Your AI Is Starving (and Data Products Feed It)
Most organisations still assume data lakes and harmonisation will solve fragmentation, yet AI initiatives continue to stall. The real issue is usability, not storage. This article introduces data products as reusable, governed datasets that power scale. It explains how Customer 360 and financial data products enable consistent, AI-ready foundations for 2026.
Originally published in SUBSTACK
For the longest time, I thought a data lake was the answer to all our data fragmentation problems. I don’t blame anyone. We have all been sold this idea by vendors. Every organisation I talk to has some data harmonisation project going on. Either in the form of a formal data lake or more informal, smaller data sets.
The problem is…despite all of this, we continue to see AI pilots stalling and the tech stack becoming more complex by the day.
The deeper problem is not the harmonisation of data. It is the usability of that data. We’re treating data as raw material, not a ready‑to‑use product.
Data Products: The Missing Link in AI Scale
Consider a data lake: a repository of structured and unstructured data without business context. An analyst must query, join, clean, and calculate for each use case — repeating the work for every dashboard or report. This inefficiency explodes with complex AI demands, especially when we are moving into an era of AI agents working autonomously.
The fix isn’t more data storage. It’s building data products — reusable, pre‑packaged datasets with defined interfaces, context, and governance. As I argue in Reimagine Finance, this is one of three core foundations for the new digital operating model.
What Is a Data Product?
A data product is a complete, reusable dataset ready for multiple use cases without transformation. Instead of accessing and transforming data for every new use case, finance, sales, operations, AI engineers, all use the predefined data products for their different use cases. It includes:
- Defined scope: what it covers (e.g., Customer 360, Financial Reporting 360).
- Interfaces: standardised access for analytics/AI.
- Governance: ownership, freshness, SLAs
Why Imperative for 2026
Fragmented data kills scale. Agents and multi‑models need consistent feeds. One Customer 360 product serves marketing campaigns, sales forecasting, AR predictions, and churn predictions. Some examples include:
Customer 360: Demographics, purchase history, payments, preferences—one profile. Finance uses for AR insights; marketing for campaigns; AI for churn prediction. No re-coding per team.
Financial Analytics: GL, AR/AP, forecasts as a reusable product. Plug into planning tools, anomaly detection, internal controls, risk management, and scenario models. One build serves close, reporting, controls, planning, and AI.
3 Concrete Actions to Start
- Inventory to Product: Audit Top 3 data sources. Prioritise data that feeds into high value/impact use cases e.g. Vendor 360.
- Build the Catalogue: Define SLAs (freshness, completeness). Assign product owners, teams and governance structure.
- Quarterly Readiness Audit: Score data products (reuse rate, quality). Tie to the AI roadmap. Iterate based on consumer feedback.
Data products aren’t optional; they’re the wiring for AI scale. Through data products, you move from data silos to an interconnected data-driven ecosystem.
Let’s grab a coffee together and learn more about this.