Dashboards and data

Management or operational?

A dashboard starts with a question, not a chart. Usually that question comes from management, from operations, or from both.

A management dashboard shows how the business is doing: revenue, margin, outstanding invoices and the KPIs you manage by. The figures come from your accounting system, and one update a day is often enough.

An operational dashboard shows what's happening right now in the operational processes, with data from machines, scanners or your planning system.

You don't have to choose: one dashboard can do both, with an overview for management and screens for operations. And if people need to step in right away, we make it real-time: the dashboard refreshes as soon as something changes, as at Granuband.

Technically, a dashboard is a web application: it runs in the browser, on its own or as part of a larger system.

Data first, then the dashboard

A dashboard is only as good as the data underneath, so we start with the layer that collects and stores your figures. For every number you should be able to see where it comes from, otherwise nobody will trust it.

We pick the database per project. Usually that's SQL, because it's the most cost-effective choice. Sometimes something else fits better: NoSQL for a recommendation engine, for example, or a vector database if you're working with AI.

The data comes from your own sources. How we connect those systems is explained on our integrations page.

Custom or Power BI?

A standard BI tool such as Power BI is enough when your data already sits in widely used systems, you have standard reporting needs and the figures don't have to be live. In that case you don't need a custom dashboard, and we'll tell you so on the first call: a standard tool gets you going faster and at lower cost.

Custom wins when one of these applies:

1. You want to see figures while the work is happening, not afterwards.
2. The data comes from machines, hardware or software you had custom-built.
3. The dashboard is part of your own application or process, and people need to take action in it too.
4. You want to control the data layer yourself: how your data is stored and how every figure is calculated.

Example: a tyre classification machine at Granuband

Around 60 million car tyres a year: that's how many Granuband sorts for sale or recycling. Experienced graders are getting harder to find, so the company bought a classification machine: hardware from Germany, software from another supplier. That machine rejected tyres without anyone being able to see why.

We worked as a technical partner between the operations, management and the suppliers. On site we looked at the machine, the software behind it and the rules used to classify tyres, and we tested at the machine itself. With real inspection data we adjusted and extended the filters. Graders got a screen to correct the machine's judgement, and we showed new versions on an industrial touchscreen, right next to the conveyor belt.

The inspection data now comes together in one data layer. On the dashboard, the team and management follow in real time which tyres are being classified and processed, how the machine is performing and how the processed tyres trend over time. Every classification code stands for an amount in euros, so they can also see straight away what the processed tyres are worth.

What it delivered: far fewer tyres that the machine wrongly failed to classify, insight into how the machine performs and what the investment is worth, and a system the team relies on every day.

How we build a dashboard, step by step

• Introduction and analysis: we start from your questions and your sources. Which figures do you want to see, where are they now, and how clean is that data? We work that out into a plan that shows, phase by phase, which screens you get; that research is on us.
• Design and visual prototyping: we design the screens around your wishes, so you can see how the dashboard works, screen by screen, before any development starts.
• Building in phases: in phase 1 we build the figures you look at every week, on your real data. You and your team check them against what you know from your own records or from day-to-day operations, and that feedback shapes the next phase. After each phase, you decide whether we continue.
• Launch and support: your team works with the dashboard every day, while we keep an eye on things in the background.

The cost depends on the number and type of sources, whether the figures need to be live, how much clean-up work the data needs and how many screens you need.