Before investing in AI capabilities, assess your organization's readiness across eight critical dimensions. This framework helps analytics leaders identify gaps, prioritize investments, and set realistic expectations for AI-driven transformation. Use it to baseline your current state, understand where AI will accelerate value, and clarify where foundational work must happen first.
How to use this assessment
This framework works best as a team exercise. Assemble analytics leaders, architects, data governance owners, and IT stakeholders. For each of the eight dimensions below, read through the level descriptions (1, 3, and 5) and score your organization on a scale of 1 to 5. You do not have to score every intermediate level—use 2 and 4 to fine-tune between levels.
Be honest about your current state
This is not a compliance exercise or a report card. The goal is clarity. If your data governance is poor, score it low. If your semantic layer is fragmented, say so. Understanding gaps is the first step to closing them.
Use this as a baseline, not a destination
Your score today is a snapshot. It will change. Use this assessment at the start of an AI initiative to establish a baseline, then re-assess quarterly or semi-annually to track progress. The goal is movement and intentionality, not perfection from day one.
Different dimensions have different urgency
Some gaps are blockers—you cannot proceed without fixing them. Others are risks—you can proceed but with mitigation. The interpretation section below will help you understand which gaps are which.
Use dimension scores to prioritize the roadmap
Once you have scored all eight dimensions, identify your lowest scores. Those are your readiness gaps. Build a remediation plan to lift those dimensions. In parallel, start AI pilots in areas where you already have strength—usually around data quality, semantic layer maturity, or strong executive sponsorship.
The eight assessment dimensions
Dimension 1: Data Governance Maturity
Data governance defines who owns data, how it is defined, and how changes are managed. Immature governance creates inconsistency. AI amplifies inconsistency at scale.
Level 1: No formal governance
Data definitions are created ad hoc. Teams define terms differently. There are no formal stewards or domains. Change management is informal. Masterdata (customers, products, accounts) is maintained by multiple teams with no coordination. When a business term is used—"revenue," "active customer," "tier"—different reports may use different logic. No single source of truth exists.
Level 3: Partial governance
Some business domains have recognized stewards (finance owns GL hierarchy, sales owns customer hierarchy). Data definitions exist for core entities and KPIs, but they are not enforced uniformly. SAP Datasphere may be deployed, but not all analytics models use it—many SAC models and reports still import raw data. Governance processes exist for critical master data but are inconsistently applied.
Level 5: Full governance
All data domains have named stewards. Datasphere spaces are governed with clear ownership and change control. KPI definitions are centralized, versioned, and documented. All analytics consumption flows through Datasphere or a well-maintained semantic layer. Changes to master data, hierarchies, or definitions follow a formal process. Data is treated as an asset with clear lineage and accountability.
Why this matters for AI: AI answers are only as good as the semantic layer they draw on. If your business terms are inconsistent or undefined, every AI query will produce inconsistent results. If your master data is not maintained, AI will inherit bad data and propagate it. Strong governance is the foundation for trustworthy AI.
Dimension 2: Semantic Layer Completeness
The semantic layer is the business-facing abstraction between raw data and analytics. It defines relationships, attributes, hierarchies, and calculated fields. AI queries work against the semantic layer.
Level 1: No semantic layer
Analytics are built on raw tables. SAC models import data directly from SAP ERP, data warehouse, or lake without a business logic layer in between. Business logic is embedded in each report, query, or model. Adding a new report means reimplementing business logic. No standard definitions exist.
Level 3: Partial semantic layer
Datasphere has been deployed and some business domains have curated models. Core entities like customers, products, and GL accounts are modeled. However, many SAC models still use direct imports to add custom logic or accelerate reporting. Some analytics teams build against the semantic layer; others circumvent it. Coverage is perhaps 40-60% of analytics consumption.
Level 5: Complete semantic layer
All analytics consumption goes through Datasphere models or an equivalent business layer. Raw tables are not accessible to BI tools. The semantic layer includes all required entities, relationships, hierarchies, calculated fields, and KPI definitions. Governance processes control changes to the layer. Analytics teams query the layer without exception.
Why this matters for AI: Joule, just ask, and other conversational AI features only work reliably against well-modeled semantic definitions—just ask answers only from the models administrators add to it, and SAP provides a model readiness check for that step. If analytics teams circumvent the semantic layer or use raw tables, AI has no reliable business context and produces unreliable results. A complete semantic layer is required for scaled AI.
Dimension 3: Data Quality and Freshness
Data quality refers to accuracy, consistency, and completeness. Freshness refers to how current data is. Both directly affect AI output quality.
Level 1: Poor quality, no monitoring
Data quality issues are known but not systematized. Missing values, duplicates, or inconsistencies occur and are caught by users or during analysis, not by systems. Master data is incomplete or contains duplicates. Reference data drifts over time. No data quality monitoring or alerting. Freshness varies—some data updates daily, some monthly, and timing is not enforced. Reports often require manual corrections before they are trusted.
Level 3: Partial quality and freshness
Major entities and transactions are clean. Core master data (customers, products, GL accounts) is governed and maintained. However, edge cases and secondary dimensions may have quality issues. Timeliness is mostly met—most data is within SLA, but some exceptions occur. Basic data quality monitoring exists for critical tables but not comprehensive. Some manual data cleanup is still required.
Level 5: High quality, enforced freshness
Data quality rules are enforced in pipelines. Duplicates, missing values, and inconsistencies are caught before data enters the warehouse. Master data matching is systematic. Freshness SLAs are met consistently (e.g., GL updates daily by 7 AM, transactions by 2 hours, forecasts by noon). Data lineage is tracked. Quality metrics are monitored and reported. Failed quality checks trigger alerts and workflow to remediate quickly.
Why this matters for AI: AI amplifies both good and bad data quality. A model trained on poor data will produce poor results consistently, and users will lose trust quickly. An AI system that answers questions using stale data creates business errors. Investing in data quality before AI deployment is one of the highest-ROI foundational activities.
Dimension 4: User Adoption and Analytics Culture
Analytics culture is how comfortable users are exploring data independently and how engaged they are in data-driven decision-making.
Level 1: IT-driven, low self-serve adoption
Users request reports from IT or business analysts. Business users rarely access data directly. Most users are passive report consumers. Self-service tools exist but are underused. There is limited feedback from users about what data they need. Questions from business go to analysts, who spend time building reports rather than doing analysis. User feedback loops are weak.
Level 3: Power users drive adoption
A small group of power users actively creates reports and dashboards in SAC. Most other users consume those static dashboards. Self-service is growing but not universal. Business users are becoming more comfortable with data but still rely on intermediaries for complex analysis. Feedback on data needs is informal and sporadic.
Level 5: Self-serve culture, strong engagement
The majority of users are comfortable exploring data independently. Self-service is the norm, not the exception. Users ask questions and expect answers without going through analysts. Analytics is seen as a core business competency, not an IT function. Feedback loops are active—users suggest improvements, report issues, and drive requirements. Data literacy training is routine. Communities of practice help users learn and share best practices.
Why this matters for AI: AI features like Just Ask, narrative insights, and anomaly alerts only deliver value if users are engaged and willing to act on insights. In an IT-driven environment where users passively consume reports, AI features will be ignored. Adoption requires a culture where users are comfortable asking questions and exploring data—and where feedback on AI quality can be collected and acted on.
Dimension 5: Technical Infrastructure
Technical infrastructure is the cloud stack, data platform, and AI services available to support analytics and AI workloads.
Level 1: On-premise only
Analytics are built on BusinessObjects or legacy BI tools on-premise. SAC is not deployed. SAP Datasphere is not in use. There is no cloud analytics platform. Cloud connectivity is limited or nonexistent. SAP's cloud AI—Joule, Joule Studio agents, the generative AI hub—is out of reach. Migration to cloud is planned but not underway.
Level 3: Partial cloud deployment
SAC is deployed and used widely. Datasphere is in progress or partially deployed. Some workloads have migrated to SAP Business Data Cloud (BDC) or other cloud platforms, but many remain on-premise. AI Core may be licensed but not yet activated. Generative AI hub access is planned but not fully configured. Cloud architecture is being built incrementally.
Level 5: Full cloud stack
Complete SAC + Datasphere + BDC deployment. Analytics pipelines are cloud-native. SAP AI Core is available and integrated with analytics workloads. The generative AI hub is configured with the large language models you need, Joule is active in SAC and Datasphere, and agents are registered in SAP AI Agent Hub. Infrastructure supports real-time analytics, batch processing, and AI inference at scale. Cloud-native development patterns are in place.
Why this matters for AI: SAC's AI features—just ask, Smart Insights, Smart Predict, and the AI-assisted features—run in SAC itself, and some AI-assisted features are Premium AI that consumes AI Units. Joule, Joule Studio agents, and the generative AI hub run on SAP's cloud platform, SAP BTP with SAP AI Core. Analytics that stay on-premise can reach that AI only through integrations you build and govern yourself. The cloud stack is the foundation for scaled AI.
Dimension 6: AI Skills and Team Readiness
AI initiatives require specific skills: prompt engineering, model evaluation, semantic layer design, MLOps, and change management. Teams without these skills will struggle.
Level 1: No AI expertise
The analytics team has no formal AI training. Prompt engineering is unknown. Model evaluation capabilities do not exist. There is no one designated to own AI workflows or govern AI output quality. The team relies entirely on vendors or consultants for AI work, with no internal knowledge building. There is no plan to hire or train AI skills.
Level 3: Some knowledge, limited hands-on
A few team members have completed AI training or worked with AI concepts in other roles. There is basic understanding of how LLMs work and what prompt engineering is. However, there is limited hands-on experience. No one has evaluated or tuned models in production. Semantic layer design for AI is not a competency yet. There is interest but not capacity.
Level 5: Dedicated AI capability
At least one—ideally more—senior team member is dedicated to AI workflows. The team has hands-on experience with prompt engineering, model evaluation, and output quality assessment. Semantic layer curators understand how to design for AI queries. MLOps disciplines are in place: model monitoring, retraining, governance. Change management skills exist to help users adopt AI features. The team can evaluate whether an AI solution is working and adjust it based on feedback.
Why this matters for AI: Someone must own the quality and governance of AI output. Without dedicated skills, AI systems will be deployed without proper monitoring, models will degrade without retraining, and users will lose trust. Skill investment is not optional—it is the difference between AI that delivers value and AI that creates frustration.
Dimension 7: Executive Sponsorship and Budget
Executive sponsorship means clear accountability, prioritization in the roadmap, and dedicated funding. Without it, AI projects are deprioritized when urgent work arises.
Level 1: No executive awareness
AI in analytics is not on the executive agenda. There is no identified sponsor. No budget is allocated specifically for AI initiatives. AI is seen as "nice to have" or a future consideration. When IT or analytics suggests an AI project, it competes with all other work and gets deferred. There is no strategic roadmap that includes AI.
Level 3: Interest, no formal backing
Executives are aware of AI and see potential value. There is discussion about AI in analytics, and interest in pilots. However, no formal executive sponsor is identified. Budget for AI is embedded in general IT budgets, not allocated specifically. AI projects are approved on a case-by-case basis, making them vulnerable to reprioritization. Strategic roadmap mentions AI but does not commit to it.
Level 5: Executive champion, funded roadmap
An executive (VP analytics, CFO, CIO, CDO) is the identified sponsor for AI in analytics. This person champions AI in strategy conversations and allocates budget. There is a multi-year roadmap that includes specific AI initiatives. Budget is protected and does not get consumed by urgent work. Success metrics are defined and reviewed quarterly. The sponsor removes blockers and escalates impediments. AI is treated as a strategic priority, not a discretionary project.
Why this matters for AI: AI initiatives without executive backing stall at the pilot stage. When an urgent business need or cost pressure emerges, resources get diverted. The dedicated team gets reassigned. The pilot never scales. With executive sponsorship, pilots get the resources to expand, learnings get documented, and patterns get reused. Sponsorship is the difference between one-off pilots and sustainable transformation.
Dimension 8: Change Management Readiness
AI changes workflows. Users need clear communication about what is changing, why, how to use new tools, and how to give feedback. Without structured change management, adoption suffers.
Level 1: Ad hoc communication only
When changes happen, communication is informal and sporadic. Training is minimal or nonexistent. There is no adoption tracking or measurement. Users discover changes incidentally and learn by trial and error. There is no structured feedback channel. Resistance is not addressed. Failures and workarounds are not documented or remediated.
Level 3: Communication and informal training
Communication about new features happens in meetings and emails. Training is offered informally—sessions are available but attendance is optional and not tracked. Documentation exists but may be incomplete. Some feedback is collected, but it is not systematically analyzed or acted on. Adoption is monitored informally. There is no formal change board or governance.
Level 5: Structured change management
Changes are managed through a formal process. Communication plans are created for each major change, with messaging tailored to different audiences. Training is mandatory for affected users, with curriculum designed for different roles. Adoption is tracked through usage metrics and feedback. A change board reviews impact and removes barriers. Feedback loops are active and acted upon quickly. Success metrics are defined and measured. Communities of practice support peer learning.
Why this matters for AI: AI changes how people work. Just Ask changes how analysts write queries. Narrative insights change how reports are written. Predictive planning changes forecasting methodology. Users need support to adopt these changes and confidence that the AI is trustworthy. Structured change management reduces resistance, accelerates adoption, and gives the AI system time to prove value.
Interpreting your score
Add up your dimension scores (each 1-5). Your total will be between 8 and 40. Use these bands to interpret your organizational readiness:
Score 8-16: Foundation building needed
Readiness level: Not ready for scaled AI
What this means: Your organization has significant foundational gaps. Data governance is weak or absent. Your semantic layer is fragmented. Data quality issues are known but not managed. The culture is not ready for self-service analytics, let alone AI-driven workflows.
What to do: Before investing in AI, focus on foundational work. First, establish data governance and assign stewards. Second, build or complete your semantic layer. Third, implement data quality monitoring and remediation. These investments can take a year or more, but they are prerequisites for trustworthy AI. In parallel, start building analytics culture—train users on self-serve tools, establish feedback loops, demonstrate quick wins.
AI strategy: Delay major AI investments until you reach a score of at least 24. Use this time to pilot small AI features (anomaly detection, basic narrative) in low-risk domains to build internal capability. When your foundation is stronger, AI will scale faster.
Score 17-28: Selective AI pilots appropriate
Readiness level: Ready for targeted pilots
What this means: You have partial capabilities in most areas but gaps in others. Your semantic layer is incomplete but progressing. Data quality is good for core entities but inconsistent in edge cases. You have some analytics adoption but not universal self-serve. You have interest but not formal executive sponsorship.
What to do: Identify your lowest-scoring dimensions (usually governance, semantic layer, or skills). Build a 6-12 month plan to improve those. In parallel, pilot AI in areas where you are already strong. Start with Just Ask queries against your best-governed semantic models. Build narrative insights for routine reports. Pick pilots where semantic definitions are mature and data quality is high. These pilots will build internal confidence and skills.
AI strategy: Run 2-3 targeted AI pilots simultaneously. Design pilots to test both technical feasibility and organizational adoption. Measure impact carefully. Use learnings to refine your roadmap. Hire or train AI skills during the pilot phase so you are ready to scale. Build executive sponsorship through pilot success stories. Target a score of 30+ within 18 months.
Score 29-36: Ready for scaled AI
Readiness level: Ready to expand AI broadly
What this means: You have strong foundations across most dimensions. Data governance is established. Your semantic layer is mostly complete. Data quality is generally good. You have an analytics culture and some user adoption. Executive interest exists, even if formal sponsorship is not yet fully activated. Skills are developing.
What to do: Scale AI across the analytics roadmap. Expand successful pilots to broader audiences. Build AI into standard workflows: make narrative generation standard for all SAC models, integrate Just Ask into core analytics processes, add predictive insights to planning workflows. Invest in MLOps and governance infrastructure now so you can monitor quality at scale. Formalize executive sponsorship and secure 2-year budget for AI expansion.
AI strategy: Move from pilots to production systems. Build centers of excellence around key AI use cases. Establish AI governance boards. Create standard templates and patterns for AI workflows so teams can adopt them quickly. Start building custom models and predictive use cases. Set a target for the share of analytics workflows that use AI, measured against today's baseline, and review it quarterly.
Score 37-40: AI-native operations
Readiness level: Ready for advanced and autonomous AI
What this means: You have mature capabilities across all eight dimensions. Strong governance, complete semantic layer, high data quality. Self-serve analytics culture is the norm. Full cloud infrastructure is in place. Dedicated AI skills exist. Executive sponsorship is active. Change management is structured. You are positioned to use AI not just to enhance workflows but to fundamentally change how analytics operates.
What to do: Push into advanced AI scenarios. Build autonomous decision systems and workflow automation. Implement predictive planning at scale. Create AI agents that can answer complex questions without human intervention. Explore generative use cases: AI that writes entire reports, builds models, or designs system changes. Establish thought leadership—share your AI practices and patterns externally. Use your maturity to attract talent and drive competitive advantage.
AI strategy: Your competitive advantage is no longer in basic AI adoption but in how deeply you have embedded AI into analytics operations. Focus on edge cases and complex scenarios that competitors have not tackled. Build partnerships with SAP and consultants to explore emerging AI capabilities. Invest in research—evaluate new models and approaches early. Your focus shifts from "can we do AI?" to "what is the highest-value AI we can build?"
Recommended next steps by score range
If you scored 8-16: Your first 12 months
Months 1-3: Governance foundation Establish a data governance model. Identify data stewards for core business domains (finance, sales, supply chain). Document current data definitions. Create a governance charter. Start a governance council to meet monthly. Hire or designate a Chief Data Officer if you do not have one.
Months 3-6: Semantic layer plan Assess your current semantic layer—what is modeled, what is not. Create a multi-year semantic layer roadmap. If SAP Datasphere is not deployed, plan deployment. If it is partially deployed, create a completion roadmap. Begin by modeling your most-used data domains (GL accounts, customers, products).
Months 6-9: Data quality initiative Implement data quality monitoring for your top 10 tables. Define quality rules and acceptable thresholds. Build monitoring dashboards. Create a process to track and remediate quality issues. Establish data quality SLAs—e.g., GL updates by 7 AM daily, zero duplicates in customer master.
Months 9-12: Analytics culture and skills Launch a self-serve analytics training program. Set up power-user communities. Hire or cross-train 1-2 team members in AI concepts; SAP's Joule Studio developer certification is free until December 31, 2026, and SAP now offers an SAP Business AI Platform solution architect certification. Plan small AI pilots for year 2 (anomaly detection, basic narrative) in domains where you have already made governance progress.
If you scored 17-28: Your first 18 months
Months 1-3: Identify priority gaps Determine which 2-3 dimensions are dragging down your score. For each gap, understand the root cause: Is it budget? Skills? Executive priority? Process gaps? Create a targeted remediation plan for each.
Months 1-6: Parallel foundation work and AI pilots While you improve your weakest dimensions, launch 2-3 AI pilots in areas where you are already strong. Pilots might include: Just Ask queries on mature semantic models, narrative generation for routine reports, or anomaly detection on well-governed master data. Design pilots to be completed in 8-12 weeks so you can gather fast feedback.
Months 6-12: Scale successful pilots Expand successful pilots to broader audiences. If narrative generation pilot succeeded, make it standard for all executive dashboards. If Just Ask worked well, make it the default query method for that domain. Use pilot learnings to refine governance and skill requirements.
Months 12-18: Foundation completion and roadmap Complete your foundation remediation work (semantic layer expansion, governance formalization, data quality improvements). Hire dedicated AI/ML roles or contract with partners. Create a 2-year AI roadmap with specific use cases, investment requirements, and success metrics. Secure executive sponsorship and budget.
If you scored 29-36: Your first 24 months
Months 1-3: Close dimension gaps Identify which dimensions are below 4 and create targeted improvement plans. For example: if executive sponsorship is weak, secure a formal sponsor and budget. If semantic layer is 70% complete, finish it. If AI skills are limited, hire or contract. These are focused, 3-6 month efforts.
Months 3-9: Scale pilots to production Take 2-3 successful AI pilots and operationalize them. Build MLOps infrastructure: monitoring dashboards, retraining pipelines, governance workflows. Create standard governance processes for AI output review. Train larger user groups on new AI-enabled workflows.
Months 6-12: Establish AI centers of excellence Create CoE structure around major use cases (forecasting AI, sales analytics AI, financial AI). Each CoE owns semantic layer quality for its domain, trains users, and iterates on AI output. CoEs become the hub for scaling AI across the organization.
Months 12-24: Expand AI across roadmap Roll out AI to the workflows where pilots proved their value. Begin building custom models for predictive use cases (forecast accuracy improvement, anomaly prediction, churn modeling). Establish an AI governance board to oversee model quality, fairness, and regulatory compliance, including an EU AI Act risk classification for each use case (high-risk obligations apply from December 2, 2027, for stand-alone systems). Plan for multi-year sustained AI investment.
If you scored 37-40: Advanced AI strategy
Months 1-6: Governance and guardrails Establish formal AI governance frameworks: a model registry, an agent and MCP server registry such as SAP AI Agent Hub, bias testing, and compliance documentation. Create AI review boards. Document your AI policies and principles. This becomes your foundation for responsible scaling.
Months 3-12: Custom model development Begin building proprietary models for your highest-value use cases. Areas to explore: predictive forecasting (forecast accuracy improvement), demand prediction, financial anomaly detection, supply chain optimization. Partner with universities, research institutes, or consultants if needed for specialized expertise.
Months 6-18: Automation and autonomous systems Move beyond AI-assisted workflows to automated workflows. Examples: autonomous anomaly response (AI detects and auto-mitigates), autonomous forecast adjustments within guardrails, autonomous report generation and distribution. Humans remain in control through guardrails and thresholds, but routine work is fully automated.
Months 12+: Thought leadership and competitive moat Share your AI practices externally: speak at conferences, publish case studies, mentor other organizations. Build relationships with SAP product teams to influence roadmap. Explore emerging models and techniques early. Your competitive advantage is continuous innovation in AI-driven analytics.
Where next
This assessment is a starting point. Once you understand your organizational readiness, explore these related topics to deepen your strategy:
Comprehensive guide to AI capabilities across SAP Analytics Cloud, Datasphere, and BTP. Start here to understand what is possible and what each capability requires.
Real-world use cases: forecast accuracy improvement, sales rep coaching, anomaly detection, demand planning. See how other organizations have implemented AI and what they achieved.
How to evaluate, prioritize, and prove ROI for AI in SAP analytics. Build the business case for your pilots and scaling roadmap based on measurement frameworks and investment models.
Templates, tools, and curated resources for AI implementation. Find readiness worksheets, governance templates, pilot playbooks, and links to SAP training and certification.