Turning customer and business signals into AI priorities
An ROI-driven prioritization framework for deciding which AI scenarios are worth building — and how the team would know they worked.
high-value Copilot scenarios shaped by the framework
Context
When a technology can plausibly do almost anything, the constraint stops being capability and becomes judgment. Roadmaps fill with things that are possible rather than things that matter.
I built a prioritization framework so that AI investment decisions could be made explicitly, defended in a room of senior stakeholders, and revisited when the signals changed.
Visual model
- 01Customer impact
- 02Revenue signals
- 03Strategic importance
- 04Investment / effort
- 05AI feasibility
Prioritization across impact and feasibility, weighted by the five dimensions.
The five dimensions
- 01
Customer impact — how much of a real workflow this removes, for how many people.
- 02
Revenue signals — what the commercial evidence says, not what the deck says.
- 03
Strategic importance — whether this compounds into where the product is going.
- 04
Investment & effort — the honest cost, including evaluation and operations.
- 05
AI feasibility — whether current model capability clears the bar the workflow demands.
Why it works
The framework's value is not the score. It is that disagreement becomes specific: two people arguing about a scenario end up arguing about feasibility or about revenue signal, not about taste.
It also forces the measurement question early. If nobody can say how we would know it worked, the scenario is not ready to be prioritized.
I don't just ask what AI can do. I ask what is worth building, for whom, and how we will know it worked.