We've written before about the five-dimension framework for prioritising AI use cases. Frameworks are easy to describe and harder to apply. This article is the applied version: an actual scorecard, filled in, with six illustrative use cases scored and ranked, so you can see exactly how the mechanic works before you run it on your own list.
The five dimensions, recapped
- Strategic alignment: how directly this supports a stated strategic priority
- Business value: the realistic business impact if executed well
- Feasibility: how achievable it is given current data, infrastructure, talent, and change capacity
- Time to value: how long before it delivers meaningful results
- Risk and governance complexity, scored as favourability: a 5 means low risk and light governance overhead, a 1 means high risk and heavy governance overhead. Scoring it this way keeps every column pointing the same direction, so a higher number is always better and the totals stay simple to read.
Score each dimension from 1 to 5. Add them up. The total is out of 25. That's the entire mechanic. Weighting the dimensions differently is a legitimate refinement once you trust the basic version, but don't start there, start with the simple sum.
A worked example scorecard
Six illustrative use cases, scored across the five dimensions, ranked by total.
| Use case | Strategic | Value | Feasibility | Time to value | Risk favourability | Total /25 |
|---|---|---|---|---|---|---|
| AI-assisted invoice processing | 3 | 4 | 5 | 5 | 4 | 21 |
| Internal knowledge-base search assistant | 3 | 3 | 5 | 5 | 5 | 21 |
| Automated marketing content generation | 2 | 3 | 5 | 5 | 4 | 19 |
| Customer service chatbot | 4 | 4 | 3 | 3 | 3 | 17 |
| Predictive maintenance for equipment | 4 | 4 | 3 | 2 | 4 | 17 |
| AI-powered credit risk scoring | 5 | 5 | 2 | 2 | 1 | 15 |
What the matrix tells you
Look at where AI-powered credit risk scoring lands. It has the highest strategic alignment and business value scores on the entire list, a 5 and a 5. It ranks last. This is precisely the failure mode we've written about before: prioritising by enthusiasm and headline value rather than evidence. The regulatory weight, data maturity requirements, and governance overhead of a high-risk, regulated use case make it genuinely harder and slower to do well, and the scorecard makes that visible instead of letting ambition hide it.
Compare that to the tie at the top: invoice processing and the internal knowledge-search assistant. Neither is glamorous. Both score modestly on strategic alignment. But both are highly feasible, fast to value, and low risk, which is exactly the profile of a use case that builds credibility early and funds the appetite for harder problems later.
The credit risk scoring use case isn't wrong to pursue. It's wrong to pursue first. A balanced portfolio still has room for it, once the organisation has the governance maturity and delivery credibility that the quicker wins help build.
The exact scores matter less than the conversation the scoring forces you to have. If two people in the room score the same use case a 5 and a 2, that gap is more valuable than any number in the total column.
How to run this with your own use cases
- List your candidate use cases, ideally no more than fifteen to twenty at a time
- Get the right people in the room: a business owner, a technical lead, and someone who can speak to risk and governance, not just one person's opinion
- Score independently first, then compare, don't average away disagreement silently, discuss it out loud, disagreement is diagnostic information about where the real uncertainty sits
- Rank by total, then sanity-check the ranking against portfolio balance across time horizons and business functions
- Revisit quarterly, scores change as data maturity, regulation, and business priorities shift
Before any use case that clears this scorecard moves to build, it should also clear a governance check specific to the UAE regulatory environment, not just this generic risk score.
Run it through the checklist first: The AI Governance Checklist for the UAE or read the full framework: How to Prioritise AI Use Cases, or talk to us about running this with your own list.