How well-defined is your organisation's AI & ML strategy? No defined AI & ML strategy Informal discussions underway with no documented strategy or expected return on investment (ROI) Documented strategy exists with understanding of ROI but is not widely communicated Strategy is communicated, owned and tracked against KPIs with clear business cases Strategy is embedded in business planning and reviewed regularly at board level
How aligned is AI & ML investment with measurable business outcomes? Investment is opportunistic with no clear business cases Some initiatives have business cases but outcomes are not tracked Most initiatives are tied to defined outcomes but tracked informally Outcomes are formally tracked and reported per initiative AI & ML ROI is a core input to portfolio prioritisation across the business
How is AI & ML championed at the executive level? No discussion or measurement of AI & ML business value Limited discussion with ad hoc spotlights on individual projects Defined key performance indicators (KPIs) exist for some initiatives and reported to executive level Self-service executive dashboard with consistent KPIs across initiatives, used to inform decision making AI & ML use cases and KPIs are tied to enterprise OKRs and championed by the board
How would you describe the quality of data and pipelines for AI & ML? Data quality is inconsistent and/or pipelines require substantial manual support Some data quality issues remain with minimal pipeline automation and issues typically discovered downstream Moderate data quality with reliable pipelines and engineered features for core use cases Fully automated and self-healing pipelines with proactive data quality monitoring to prevent downstream issues AI & ML use cases and KPIs are tied to enterprise OKRs and championed by the board
How mature is your data governance (i.e., lineage, ownership, cataloguing)? No formal governance processes in place Governance exists in pockets but not fully enforced Policies exist and apply to critical datasets Governance is enforced across the enterprise with clear ownership Automated governance and audit processes which are continuously improved
To what extent is data accessible across business units for AI & ML use cases? Data is fragmented across isolated systems with no integration Point-to-point data sharing integrations exist for a few use cases, with limited documentations Shared data platforms support several business units Most business units consume from a shared, well-governed data platform There is a unified data fabric, facilitating self-service, enterprise data products and monitoring
What best describes your AI & ML platform for model training, deployment and monitoring? No shared platform, mostly or entirely localised (e.g., individual laptops, on-premise servers) Disparate cloud platforms and tools exist but aren't integrated A shared and scalable cloud platform exists, mostly manually configured A managed platform covering most needs for model training, deployment and monitoring A unified platform that seamlessly integrates data management, governance and advanced AI & ML capabilities
How standardised are AI & ML tools and frameworks across teams? Every team picks its own stack with no coordination Some informal conventions exist within teams Recommended tools and frameworks are documented but not enforced A consistent core tech stack and set of frameworks are adopted by most teams Tooling and frameworks are best in class, automated and centrally supported
How is model deployment handled (AIOps & MLOps maturity)? Model development through to deployment is manual or no models make it to production at all Versioned scripts in code repositories with manual deployment for each model Automated deployment for some models with basic/patchy monitoring Standardised deployment pipeline with model monitoring and rollback for most models Fully automated and closed-loop re-training, drift monitoring and explainability checks with full model lineage 
