An afternoon spent answering the right questions before an AI transformation can save months of wasted effort. You will not have perfect answers to all of them, and that is fine. The point is to know where you are clear, where you are guessing and where you need to find out more before committing.
Purpose
- What business problem are we trying to solve? In one sentence, with a number if possible.
- Why now? What makes this a priority this year?
- What does success look like in 12 months?
- What are we not going to do with AI?
Ownership and leadership
- Who is the single accountable owner?
- Who is the executive sponsor?
- Do our senior leaders use AI themselves?
- How will decisions about AI tools and uses be made?
People
- How are our people likely to feel about this, and why?
- What will we tell staff about how AI affects their roles?
- Who needs training, and what kind?
- Who are our potential champions?
Risk and governance
- Which AI tools will we approve, and on what basis?
- What information must never go into AI tools?
- Where must a human always review AI output?
- How will we handle UK GDPR and any sector rules?
Process and measurement
- Which workflow will we start with, and why?
- What is the baseline for that workflow today?
- How will we measure adoption, efficiency, quality and outcomes?
- When will we review progress and decide what to scale or stop?
How to use the questions
Work through them with a small group: a senior sponsor, the intended owner, a manager close to the work, someone from IT and someone from data protection. Mark each answer as:
- Clear: we know the answer.
- Assumption: we think we know, but should test it.
- Unknown: we need to find out.
Questions marked "unknown" become your first actions. Questions marked "assumption" become things to test in your first pilot.
The questions people most often skip
In our experience, the most commonly skipped questions are 4 (what we will not do), 10 (what we will tell staff), 18 (the baseline) and 20 (when we will decide). Each of these, left unanswered, is a common cause of AI projects drifting or failing.
From questions to plan
Once you have answers, you have most of the content for a one-page AI direction or a 90-day plan. See what an AI strategy looks like on one page for how to turn them into a document. For more strategic questions, see strategic questions to ask about AI.
Our Leading AI Innovation & Organisational Transformation course helps you answer these questions for your own organisation, finishing with a practical 90-day plan.
Frequently asked questions
Do we need answers to all 20 questions before starting?
No. You need clear answers to the ownership, data and first-workflow questions. Others can be refined as you learn.
Who should answer these questions?
A small cross-functional group led by the intended AI owner, with input from leadership, IT, data protection and frontline managers.
How often should we revisit the questions?
At each major review point, typically every quarter, because answers change as you learn.