From Traditional BI to Modern AI: How Small and Medium-Sized Companies Strategically Use Data

Executive Summary

Traditional business intelligence solutions are limited to static dashboards and predefined metrics – an approach that holds modern companies back. Elena Alfonsi and Bianca Frost from Gemma Analytics demonstrate in the podcast "Das Gelbe vom AI" how the transition to modern data architecture and Artificial Intelligence delivers concrete business results – from predictive models with 88% accuracy to 5% revenue growth. The key point: technology alone is not enough. Success only emerges when companies simultaneously transform their data culture.

People

• Elena Alfonsi
• Bianca Frost
• Michael (Moderator)

Topics

• Business Intelligence
• Data Architecture
• Machine Learning
• Data Culture
• Change Management
• Women in Tech

Clarus Lead

The transition from legacy BI to modern data platforms is no longer just a technical question for small and medium-sized enterprises – it's a cultural one. In the podcast interview, Gemma Analytics reports on projects where companies increased operational efficiency by measurable percentages through intelligent predictive models. Yet the central insight is: where technology and human-oriented change do not go hand in hand, even the best systems fail.

Clarus Original Analysis

• Clarus Research: Systematic differentiation between static KPI dashboards (reporting function) and proactive insight generators. Concrete case examples: predictive models in property management achieve 88% accuracy in inspection distribution; churn forecasts for SaaS providers measurably reduce errors of rule-based predecessors.

• Classification: The greatest risk is "data theater" – performative analytics without genuine business integration. Opportunities emerge through iterative collaboration between data teams and business stakeholders; MVP approach instead of perfectionist big-bang rollout reduces change management resistance.

• Consequence: Small and medium-sized companies must launch two measures in parallel: (1) modernize infrastructure (automated pipelines, data quality checks, cloud-native stacks), (2) train leadership and teams in genuine data-driven decision making – otherwise they invest in dead systems.

Detailed Summary

The Shift from Static to Intelligent Reporting

The traditional BI model is defined by passivity: predefined dashboards, manual reports, dependency on analysts for additional questions. Data is understood as a pure reporting function, isolated from operational processes. The result: slow response times, faulty Excel exports, data sources manually pieced together multiple times.

Modern analytics reverses this approach. The focus shifts from What happened? to What will happen? and What should we do?. Intelligent predictive models, segmentation algorithms, and optimization systems are embedded directly in business workflows – not in isolated dashboards.

An example from property management illustrates the difference: A company conducting inspections across 100+ regions used a predictive model for staff allocation. With 88% accuracy over a forecast horizon of 3-6 months, the operations team could proactively allocate resources – no longer reacting to inspection backlogs.

Data Infrastructure as Foundation, Culture as Success Factor

However, modernization regularly fails when approached purely technically. A typical scenario: a company builds modern data architecture with automated pipelines but discovers that its leadership and teams don't use the new systems – or worse, misinterpret them.

Elena Alfonsi and Bianca Frost introduce the concept of "data theater": these are beautiful, technically sophisticated dashboards that have no real influence on business decisions. They are ornament, not tool.

Real success emerges through parallel movement:

1. Technical: Solid foundations with BI fundamentals (data quality, standardized modeling, governance), automated pipelines, dashboards that serve a clear business decision.

2. Cultural: Teams understand why data matters. Business stakeholders have visibility into operational metrics. Feedback loops are built in.

A real case: With a customer using manual reporting processes, Gemma Analytics first set up a modern pipeline – this reduced daily export work by hours. In the second step, they integrated insights directly into CRM and marketing planning. Result: 5% revenue growth in the first year. But not because of the dashboard alone – but because executives actually made different decisions when they could work daily with fresh, trustworthy data.

Change Management: From Resistance to Shared Goals

The most typical mistakes in migration are not technical, but human:

• Timing conflicts: Management doesn't understand why secure migration takes 8 weeks, not 2.
• Loss of trust: Sales teams trust Excel forecasts more than machine learning models – even if these are more precise.
• Unclear ownership: Nobody knows who "owns" which KPI and therefore can't decide quickly.

The solution strategy is simple, but demanding:

1. Transparent communication: Why does this component take longer? What are the quality assurance steps?
2. Stakeholder involvement: Co-creation instead of top-down rollout. MVPs with quick feedback instead of perfect final release.
3. Leadership alignment: If management doesn't decide data-driven themselves, it signals to teams: "Data is optional."

In a churn prediction project, a procurement manager discovered that the model didn't correctly account for seasonal promotions – feedback that directly led to better forecasts. The success was not the model, but the collaboration.

Talent Development and Data Culture in Mid-Market

A surprisingly large obstacle: German mid-market companies find qualified data professionals – but lose them again because the culture isn't right.

Successful teams are characterized by the following features:

• Clear, open communication: Ability to ask questions and address challenges without surprises.
• Shared goals: Not "my ticket vs. your ticket," but collective ownership of outcomes.
• Learning time as safe space: Regular protected time to experiment with new tools and methods. This often brings operational innovations.
• Proximity to business: Analysts who understand how the company works are 10 times more valuable than pure technicians.

Particularly with women in tech, it becomes clear: it's not about "lowering perfection requirements" (that would seem sexist), but about psychological safety. Women who dare to ask, experiment, and speak critically – in teams with supportive leadership – often show higher career ambitions and retention rates.

Key Takeaways

• Modern BI is not optional: Static dashboards are too slow for fast-moving markets. Predictive models and self-service analytics have become standard.

• Data quality is the foundation: Even the best AI produces errors if input data is unreliable. BI fundamentals (governance, modeling, QA) are not "optional."

• Technology + Culture = Success: Hardware and software are means, not ends. Success is measured by actual business decisions that change – not by dashboard aesthetics.

• Migration requires a plan and patience: Parallel operation, QA checkpoints, stakeholder approval, and change management take time. Shortcuts lead to technical debt.

• Leadership exemplifies data-drivenness: If management doesn't itself decide with data insights, the entire data team is degraded to performance theater.

• Learning and safety are retention factors: Especially in tech talent development, those who protect explicit learning time and offer psychological safety win.

Stakeholders & Affected Parties

Opportunities & Risks

Actionable Relevance

For Executive Leadership

1. Clarify strategic question: Is data-drivenness a competitive factor for us? If yes, commit to it publicly – in meetings, objectives, budget.
2. Set concrete business goals (not "we need better dashboards"): e.g., "15% churn reduction through better customer segmentation" or "reduce inventory by 10%."
3. Budget resources realistically: Modern BI with cultural change takes 6–18 months, not 3.
4. Work with data yourself: Leadership must visibly decide with dashboards – otherwise you signal that it's optional.

Indicators to watch: Are you actively using the dashboard the first week after go-live? Do your teams ask questions about it, or do they keep using old Excel files?

For Data Teams

1. Build business proximity: Don't just deliver dashboards, understand your stakeholders' operational processes.
2. Communicate MVP strategy: Explain why iteration is better than perfectionism; early wins build trust.
3. Make QA processes transparent: Why does it take time? Show checklists, validation steps, not just code.
4. Protect learning time: Regularly block time for experimentation – this is both retention and innovation.

Indicators to watch: Do stakeholders use dashboards independently? Or do they call you every time? Do teams ask