Prioritizing data initiatives, from foundational data engineering work to advanced AI/ML use cases, is a significant challenge for enterprise businesses. With limited resources and budgets, companies need to focus on projects that maximize business impact, align with strategic goals, and have a high chance of success.
Several frameworks and approaches can guide prioritization. Below, I’ll outline a general framework and considerations for prioritizing across Data Foundation/Migration/Transformation, Data Analytics, BI, and AI/ML. This framework is adaptable and scalable across various organizations, but it requires tailoring to each enterprise’s goals, resources, and maturity level in data and analytics.
Framework for Prioritization
A holistic framework that factors in business impact, feasibility, strategic alignment, and data readiness is highly effective. Here’s a structured, step-by-step approach:
1. Define Business Objectives and Data Strategy
- Purpose: Aligning data initiatives with core business goals ensures relevance. This includes objectives like revenue growth, cost reduction, customer satisfaction, and operational efficiency.
- Considerations: Start with high-level strategic objectives and identify how data and AI can support them. For instance, if the objective is to increase customer retention, both foundational data (like unified customer data) and analytics (like customer segmentation) can be critical.
2. Categorize Projects by Domain and Maturity Level
- Domains: Separate use cases into categories such as Data Foundation (Migration, Transformation), Data Analytics & BI, and Advanced AI/ML. This categorization helps avoid prioritizing advanced AI/ML before foundational data issues are addressed.
- Maturity Level: Assess each domain’s current maturity within the organization. For instance, some enterprises may still need a strong data foundation, while others are ready to focus on AI/ML use cases.
3. Assess Impact, Feasibility, Data Readiness
- Impact (Value to Business): Rank projects based on their potential impact. Impact can include revenue generation, cost savings, risk reduction, or strategic enablement.
- Feasibility (Technical & Resource Feasibility): Assess each project based on technical requirements, data availability, resource allocation, and timeline.
- Data Readiness: Some use cases, particularly AI/ML, may require extensive data, model training, or data transformation. Assess if the foundational data is ready or if additional data work is required.
4. Evaluate ROI and Time-to-Value
- ROI (Return on Investment): Calculate a rough ROI for each project, considering both tangible and intangible benefits. For instance, BI dashboards may have quicker returns compared to more complex AI use cases.
- Time-to-Value: Projects that provide quick wins help build momentum and show stakeholders the value of data initiatives. Start with projects that require less time and yield faster results.
5. Prioritize Based on Business and Technical Dependencies
- Dependency Mapping: Many advanced projects depend on foundational data readiness. For example, AI/ML use cases often require high-quality, well-structured data. Migration and foundational data projects may be prerequisites for these use cases.
- Sequential Prioritization: Start with foundational data projects, followed by analytics and BI, and then move toward AI/ML projects. This progression builds the foundation necessary for more advanced analytics and AI.
6. Risk and Change Management
- Risk Assessment: Evaluate potential risks associated with each project. Migration and transformation projects may come with higher risks if they involve core systems, whereas BI projects might have relatively lower risks.
- Change Management: Consider the level of change management needed. For instance, AI projects that introduce predictive analytics into decision-making might require more user training and change management than BI reporting tools.
List of Criteria:
| Criteria | Key Considerations |
| Business Objectives | Align use cases with enterprise-wide goals like revenue growth, operational efficiency, customer satisfaction, or cost savings. |
| Project Category | Classify into Data Foundation, Data Analytics, BI, and AI/ML. Ensure foundational data is prioritized before advanced use cases. |
| Impact & Value | Rank projects by potential business impact, like revenue generation, cost reduction, and strategic enablement. |
| Feasibility | Assess technical, resource, and data feasibility. Check if needed data is available, and gauge technical complexity. |
| ROI & Time-to-Value | Estimate ROI based on potential returns and timeline. Shorter time-to-value projects can act as quick wins. |
| Risk Assessment | Identify risks such as system downtime, data migration errors, or user adoption hurdles. Projects with low risk may be prioritized for initial wins. |
| Dependency Mapping | Map dependencies (e.g., foundational data needed for AI/ML). Prioritize foundational and dependent projects first. |
Example Prioritization in Practice
- Data Foundation / Migration / Transformation
- Use Case: Migrate on-premise data to a cloud environment for scalable access and analytics.
- Impact: High, as it enables all future analytics and AI/ML initiatives.
- Feasibility: Moderate to high, depending on legacy systems.
- Dependencies: Essential for advanced analytics and BI/AI.
- Priority: High due to its foundational role in enabling other projects.
- Business Intelligence (BI) / Data Analytics
- Use Case: Develop a sales performance dashboard for real-time monitoring.
- Impact: Medium, as it empowers immediate decision-making.
- Feasibility: High, assuming foundational data is already migrated and transformed.
- Dependencies: Low, but enhanced with foundational data in place.
- Priority: Medium to High as it provides a quick win with visible business impact.
- Advanced AI/ML Use Cases
- Use Case: Predictive maintenance for manufacturing equipment to reduce downtime.
- Impact: High, with potential cost savings and efficiency gains.
- Feasibility: Moderate to high, dependent on historical data availability.
- Dependencies: Requires clean, transformed data and may depend on IoT integrations.
- Priority: Low to Medium initially but could move higher once foundational and analytics components are established.
Additional Industry Frameworks for Reference
- RICE (Reach, Impact, Confidence, Effort): Typically used in product development, RICE can be adapted for data projects to weigh the reach (how many users benefit), impact, confidence in success, and effort involved.
Credit: https://www.product-frameworks.com/Rice-Prioritization.html
- DICE (Data, Impact, Complexity, Effort) Framework: A commonly used method for assessing the prioritization of data projects based on four factors—Data readiness, Impact, Complexity, and Effort.
- MoSCoW (Must-have, Should-have, Could-have, Won’t-have): MoSCoW is a simple prioritization tool often used in Agile projects to rank features or projects by necessity, which can work well for data project prioritization.
Credit: https://workflowy.com/systems/moscow-method/
Final Recommendations for Prioritizing Data Projects in Enterprises
- Establish a Data Governance and Prioritization Committee: Include stakeholders from various departments (IT, data science, business units) to ensure alignment.
- Start with Foundational Data Projects: Lay a strong data foundation before tackling analytics and AI. Migrating to scalable, unified data platforms can enable more complex projects.
- Balance Quick Wins with Long-Term Initiatives: Choose a mix of high-impact but feasible projects (e.g., BI dashboards) to show results quickly, while laying the groundwork for complex AI initiatives.
- Iterate and Reassess Regularly: Priorities can change as business needs evolve. Reassess the prioritization every quarter or as major strategic shifts occur.
By following this structured prioritization framework, enterprises can focus on the right projects at the right time, maximizing the impact of their data initiatives and ensuring alignment with broader strategic goals. This approach also builds a data-first culture by prioritizing foundational data needs, which is essential for the success of future AI and ML initiatives.