Good KPI examples for an SME are gross margin per customer, revenue against budget, days sales outstanding, on-time delivery, absence rate and conversion per channel: each with a fixed definition, a target and an owner. Those last three parts are what separate a KPI from a number. A number without a target is a reading, not a signal. A number without an owner is a report everyone reads and nobody picks up. And a number without a fixed definition is the start of an argument that gets repeated every month.
That is why the examples below are not a loose list, but always in the same form: name, formula, source, target. First what a KPI is and is not, then the examples per department, and then how to set your own in five steps.
What is a KPI, and what is not?
A KPI is a measurable figure that shows whether you are on course towards a goal. The K stands for key: the few figures a decision hangs on, not everything you can count. Number of orders, number of website visitors and number of hours booked are metrics. They only become a KPI once someone does something differently because of them.
That distinction is not theory. It decides how many figures belong on a screen. An organisation has hundreds of metrics and should have three to five KPIs per role, five to eight per dashboard. More is not extra control, it is less: when everything is important, nothing is.
SMART, the familiar test for specific, measurable, achievable, realistic and time-bound, helps to phrase a KPI sharply. But SMART says nothing about the question that comes before it: which decision should this figure improve? A KPI can be neatly SMART and still move nobody to act.
KPI examples per department
The CFO wants to see earlier where margin is leaking. The COO wants to know which step in operations is holding up the work. The CEO wants to know which growth actually turns a profit. Three roles, three steering questions, and so three different KPI sets drawn from largely the same data. These are the examples that come up most often in practice, per department, always with formula, source and a common target.
Finance (in more depth in our piece on the financial dashboard):
- Gross margin in percent. Revenue minus cost of goods, divided by revenue. Source: accounting. Target per product group, because an average hides exactly the group that is losing money.
- Revenue against budget. Realised revenue as a percentage of the budget for the same period. Source: accounting plus the budget file. Target: 100%, with a bandwidth of 5%.
- Days sales outstanding (DSO). Outstanding receivables divided by the period’s revenue, times the number of days. Source: receivables ledger. Target: the payment term plus ten days.
Sales (see sales dashboard):
- Quote-to-order conversion. Quotes won divided by quotes issued, per salesperson and per segment. Source: CRM.
- Pipeline coverage. Weighted pipeline value divided by the revenue target for the coming quarter. Source: CRM. Target: three times the goal.
- Average order size. Revenue divided by number of orders. Source: ERP. If it rises, you are selling better; if it falls at the same revenue, you are working harder for the same result.
Operations and logistics:
- On-time delivery. Percentage of orders delivered complete and on the agreed day. Source: ERP or warehouse system. Target: 95% or higher.
- Lead time from order to delivery. Average number of days between order date and delivery date. Source: ERP.
- Days of inventory. Inventory value divided by the daily cost of goods sold. Source: ERP.
HR (see HR dashboard):
- Absence rate. Sick days as a percentage of available working days, weighted for part-time, following the definition used by Statistics Netherlands (CBS). Source: HR system. Count in headcount instead of FTE and you get a different number, which is exactly why the definition has to be fixed.
- Turnover. Leavers divided by average headcount, per year. Source: HR system.
Marketing (see marketing dashboard):
- Cost per qualified lead. Marketing spend divided by the number of leads sales has accepted, per channel. Source: ad platforms plus CRM.
- Conversion per channel. Leads divided by visitors, per source. Source: web analytics plus CRM.
Every line has the same four parts. If one is missing, it is not yet a KPI but a candidate.
How do you set KPIs? Five steps
The list above is a starting point, not a shopping list. Which KPIs it should be in your case follows from the steps below, and the first step is the one most projects skip.
- Start with the decision, not the data. Which decision has to be faster, better or sharper? Raise prices or not, hire an extra planner or not, supply a customer on credit or not. Without that question you pick KPIs based on what happens to be measurable.
- Pick one result figure and one or two predictors per decision. For “do we need another planner”, on-time delivery is the result, and planned hours against available capacity for the next four weeks is the predictor.
- Write down the definition: formula, source, filters. Including or excluding VAT, with or without intercompany revenue, order date or delivery date as the reference date. Write it down, in one place, and have it signed off by the owner from step 4.
- Set the target and the owner. A target is a goal value with a bandwidth, so that not every wobble is an alarm. The owner is one person who corrects course when the figure falls outside the band. A department cannot be an owner.
- Agree the rhythm and automate. Daily, weekly or monthly, from which source, and who looks when. As long as someone still has to load an export by hand, that becomes the weakest link. Review every quarter whether each KPI still hangs on a decision, and drop what is no longer used.
Skip step 3 and you will notice within a month. Then every department turns up to the meeting with its own revenue figure, and what you have is not a KPI but a pub quiz.
Leading and lagging KPIs: looking ahead instead of back
Almost all the KPIs organisations track by default are lagging: revenue, margin, absence, customer satisfaction. They measure the result once it is settled. Useful to know, but a KPI that only changes once the month is over shows you only what you can no longer fix. That is accountability, not steering.
You steer on the figures that can still change next month’s result. Those are the leading KPIs, and they are almost never on the first dashboard:
- Quotes without follow-up after seven days. Predicts next month’s conversion.
- Pipeline coverage for the coming quarter. Predicts whether the revenue target is reachable before the quarter starts.
- Planned hours against available capacity, four weeks ahead. Predicts delivery problems before the first customer calls.
- Invoices outstanding for more than thirty days. Predicts the cash position six weeks out.
- Short-term absence per team over the past eight weeks. Predicts long-term absence.
A good KPI set has something of both: the result to know whether it is working, the predictor to still be able to do something about it. Only lagging KPIs is the rear-view mirror. Only leading KPIs is steering without knowing where you are.
What is a KPI tree?
A KPI tree is the way to tie those two kinds together. At the top is the goal, and each layer below splits that goal into the figures that determine it, until you arrive at something someone can actually turn.
For a wholesaler it looks like this. Profit is revenue times margin. Revenue is number of orders times average order size. Number of orders is quotes issued times conversion. Quotes issued is the number of leads times the percentage followed up within two days. So at the bottom of that tree sits a figure like “follow-up within two days”, owned by sales, and at the top sits profit, owned by the board. Everything in between shows why a deviation at the top arises.
Anyone looking for a KPI framework will come across names like OKR, the Balanced Scorecard or OGSM. They help to order goals and spread them across perspectives. They do not pick the figures for you. The tree does, because it forces you to name, layer by layer, what causes the figure above it. If you cannot fill in a layer, you do not yet know how your business works, and discovering that is more useful than which framework you use.
KPI dashboard example: what belongs on screen per KPI
A KPI on a dashboard is more than a big number. Per KPI your screen should show four things: the current value, the target, the gap between them and the trend over recent periods. Power BI partly enforces that: a KPI visual requires a base measure, a target value and a threshold, otherwise it cannot turn green or red. Anyone who has not fixed a target sees that come back at this point in the build.
If you also want to record the owner and status per KPI, goals in Power BI are built for that: a scorecard with the target value, the progress and who is responsible, per KPI. What such a screen looks like as a whole, with tiles at the top and the explanatory charts underneath, is in our piece on Power BI dashboard development. The dashboard is the last step. The four before it decide whether it is worth anything.
Pitfalls when setting KPIs
Five things go wrong most often, and all five can be prevented up front:
- Too many KPIs. Thirty figures on a screen means nobody knows which three matter.
- Measuring what is easy to measure. Visitor numbers are one click away, margin per customer takes an afternoon of calculation. The second figure hangs on a decision, the first does not.
- A target without a bandwidth. Then every week is an alarm, and after three weeks nobody looks anymore.
- No owner, or a department as owner. A figure everyone is responsible for, nobody picks up.
- Never reviewing. KPIs chosen two years ago measure the business of two years ago.
One more, which does not fit the list because it comes before it: if you do not yet have three decisions you would like to take differently every week, a BI project is too early. Start with a list of five KPIs in Excel, with the definition, the target and the owner for each. If that list works after a quarter, that is the moment to automate it. If it does not work, a dashboard would not have fixed that either.
Does every KPI in your organisation hang on a decision, or on a chart that happened to be available? Want to know which five to eight KPIs make the difference for your organisation, and what it takes to refresh them automatically? Get in touch and we will work out the KPI tree together.
Frequently asked questions
What is a KPI?
A KPI (key performance indicator) is a measurable figure that shows whether you are on course towards a goal, and that someone acts on when you are not. What separates it from an ordinary number is three things: a fixed definition, a target and an owner.
What is the difference between a KPI and an ordinary metric?
A metric is anything you can count: number of orders, number of visitors, number of hours booked. A KPI is the small selection from those that a decision hangs on. An organisation has hundreds of metrics and should have no more than a handful of KPIs per role.
How many KPIs should you have?
Three to five per role, five to eight per dashboard. More KPIs does not mean more control but less: when everything is important, nothing is. Pick one result figure per decision and one or two figures that predict that result.
How do you set a KPI?
Start with the decision that has to get better, pick one or two figures for it, write down the definition with formula and source, set a target and an owner, and agree how often the figure refreshes. Then review every quarter whether the KPI still does what it was meant to do.
What is a SMART KPI?
A KPI that is specific, measurable, achievable, realistic and time-bound. SMART helps you phrase a KPI sharply, but says nothing about whether the figure hangs on a decision. A KPI can be perfectly SMART and still move nobody to act.
What is a KPI tree?
A KPI tree is a cause-and-effect diagram that breaks a goal at the top down into the figures that determine it. Profit splits into revenue and margin, revenue into number of orders and order size, orders into quotes and conversion. At the bottom of the tree are the figures you can actually turn.
What is the difference between a leading and a lagging KPI?
A lagging KPI measures the result after the fact, such as revenue or absence over the past month. A leading KPI measures something that predicts that result and that you can still influence now, such as pipeline coverage or the number of quotes without follow-up. You steer on the leading KPIs.
What does a KPI dashboard example look like?
For each KPI the screen shows the current value, the target, the gap between them and the trend over recent periods. Five to eight of those tiles at the top, with the charts that explain why a figure deviates underneath. What does not belong on it: every figure you happen to be able to measure.
