← All case studies
Case studies · Public sector

From ambition to a decided path

How a Swedish municipality became AI-ready.

The municipal leadership had a clear ambition: the organisation would become AI-ready and make more decisions on data than on gut feel. What was missing was the path there. In a quarter, Rebtech delivered an agreed data vision, a data strategy across six focus areas and a first-year action plan that can be started with existing resources.

About the engagement

Client
Swedish municipality, anonymised
Industry
Public sector
Engagement
Pre-study: data vision, data strategy, action plan
Duration
Roughly one quarter, two consultants
Roles
Senior advisor, business analyst
Delivery
Decision material for municipal leadership and elected officials
Method
Traditional development
The challenge

The questions nobody owned

The municipality wanted to build its own capabilities around data and AI, not merely use general-purpose tools. Between the ambition and reality lay questions nobody owned. Information ownership was formally assigned but unclear in practice, the requirement for benefit assessment lacked both method and template, and the organisational structure looked different in the finance system, the HR system and the directory services.

The integration landscape was fragmented, with point-to-point solutions and manual file transfer. Without an in-house systems development organisation, access to the underlying data also became a question of contracts and requirements rather than of technology.

The solution

The business question first, the technology second

The engagement was to describe how the municipality becomes an AI-ready organisation with data as a strategic resource, addressed to the municipal leadership and elected officials. The scope was deliberately bounded: no technical architecture, no platform selection, no implementation.

Two principles carried the delivery. Load neutrality meant that every new data-driven task had to replace or simplify something existing. The second principle was to start from the business question rather than from the technology, since being AI-ready turned out to be primarily an organisational matter.

Delivery

Eight in-depth interviews through the whole chain of command

Eight in-depth interviews through the whole chain of command, from the chief executive to the IT architect and business developers, formed the basis of the analysis. What carried most weight were the patterns that recurred independently: what only one person said became an observation, what five described from their separate vantage points became a conclusion.

When information ownership turned out to be the recurring knot, the question was put sharply. Is an information strategy missing, or is the existing one not being followed? The difference determines which recommendation is right, and it was presented as a hypothesis to validate rather than as settled fact.

Every claim could be traced to a named source, and where evidence was missing the gap was written out plainly. Vision and strategy were then developed in workshops with the client's own project group, so that the result was already agreed by the time it reached the leadership.

Results

The foundational questions moved to the start of the plan

The most important effect concerned sequence. The municipality was about to build a data warehouse and analytics capability on top of organisational master data that would not hold. The pre-study moved the foundational questions of information ownership, master data and shared definitions to the start of the action plan, where they belong.

The delivery consisted of a data vision, a data strategy across six focus areas, a first-year action plan with ownership and timing, and concrete candidates for the first delivery areas. The municipality went from an ambition nobody owned to a decided path the organisation can sustain.

In brief
  1. 01

    Data vision and data strategy across six focus areas, agreed in workshops with the client's own project group

  2. 02

    A first-year action plan with ownership and timing, startable with existing resources

  3. 03

    Information ownership, master data and shared definitions moved to the front of the queue

Does this problem sound familiar?

The ambition is there, but the path is missing