Interactive dashboards: what management actually needs
A dashboard nobody opens costs more than none at all: it consumed effort and creates the impression that the topic is handled. The difference between a used and a forgotten dashboard rarely lies in the tool. This article describes how to build a drill-down that a management team actually uses.
What drill-down means
Drill-down is the ability to move from an aggregated figure step by step into its components. Group revenue leads to the region, the region to the product group, the product group to the individual customer. What matters is that this path follows the way users think, rarely the structure of the database.
Management asks questions in a fixed order: does the overall picture hold? Where does it deviate? What is behind the deviation? A dashboard that answers those three questions in three clicks gets used.
Three levels are usually enough
| Level | Content | Audience |
|---|---|---|
| 1 Overview | Five to seven metrics with a traffic light against budget and prior year | Management, board |
| 2 Structure | Breakdown by region, product group, customer segment | Division heads |
| 3 Individual case | Document line, single customer, single order | Controlling, business units |
Restricting level 1 is the hardest decision in the whole build. Every department would like to see its metric at the top. Give way and after three rounds you have forty tiles, and the variance that matters is lost among them.
Why self-service alone falls short
The expectation that business departments will build their own analyses after a training course often fails in practice. Dennis Hoffstädte of DatenPioniere addresses this openly in his conversation on the BI or DIE channel: training and self-service do not always work. His comparison with a fast-food chain points to standardised, repeatable processes rather than individual bespoke work.
Maximilian Laturnus describes self-service reporting at Carl Zeiss Vision as a path with defined steps. That is precisely the condition: self-service works on a verified data model with defined metrics. Without that foundation, competing analyses appear and the discussion shifts from the substance to the data quality.
Six design rules
- One message per view. If you can summarise a page in one sentence, you have cut it correctly.
- Variance before absolute value. The figure alone informs; only the comparison against budget or prior year prompts action.
- Use colour sparingly. Colour marks deviation. A colourful dashboard loses exactly that signalling power.
- Consistent scales. Two charts side by side with different axes reliably produce wrong conclusions.
- Show the data as-of time. The date and time of the last refresh belong on every page. They answer the most common follow-up question in advance.
- Check it on a phone. Management looks while travelling. A view that only works on two monitors rarely gets opened.
The author's assessment
The most common design error is confusing a dashboard with a report. A report documents completely and serves traceability. A dashboard directs attention and serves steering. Trying to fit both into one view produces a tool that does neither job.
On adoption: a dashboard establishes itself once it has a fixed place in a recurring meeting. As long as the management meeting runs on printed slides, the dashboard stays peripheral. The most effective step is therefore organisational: the monthly review is held on the dashboard, and the slides go away.
On tooling I stay pragmatic. For Swiss SMEs Power BI is usually the obvious choice because Microsoft 365 is already in place and the licensing hurdle is correspondingly low. The design rules above apply to any tool.
The short version
- Three levels suffice: overview, structure, individual case. Level 1 carries five to seven metrics.
- The drill-down follows how users think: does the picture hold? Where does it deviate? What is behind it?
- Self-service only works on a verified data model with defined metrics.
- Six rules, among them: variance before absolute value, colour sparingly, refresh time visible.
- A dashboard establishes itself once it holds a fixed place in a recurring meeting.
Read on
Sources
This article draws on the metadata and full descriptions of the YouTube contributions listed below, together with the chapter markers and quotation highlights the channels published themselves. YouTube no longer releases verbatim transcripts, so statements are attributed to their originator and marked as that person's position.
- AI or DIE (BI or DIE) with Dennis Hoffstädte, DatenPioniere: BI in mid-sized companies (in German), 4 Apr 2023, 33:56 min.
- AI or DIE (BI or DIE) with Maximilian Laturnus, Carl Zeiss Vision: Self-service reporting in practice (in German), 29 Nov 2022, 4:16 min.
- BI or DIE: BI for everyone – why BI makes sense for mid-sized companies too (in German), 2022, — min.
- KIundTECH Podcast with Jens Horstmann, TREVISTO AG: Controlling with Power BI: automated reports without Excel (in German), 18 Jul 2023, 35:26 min.
Does your management team use its dashboard?
If the answer hesitates, the tool is rarely the reason. Let us look at what it depends on.
Book an initial consultation