Data strategy and data governance

Use a maturity model as a map

Building step by step is safe, but too slow when competitors leap ahead. The art is knowing where to climb and where you can jump.

· · 3 min read

Every organisation that starts with AI runs into the same dilemma. Technological change works best step by step: get the culture and the technical foundation in order, then take the next step. Be too cautious, though, and competitors who build on AI from the start may overtake you.

Faisal Hoque and Paul Scade argue that you don't have to choose. You can do both at once, as long as you know exactly where.

Five maturity levels

Their model has five levels. At the first level, initial, there are isolated experiments without a strategy. Next come the first repeatable applications (developing) and an AI strategy aligned with business goals, with mature data management and risk frameworks (defined). At the fourth level AI is fully integrated and performance is measured (managed). The fifth level is a culture of continuous innovation in which people and AI work together as a matter of course (optimising).

The authors refer to research among 500 listed companies. It showed that technology projects matched to the organisation's maturity level are an important predictor of financial success. Moving too fast wastes money.

Use the model as a diagnosis

Hoque and Scade's main point: a whole organisation cannot skip levels, but a single department or application sometimes can. A company with little experience in predictive analytics might, for example, deploy a generative AI application for knowledge management straight away in one department. Their diagnosis starts from the end goal:

  1. Identify which AI applications deliver the most business value.
  2. For each application, map what blocks adoption, from the current to the desired situation.
  3. Decide whether those blockers can be removed quickly and affordably.

Leaping works best when there is little legacy, a strong mandate with concentrated resources, a clear scope, a team that is ready and real strategic urgency. With deeply interwoven systems, processes where a mistake does serious damage, complex regulation or an organisation that isn't ready yet, patience is the fastest route.

Eight questions before you scale

The business case for AI only appears at scale. Didier Bonnet and Achim Plueckebaum offer a checklist you should already work through at the start of a pilot:

  • Does the scaled application contribute materially to the strategic goals?
  • Does it solve a recurring problem rather than a one-off question?
  • Are there resources and budget to scale?
  • Is there a scalable technical foundation with shared standards?
  • Are the data sources of sufficient quality and well governed in the long term?
  • Can we safeguard security, compliance and ethics at scale?
  • Do we have the skills and the organisation to roll it out widely?
  • Is there a business owner with budget responsibility for the result?

That is why I prefer to see a maturity assessment as a map rather than a report card. It shows where you need to lay the foundation first and where you can speed up today. Curious where your organisation stands? The data maturity check on this site gives you a first picture.

Sources

  1. Hoque, F. & Scade, P. (2026). Strategic leapfrogging: Moving past AI maturity models with diagnostic precision. I by IMD, 18 February 2026.
  2. Bonnet, D. & Plueckebaum, A. (2025). How to scale your AI use cases: A checklist. I by IMD, 27 November 2025.

Read next

Let's talk

Would you like to exchange ideas on data, analytics, AI or FP&A? I would be glad to connect.