AI Strategy and Operating Model
AI use case prioritisation
Reduce a long list to a defensible shortlist using three gates: strategic consequence, measurability, feasibility
- Duration
- 2 weeks
- Ladder stage
- Pre-Proof
- NATIVE stages
- Target
The situation
When this engagement is the right one
Long lists are easy to produce and impossible to fund. The work is subtraction. We run the three gates in a single structured session and publish the reasons for every cut, so the decision survives the next executive who asks why their idea is not on the list.
You are probably seeing
- The list has grown every quarter and nothing has left it.
- Prioritisation is currently a two-by-two where everything is top-right.
- Teams are lobbying for their own candidates rather than arguing about metrics.
What we do
The work, in the order it happens
- 01Normalise every candidate to a single sentence: workflow, metric, and who is affected.
- 02Apply gate one, strategic consequence: what breaks if this never happens.
- 03Apply gate two, measurability: is there a baseline that exists today.
- 04Apply gate three, feasibility: data, integration, and governance capacity.
- 05Publish the rejection reasons alongside the shortlist.
What you get, and keep
- A shortlist of one or two candidates with completed value maps.
- A rejection register with the gate each candidate failed.
- A one-page prioritisation standard your teams can reuse without us.
Prerequisites
- An existing candidate list, however messy.
- Access to the people who own the metrics being claimed.
Not included
- We do not estimate ROI for candidates that have no measured baseline. We record that the baseline is missing.
Duration and price shape
2 weeks
Fixed fee. Often bundled into the AI Readiness Diagnostic.
Where this sits in the method
Commercially this is a Pre-Proof engagement on the Proof, Product, Platform ladder.
FAQ
Questions we get asked
- What if leadership disagrees with a cut?
- The rejection register exists for exactly that conversation. Every cut names the gate it failed, so the argument is about evidence, not preference.
- Can a shortlist have five items?
- It can, but it will not be funded properly. A shortlist that still has eight items has been transcribed, not triaged.
Related engagements
Work that usually sits either side of this
AI Strategy and Operating Model
AI Strategy and Roadmap
A sequenced twelve-month plan built from a triaged use-case shortlist, each item carrying a metric, an owner, and a validation method
AI Strategy and Operating Model
AI Readiness Diagnostic
A scored assessment across leadership commitment, data usability, workflow clarity, and governance, with a named owner assigned to the weakest pillar
Next step
Ready to scope use case portfolio design?
Bring the pilots you already have running. The first call is a scoping conversation, not a pitch.