BI transformation in finance: the road SMEs actually travel

26 July 2026 · 8 min read · Manfred Schönberger

BI initiatives are usually discussed as a tooling question: Power BI or Tableau, cloud or on-premises. In the mandates I run, success is decided somewhere else entirely. I analysed four widely watched expert discussions from the German-speaking BI community and combined them with current figures on the Swiss SME landscape. The pattern is unambiguous.

What BI transformation means in finance

Business intelligence describes the path from raw data to a basis for decisions: data is pulled from source systems, consolidated, checked and prepared so that a management team can work with it. The word transformation belongs in the phrase because the way the finance department works changes, rarely just its tooling.

For mid-sized companies the figures show a clear pull in this direction. According to the SME Labour Market Study published by AXA Switzerland in October 2025, 32 per cent of Swiss SMEs use artificial intelligence for data analysis, up from 22 per cent a year earlier. Over the same period the share of companies deliberately integrating AI into their working processes rose from 22 to 34 per cent.

The most common mistake starts with the tooling question

One sequence stands out across the discussions I reviewed. Dennis Hoffstädte, founder of DatenPioniere, has guided German mid-sized companies through Power BI rollouts for more than a decade. In his conversation with Andreas Wiener on the BI or DIE channel he deliberately places the tooling question behind objectives and delivery model – his chapter list starts with the target group, moves through goals to the delivery model, and covers software after that.

The vendor side reports the same finding. Jens Horstmann, board member at TREVISTO AG, names four symptoms of a grown spreadsheet landscape in the announcement to his late-2022 podcast: rising manual effort, multiplying errors, results that are "often no longer traceable", and maintenance that "depends on the employee who originally built the spreadsheet". Buying a licence removes none of the four.

Four maturity levels

To place a company I work with four levels. They are my own working model drawn from mandates, and they make a useful map because each level carries its own next task.

Most SMEs that approach me sit between levels 1 and 2. The move that pays off most takes them to level 3, and it begins with unglamorous work: defining which figure comes from where.

An approach that works in mid-sized companies

Hoffstädte stresses that the approach for a mid-sized company differs from the one used in a group, and describes a planning and information workshop as the first step. That sequence matches my experience. Four phases have proven themselves:

1. Agree the target picture and the metrics. Which ten to fifteen figures steer the business? That list is drawn up with management, not in the IT department.
2. Establish where the data comes from. One source, one definition and one accountable person per metric. This step regularly reveals that two departments calculate the same term differently.
3. Build and launch a first cockpit. A contained scope that runs after a few weeks, rather than a complete reporting suite after a year.
4. Extend and hand over. Connect further areas, transfer ownership to the team.

Why self-service often disappoints

One point from the Hoffstädte discussion deserves particular attention because it contradicts a widespread expectation: training and self-service do not always work. The idea that business departments will build their own reports after a one-day course rarely survives contact with practice.

Maximilian Laturnus describes self-service reporting from the user side at Carl Zeiss Vision, presenting it as a path with defined steps. That is precisely the condition. Self-service works once a verified data model sits underneath and the metrics are defined. Without that foundation, multiple versions of the truth simply migrate into new tools – the same situation as before, at higher licence cost.

The author's assessment