Free spreadsheet
Score every AI idea on the same ten elements
A scorecard that lets you compare any two AI ideas, from any function, like for like, and rank them by readiness as well as value.
Two ideas can both be "high priority" and be a year apart
A productivity assistant and an AI feature inside your product can both score high on benefit. One can start on Monday. The other is a year of research before it is a project.
Without a common scorecard, both go to the board as "high-priority AI ideas", and the board cannot tell them apart. With one, the first becomes a quick win and the second a research programme, and both are funded for what they are.
What each idea is scored on
| # | Element | The question | Scale |
|---|---|---|---|
| 1 | Priority | How urgent is it, against the other ideas? | High, medium, low |
| 2 | Strategic fit | Does it serve what the business is trying to do this year? Does it save money or make it? | 1–5 |
| 3 | Technical feasibility | Can it be built or bought now, in this environment? | 1–5 |
| 4 | Complexity | How much integration, change and dependency does it carry? | 1–5 (5 = simplest) |
| 5 | Estimated cost | Proof of concept, build and running cost, together | £, ££, £££ |
| 6 | Estimated benefit | The net value, in efficiency, quality, revenue or risk | 1–5 |
| 7 | Risks | Technical, safety, legal, adoption and data privacy | Red, amber, green |
| 8 | Data readiness | Low dependency, in hand, siloed, partial or absent | Five positions |
| 9 | Vendor options | Buy or build, and the candidate products | Notes |
| 10 | Recommendation | Do now, prepare, research, or park | One of four |
The scores are relative and deliberately coarse. A £ band is not an estimate. The numbers get firmed up when an idea becomes a business case.
Measure the process, not the AI activity
A benefit is net: the time saved producing something, less the time spent checking, correcting and reading it, less the cost of running the system.
Not time saved generating content
Total time to produce, review and consume it.
Not tickets answered
Cost per customer issue actually resolved.
Not documents processed
Cost per accurate, completed transaction.
Not employees using AI
A measurable change in a business outcome.
Group by readiness, not by enthusiasm
The spreadsheet ranks the ideas, then sorts them into four groups. These are priorities, not phases: different teams can work on different groups at once.
- Ready now. High value, low complexity, data in hand or not needed. Start this quarter.
- Nearly ready. High value, needing data preparation or coordination first.
- Strategic. High value, but gated by data and complexity. Phase it through bounded pilots.
- Research. The most differentiating and the longest horizon. Fund it as research.
Underneath all four sits the data foundation: the shared data the siloed and partial ideas all need. Start it alongside the quick wins, not after them. The four routes an idea can take to production are in build, buy, partner or wait.
Using the scorecard
Who should do the scoring?
Score every idea yourself first, as a straw man, then let the people who do the work correct it. A wrong score draws out specifics that a blank sheet never would.
How are the ten elements weighted?
The spreadsheet starts with equal weights on the five numeric elements. Change the weights at the top of the sheet to match your priorities, and the totals and ranks update.
Can we use our own business case template instead?
Yes. The ten elements are designed to map onto the investment templates most businesses already use, so a scored idea lifts straight into one. Add the four governance fields if your template lacks them.
Score your ideas with the people who know
CONVERGE chapters 8 and 9 set out the workshops that produce and correct the scores. Or we can run them with your teams.