Momentum is one of the most under-examined forces in AI decision-making.
Once a solution gains executive sponsorship, communications support, budget line and project identity, reversing or narrowing it becomes socially and politically expensive. That is why the most useful questions are asked early, while the decision is still movable.
These four questions are designed to be asked together. They map to Value, Trust and Alignment in the VTA Framework.
1. What outcome are we trying to improve, and for whom?
This question sounds basic. It is often skipped.
Teams can describe the tool in detail while struggling to describe the outcome in language that would make sense to a service user, frontline worker or finance reviewer. If the outcome is vague, every later success metric becomes negotiable.
A strong answer names:
- The group affected
- The change that would matter to them
- Why that change is important now
- What evidence would show improvement
If the answer relies mainly on words like efficiency, innovation or transformation, keep asking.
2. Why is AI a better response than the credible alternatives?
AI should not win by default.
Alternatives may include process redesign, role clarification, better data stewardship, policy change, partnership working, training, or targeted non-AI automation. This question does not assume those options are always better. It requires the organisation to justify why AI deserves the risk, cost and change load.
A strong answer explains:
- What AI can do that alternatives cannot do proportionately
- What risks AI introduces that alternatives avoid
- Why now is the right moment
If the only answer is that the tool is impressive, the case is incomplete.
3. What has to be true in our organisation for this to work?
This is the alignment test.
Many AI initiatives fail not because the model is weak, but because the organisation is not arranged to use it well. Data may be fragmented. Workflows may be inconsistent. Ownership may be unclear. Teams may already be at capacity.
A strong answer identifies:
- Required workflow conditions
- Capability and support needs
- Integration and maintenance responsibilities
- Dependencies that could undermine reliability
If these conditions are not in place, the decision may still proceed, but only with explicit preparation or conditions attached.
4. How will the people affected experience this decision?
This is the trust test.
People do not experience a business case. They experience changed expectations, new error types, altered accountability, and messages about what the organisation values.
A strong answer considers:
- Whether staff and users will understand the purpose
- Whether confidence in outputs is realistic
- Whether there is a credible route to challenge mistakes
- Whether benefits and burdens are distributed fairly
If the likely experience is confusion, extra checking work, or fear, that is decision-relevant evidence.
Using the four questions well
These questions work best in a leadership conversation with the right people in the room: operational owners, technical leadership, and someone who can speak to lived experience.
The aim is not to create a long document. The aim is to reveal whether the proposal is ready for momentum.
If the four answers are strong, proceed with confidence. If they are mixed, set conditions. If they are weak, reframe or pause.
Momentum should follow understanding, not replace it.
Who should be in the room
These questions fail when only the technology function answers them. Value needs the owner of the outcome. Alignment needs the owner of the workflow. Trust needs someone who can speak to lived experience without being asked to endorse a decision already made.
If an AI solution in your organisation is gaining momentum before the case has been tested, Get in touch. We can facilitate the four-question conversation with your leadership team.
Start by linking your enquiry to this guide.