From data strategy to AI maturity - what companies really need before they get started

In this interview, Dr. Gillian Kant, Senior Data Scientist at Finanz-DATA GmbH and based at Georg-August-Universität Göttingen, provides insights into the key steps on the path from a well-thought-out data strategy to company-wide AI maturity. He talks about specific practical examples, typical challenges and success factors for data-driven organizations.

Many companies want to 'do something with AI' - but often lack a solid data strategy. Why is this so crucial before even thinking about models?

Before thinking about AI models, it must be clear what data is available, how it is collected, structured and secured. Without uniform definitions and clean data sources, inconsistencies and incorrect conclusions will arise. A data strategy defines which data is relevant for which use case, who is responsible for quality and management and how data protection and compliance are ensured. This prevents isolated solutions, saves effort in data preparation and creates the basis for scalable AI projects.

In your opinion, what are the most important building blocks on the path to AI maturity - technically, organizationally and culturally?

I think on a technical level, you need a flexible infrastructure: scalable computing resources and automated pipelines for data processing. Organizationally, clear responsibilities are crucial. An agile project methodology ensures fast feedback cycles. Culturally, on the other hand, employees must become data literate: Training in "data literacy", open exchange about successes and failures as well as motivation for data-driven improvements. I believe that technology, organization and culture together form the framework on which an AI-enabled corporate culture can grow.

How can the maturity level of a company be realistically assessed - and which tools or frameworks help to determine the status quo?

The maturity level can be determined through analyses with frameworks such as the "Data & Analytics Maturity Model" from Deloitte. This framework evaluates dimensions such as data quality, technology stack, organizational processes and culture along defined levels from "ad hoc" to "optimized".

What would you say: When is a company 'ready' for AI - and how can you tell?

I think a company is ready for AI when it has a clear data strategy, validated use cases and initial PoCs.

Such a company operates automated ETL pipelines, uses a central platform for analyses and deployment and has an interdisciplinary team that translates business issues into technical solutions. Established data guidelines, approved budgets for AI projects and demonstrable increases in efficiency or revenue as a result of data science initiatives are crucial.

If you look back in three years - how would you recognize that a company has started its AI journey correctly?

When we look back in 2028, we will recognize successful AI journeys by the fact that AI-supported processes are part of everyday operations. Companies that implemented their data strategy early on will have high data quality and end-to-end MLOps pipelines. A practical example could be a predictive maintenance solution for our AWS deployments that reduces unplanned outages by a certain percentage. Such successes document that the AI strategy has not only been launched, but has been properly thought through.