The slideware graveyard
Every large company now has an AI strategy document. Most of them share a fate: presented once, praised, filed. The pattern behind the graveyard is consistent — the strategy lists possibilities instead of decisions. Forty use cases, no ranking. Ambitions, no costs. A vision, no first project.
A roadmap that gets funded looks different. It fits on a page, names the first three builds, attaches a number to each, and says out loud what you're not doing this year.
Rank by ROI, not by excitement
The ranking exercise is where strategy becomes decision-making. For every candidate use case, we score four things with the people who own the process today.
- Value: hours, error cost, or revenue at stake — measured from the current process, not estimated from a benchmark report.
- Feasibility: is the data accessible and clean enough? Is the decision policy writable? Can success be evaluated objectively?
- Risk: what's the blast radius of a wrong answer, and is it recoverable? Customer-facing and compliance-heavy cases rank later, not never.
- Time-to-proof: how fast can this show a number the CFO believes? Momentum is a strategic asset — the first win funds the second.
Cost the whole iceberg
AI initiatives don't fail on model costs; they fail on everything under the waterline — data cleanup, integration with systems of record, evaluation infrastructure, change management, and the ongoing human review that responsible autonomy requires. A roadmap that prices only the build is a roadmap that dies at its first budget review.
We put the iceberg in the plan: build cost, run cost per month, and the internal time the client's own team must commit. Boards fund plans they can hold accountable. Vagueness reads as risk.
Sequence for compounding
The right first project is rarely the biggest one. It's the one that builds reusable foundations — the data pipeline, the evaluation harness, the governance pattern — that every later project inherits. Sequenced well, project three costs half of project one and ships in half the time. That compounding, not any single use case, is what an AI strategy is actually for.
Written by SCORPBIT Strategy Practice — humans working with AI at every step, accountable for every word.
