Most SME owners don't have a data problem. They have a problem with data being everywhere while the decision has to be made on Monday morning. A management report doesn't have to be a six-month project — but it does have to answer the right questions.
Here's a scene we see regularly: on Friday afternoon the owner or the head bookkeeper exports sales from Comarch Optima into Excel, pastes in numbers from a sheet where the sales reps track discounts, and spends two or three hours assembling a report that's already stale by Monday. Sometimes it's simply wrong — because someone dragged a formula down one row too few.
This isn't a tooling problem. It's a problem of the numbers you base decisions on being produced by hand, once a week, in one person's head. It can be fixed — and it costs less than you think.
The Friday Excel costs more than you think
At one of our clients — a 30-person wholesaler of technical parts — the owner spent about 3 hours every Friday building a sales report from Comarch Optima exports. That's more than 12 hours a month of his own time spent retyping numbers. The data was "as of Friday" anyway, so on Monday he was making decisions on a picture from before the weekend.
The worse part was something else. When we built him a single automated dashboard that calculated margin after discounts and shipping costs, it turned out one of his main product groups had a real margin of 4%, not the assumed ~18%. For years he'd sold it convinced it was a good product. The manual Excel never showed this, because discounts and shipping lived in a different sheet that nobody pulled into the report.
That's the real cost of the Friday Excel: not three hours a week, but decisions made on numbers that look solid and aren't.
You don't need a data warehouse or expensive BI
The moment "reporting" comes up, someone usually pitches a data warehouse, a Premium BI platform and a multi-month rollout. For a 10–50 person company that's almost always overkill — you're buying a jet engine to drive to the corner shop.
In SME reality you need far less: one trustworthy data source, a handful of well-calculated metrics, and a dashboard that refreshes itself. The tool can be free or nearly free (Power BI in its basic tier, Looker Studio, sometimes a well-built sheet that pulls data through an API). The value isn't in the tool. It's in having the numbers calculated once, correctly, and always the same way.
Expensive BI on top of messy data gives you exactly what Excel does — just prettier and pricier.
First 5–7 numbers, not fifty
The most common mistake with a first dashboard is trying to show everything. You end up with a board of forty charts that nobody looks at, because it's unclear what matters.
Start from the question, not the data: which 5–7 numbers actually change your decisions this week? For most companies that's roughly:
- sales year-to-date vs. the same period last year,
- margin after discounts and costs (not just sales value),
- overdue receivables — who and how much,
- key inventory levels,
- the quote funnel: how many went out, how many closed.
The rest is trivia. If a number doesn't change a decision, it doesn't belong on the management dashboard — at most it goes into a detailed report you open once a quarter.
The data needs a single source
Here's the theme that runs through almost every project we do: data quality. A dashboard is only as good as the data flowing into it. If the same customer appears in the ERP under three names, and the reps keep discounts in a private sheet, no chart will fix that — it'll just neatly sum up the mess.
Before you build a dashboard, decide what the source of truth is for each number. Sales and margin — from the ERP. Quote status — from the CRM or a sheet, but from one of them. Shipping costs — from one agreed place. If a number is produced today by manually stitching two files together, that's the first candidate to sort out before you connect any chart at all.
It's boring work. But it's what decides whether the dashboard becomes a decision tool or just another pretty screen nobody trusts.
How we usually build it
In practice the project is simpler than you'd expect:
- A conversation about decisions. We agree on those 5–7 numbers and exactly how they should be calculated (e.g. margin "after everything", not just purchase cost).
- Connecting to the source. Most ERPs — Comarch Optima, enova365, Subiekt — expose data via an API or the database. Instead of a manual export to Excel, the dashboard pulls it itself.
- One place to do the math. The rules (discounts, costs, exchange rates) are written once, in one place, so everyone sees the same margin.
- Automatic refresh. The dashboard updates every night or every hour. Nobody "builds the report" — the report just is.
The first version usually takes 2–4 weeks, not six months. We start with one area (most often sales and margin), because it's better to have one number everyone trusts than forty that nobody checks.
Three traps that ruin dashboards
- Vanity metrics. Numbers that look good but change nothing (e.g. "website visits" with no link to sales). They take up space and distract.
- Margin calculated the lazy way. If you show sales value instead of margin after discounts and costs, the dashboard can lie to you just as effectively as Excel — see the wholesaler example.
- A dashboard nobody opens. If after a month nobody looks at it, that's a sign it shows the wrong numbers. Better to go back to the conversation about decisions than to pile on more charts.
Where to start
Don't start with the tool. Start with a sheet of paper where you write down the 5–7 numbers that actually change your decisions, and next to each one note where it really comes from today. That exercise alone usually reveals where data is being stitched together by hand and where the risk is hiding.
If you want to claw back those dozen-plus hours a month and stop making decisions on "Friday data", talk to someone who has connected dashboards to Optima, enova or Subiekt and knows where the mines usually lie. The first call is free and takes half an hour.
A management dashboard is only as good as the data flowing into it. First a single source of truth, then the charts.