About This Episode
In episode 4 of Data & AI Heads, host Ian Allison speaks with Preethi Singh, a program and project management leader with deep roots in mainframe programming, banking, retail, and enterprise BI. Preethi introduces the concept of Business Intelligence Experience Management, arguing that technical success in BI migrations and AI initiatives depends less on the technology itself and more on how well teams embed business users throughout the journey. She draws on a real Tableau-to-Power BI migration program where an obvious 70 to 80 percent cost reduction still required rigorous change management, phased milestone planning, and domain-level champion networks to achieve genuine adoption.
Preethi also addresses the current AI landscape with a people-first perspective. She outlines practical guardrails for responsible AI development, including compliance with GDPR and CCPA, domain-organized data architectures that support faster retrieval and marketplace-style flexibility, and a disciplined start-small-then-scale approach to evaluating AI use case ROI before committing to infrastructure investments in the millions or billions of dollars. Her core message is that whether the initiative is a BI platform migration or a large language model deployment, the discipline of customer experience management and phased value realization is what separates projects that land from those that stall.
This episode is essential listening for data leaders, BI program managers, and AI practitioners who need frameworks for driving business adoption, managing stakeholder trust, and building responsible data practices that do not sacrifice speed for safety.
Key Takeaways
- Embedding business users in the discovery and migration planning phases, not just informing them at the end, is the single most effective way to build trust and drive BI adoption.
- A compelling business case alone, such as a 70 to 80 percent cost reduction in a Tableau-to-Power BI migration, is not sufficient to secure buy-in. Users need to feel ownership of the journey itself.
- Phased, time-bound ROI milestones, such as showing measurable cost savings at six months and again at one year, give business stakeholders concrete confidence rather than a distant promise.
- Appointing domain-level pilot champions or SMEs to represent each business division during a BI migration ensures feedback reflects the full range of users, from engineers running daily reports to executives running quarterly board decks.
- Applying agile best practices and conducting deep dives into both IT and business processes can reduce the cycle time to stand up BI infrastructure and applications by 20 to 30 percent.
- Organizing data into domain-specific, customer-specific, and product-specific clusters creates a flexible, marketplace-style architecture that accelerates both data retrieval and future AI integration.
- AI safety is a product feature, not an afterthought. Data sourcing, containment, and sharing practices should be built to comply with CCPA, GDPR, and applicable local regulations from day one.
- Before committing to large AI infrastructure investments, teams should start with small use cases, validate ROI on a phase-by-phase basis, and honestly assess whether a custom-built application might deliver the same outcome at lower cost.
What We Cover
Frequently Asked Questions
Why do BI migrations fail to gain business adoption even when the cost savings are obvious?
According to Preethi Singh, users resist migration not because the business case is weak but because they are comfortable with their existing tools and are not made to feel part of the process. In a Tableau-to-Power BI migration she led, a projected 70 to 80 percent cost reduction was not enough on its own. Adoption improved significantly only after the team included business users in the discovery phase, showed them the migration approach, and provided a detailed, phase-by-phase ROI plan rather than a single end-state promise.
What is the most effective way to manage different stakeholder needs during a BI platform migration?
Preethi Singh recommends appointing pilot champions or SMEs who represent each business division or region. These individuals gather feedback from their entire user base, whether that is engineers running daily operational reports or a CFO running a quarterly board report. This structure ensures the migration team understands the priority requirements for every audience segment without trying to interview every individual user, and it gives domain owners a clear accountability role in the target-state platform.
How should program managers structure ROI communication to keep business stakeholders confident throughout a multi-year data initiative?
Rather than presenting a single multi-year savings figure, Preethi Singh advocates breaking the ROI plan into measurable, time-bound milestones, for example showing a first financial milestone at six months and a second at twelve months. This approach allows business stakeholders to verify that the program is delivering value incrementally rather than waiting years to see results, which sustains confidence and reduces the risk of executive sponsorship eroding mid-program.
What are the key AI safety practices data leaders should implement before deploying AI on customer or operational data?
Preethi Singh recommends treating data safety compliance as a core product feature rather than a governance checkbox. Specifically, teams should verify that data sourcing, storage, and sharing practices conform to CCPA, GDPR, and applicable local regulations at every stage. She also advises organizing data into domain-specific clusters from the start, because well-structured data improves AI model quality and reduces the risk of inadvertently sharing data outside intended boundaries.
How should organizations evaluate whether a new AI use case is worth the infrastructure investment?
Preethi Singh advises organizations to start with small, well-scoped AI use cases and build a phase-by-phase ROI plan before committing to large GPU or CPU infrastructure spend, which can run into millions or billions of dollars. Teams should ask whether the expected return is realizable within six to twelve months, and whether a custom-built application might achieve the same goal at a fraction of the cost. The goal is disciplined due diligence, not suppressing innovation.
How does domain-organized data architecture support both BI performance and AI readiness?
Preethi Singh describes organizing data into domain-specific, customer-specific, and product-specific clusters as a foundational design choice that benefits both BI and AI workloads. This structure enables faster data retrieval, allows business users to introduce new data domains independently in a marketplace-style model, and gives AI models access to higher-quality, well-governed inputs. It also makes it easier to enforce data sharing boundaries required by regulatory frameworks.
What can a project manager do early in a BI initiative to reduce delivery cycle time?
Preethi Singh credits two practices with reducing cycle time by 20 to 30 percent in her programs: implementing agile best practices in product management, and conducting a thorough deep dive into both internal IT processes and business processes before build begins. Understanding these processes together surfaces bottlenecks that slow down application and infrastructure stand-up, allowing teams to remove friction before it compounds later in the delivery cycle.
About the Guest
Preethi
Preethi Singh is a program and project management leader whose career spans mainframe and back-office programming, enterprise banking systems, retail, and large-scale global BI implementations. She began her technology career writing complex reporting modules and account reconciliation programs as a mainframe programmer, an experience that gave her a foundational understanding of business logic, data relationships, and data patterns. She has since led program management engagements across sales, marketing, supply chain, and banking domains, with a focus on BI platform migrations, agile delivery, and change management. To deepen her expertise in management and information systems, she completed an executive CIO program at New York University. She currently serves as an advisory board member at Our Lady of the Lake University, where she specializes in customer experience management, and is an active member of the International Association for Women. Her current role is Chief of Staff.
Connect on LinkedIn →More Episodes