“That’s Not Quite What I Wanted”: Closing the Last Mile of Information Product Delivery

Every data team knows that moment. Months after the requirements were gathered, the dashboard lands and the stakeholder says: “That’s not quite what I wanted.” The traditional flow, requirements, data design, data build, dashboard design, dashboard build, then feedback, puts the most valuable feedback at the end, when change is most expensive to act on. The ad hoc alternative is no better: guess from a one-line JIRA ticket, build it, and land back in the same loop.

This session shows you how to change your Information Value Stream. Starting from a completed Information Product Canvas, a light but well-formed set of requirements captured in 30 minutes, Shane shows how to use common GenAI tools to generate a working prototype and put it in front of the stakeholder in hours, not months. Feedback on the prototype then drives the data design, the dashboard design and whether to build at all. Learn how to embrace agility without becoming ad hoc, and where context, glossary definitions and concept models fit when you prototype first.

• Why the Information Value Stream breaks at the last mile: how the traditional flow of requirements, data design, dashboard design and build pushes stakeholder feedback to the very end, when it is most expensive to act on, and why the ad hoc alternative of guessing from a one-line JIRA ticket lands you back in the same “that’s not quite what I wanted” loop
• Start light but well-formed: the completed Information Product Canvas as the starting artefact, business questions, actions and outcomes, personas, delivery type, core business events, and why 30 minutes of shared-language requirements is enough to begin prototyping
• From canvas to prototype with LLMs: using an LLM to turn a completed canvas into a working prototype of the Information Product in hours; what context to give the model, what to check, and what a good first prototype looks like
• Feedback first: putting the prototype in front of the stakeholder, iterating on it, and using what you learn to drive the data design, the dashboard design and the build / don’t-build decision, including a team where one in five ideas stopped at the prototype and saved the cost of building it
• From prototype to production: pointing the validated prototype at real data , modelling, transformation, testing, go-live, without losing what the stakeholder already agreed to, and how the canvas and the prototype become the acceptance criteria
• Agility, not ad hoc: running prototype-first as a disciplined value stream rather than 5,000 unmanaged dashboards, and where context, glossary definitions and concept models do their heavy lifting when you prototype first, just in time from the canvas, not two years up front.