Answer · AI Transformation
AI Opportunity Assessment: How to Rank What to Build First
Aiden Wayne · Updated August 17, 2026
Short answer
An AI opportunity assessment finds the places in a business where AI would change a number, then ranks them by value, effort, and dependency. Opportunities come from observed friction — rework, manual transcription, waiting, chasing, judgement bottlenecks — not from a vendor's feature list. The output is a short ranked portfolio where each item names the workflow, the number it moves, and what has to be true first.
Where real opportunities come from
Every durable AI opportunity is sitting inside work someone already does. The reliable places to look are the seams: where information changes hands, systems, or format.
- Transcription — a human moving data between two systems.
- Reconciliation — a human deciding which of two records is true.
- Summarisation — a human compressing a long thing into a short thing, repeatedly.
- Triage — a human routing work using rules they could explain.
- Chasing — a human following up because the system does not.
- Judgement bottlenecks — everyone waiting on one person's read.
Scoring the portfolio
Rank by value ÷ effort, then re-sort by dependency. The top of the final list is rarely the item with the highest raw score — it is the highest-scoring item nothing else is blocking.
| Dimension | What to ask |
|---|---|
| Value | What number moves, by roughly how much, how often does it recur? |
| Effort | Build, integration, and change management — including who has to work differently. |
| Dependency | What must exist first: data access, a definition, a decision owner, a system hook? |
| Risk | What happens when the output is wrong, and who sees it before the customer does? |
What to deliberately not do
- Do not start with the most visible workflow. Start with the one you can finish.
- Do not score opportunities you have not watched someone perform.
- Do not let tool availability define the portfolio.
- Do not carry an item forward that has no owner.
What good output looks like
A one-page portfolio: eight to fifteen candidates, scored, with the top three expanded into a workflow view, an estimated cost of the current state, and the dependency each one carries. Short enough that an executive reads all of it, specific enough that an engineer could start on Monday.
Frequently asked
- What is an AI opportunity assessment?
- A structured review that identifies where AI would change a business number, then ranks those opportunities by value, effort, dependency, and risk, ending in a short prioritized portfolio.
- How many opportunities should the assessment produce?
- Eight to fifteen candidates is typical, with the top three developed in detail. Longer lists reliably go unread.
- Can we do this before our data is in order?
- Yes — data readiness is one of the dimensions being scored. The assessment tells you which opportunities are blocked by data and which are not.