data is there, but it is presented as a single number. Adding a variance column, a percentage change, or a simple conditional format that flags when the number is outside the expected range transforms the dashboard from a data display into a decision support tool.
Owned by one person
The most overlooked attribute of a useful KPI is ownership. Someone has to be responsible for the number, not in the sense that they control the outcome, but in the sense that they are responsible for updating it, investigating anomalies, and raising the flag when the metric moves outside the expected range.
When nobody owns a KPI, it degrades quietly. The update gets skipped during a busy week. An anomaly goes uninvestigated because everyone assumed someone else would look into it. By the time anyone notices, the data is stale and the dashboard has lost credibility.
In a small business, ownership usually maps to the person closest to the data source. The bookkeeper owns the financial KPIs because they are entering the transactions. The operations manager owns the labor and efficiency KPIs because they are managing the schedule. The owner reviews the dashboard, but the maintenance responsibility sits with the person who touches the data.
The four-part test
Before I add any metric to a client's dashboard, I run it through these four criteria. Does it connect to a decision the owner or manager makes regularly? Can it be updated at a cadence someone will actually maintain? Does it display with a comparison point? Is one person responsible for keeping it current?
If the metric fails any of those tests, it does not belong on the primary dashboard. It might be useful reference data, and it can live in a supporting tab for deeper analysis. But the main view, the one the owner looks at every week, should contain only metrics that pass all four. Five KPIs that meet these criteria will do more for a business than 40 metrics that do not.
