A KPI Dashboard for SMBs: Which Numbers Actually MatterAKPIDashboardforSMBs:WhichNumbersActuallyMatter

Start with questions, not with data

The usual approach runs backwards: look at what the ERP, shop, and accounting system can export, then turn it into charts. The result is a dashboard that shows a lot and answers little.

The better direction is the reverse. Write down the questions you ask yourself every week anyway. For most SMBs we work with, they are variations of: how is the month running against plan? Who owes us money, and since when? How busy are we next week? What is sitting in the warehouse without moving? Each of these questions yields one metric. Anything that answers no question does not belong on the dashboard.

The metrics that show up in almost every SMB

The exact selection depends on the business; a plumbing company needs different numbers than an online shop. A few candidates return in nearly every project, though, because they connect directly to cash and capacity:

• Revenue for the current month, set against last month or the plan, so the number has a yardstick
• Open invoices by age, because there is a dunning run between “€18,000 outstanding” and “€18,000 outstanding, €12,000 of it older than 30 days”
• Order backlog or utilization for the next two to four weeks, because this is where you see whether sales or delivery is the bottleneck
• Contribution margin instead of plain revenue where the data allows it, since revenue alone also rewards bad orders
• One or two numbers unique to your business: return rate, cancellation rate, material cost per job

What you can safely leave out

Website visitors, social media reach, newsletter open rates: all measurable, all visible in their own tools anyway, and almost never a number that changes what an SMB owner decides on a Tuesday. These belong in the marketing toolbox, not on the steering dashboard.

Leaving things out is hard, because every tile looks interesting on its own. A useful test: if this number drops 20 percent next week, will you do anything? If the honest answer is “no” or “I wouldn’t know what”, it is a number to observe, not a number to steer by.

The argument about definitions matters more than the chart

In many companies, sales and accounting do not mean the same thing by “revenue”. One counts gross order intake, the other invoiced work net of tax, and both are right from where they sit. As long as the number lives in two spreadsheets, this only surfaces in meetings, where people then argue about the number instead of the decision.

A dashboard forces you to settle these definitions once: revenue is invoiced work net of tax, the cut-off is the invoice date, credit notes are deducted. That is uncomfortable work, one or two workshops that can get loud. In our experience, this work is half the value of the whole project. Displaying the number afterwards is almost a formality.

Live does not mean everything has to blink every second

Live data sounds great but costs very different amounts of effort depending on the source system. Honestly assessed, a nightly refresh is enough for most metrics: monthly revenue says the same thing whether it is from 6 a.m. or from just now. Only numbers someone reacts to during the day genuinely need to be current, such as open jobs on a dispatcher’s screen.

This distinction saves money. A dashboard that reads three systems overnight is considerably cheaper to build and run than one that needs real-time connections everywhere. So for every metric we ask: how old can this number get before it hurts you?

Start small: six tiles, then take stock

Our advice for the start: one page, around six metrics, most important questions first. Live with that for four to eight weeks. By then you know from experience which number you check daily, which you ignore, and which is missing. Only then is the second stage worth it, with drilldowns, per-team views, and role-based access.

The opposite route, planning the big dashboard for every department up front, takes months to show a first number and guesses half the requirements wrong. We say this even though the bigger project would earn us more.

And if six tiles feel too modest: according to Statistik Austria’s ICT survey (2023), only one in four Austrian companies has run data analytics at all, in-house or outsourced, against an EU-27 average of 33 percent. Checking six cleanly defined numbers every week already puts you ahead of most.