AI Product Feasibility Checklist
A checklist for deciding whether an AI product idea is feasible: the problem, the data, the approach, how it can fail, the people, the delivery and the approvals.
Free to download. Tick the boxes on screen or on paper.
What it checks
- The problem: whether two people would agree what a good answer looks like, and whether a rule, a search or a better form was considered first
- The data: that it exists, that use for this purpose is permitted in writing, and that it covers the awkward cases and matches what the live product will see
- The approach: a defined first version, quality judged on examples kept apart from those used to build it, and what happens if an outside model changes
- How it can fail: each wrong answer listed with who is affected, whether a person can review or reverse it, a fallback, and who checks quality after launch
- People and process: a named owner who has agreed, a redrawn working process, a stop date for the old way and a sponsor who funds running
- Delivery and operation: known running cost, safe and reversible change, notice of supplier changes, user support and a plan to retire the product
- Risk and approvals: the live design seen by security, privacy and risk, applicable rules identified by someone qualified, and records of testing
- A decision guide and a decision record
What it cannot tell you
It is a guide for a conversation, not a score, and it tests no model or data. It cannot say whether the idea will succeed. The AI Project Brief Builder turns the answers into a brief, and the Product Discovery Canvas covers the problem if the first section is thin.
On this page
From ticks to a decision
- 01
Tick what you can show
A statement counts only if it is true today and you could point to why.
- 02
Read by section
Look for the thin section, not the total.
- 03
Test the weakest
Pick the cheapest test that could prove the idea wrong, and run it before building.
- 04
Record it
Write the decision, who took it, the weakest section and who owns the next step.
What do people ask before using the AI product feasibility checklist?
When should I use the AI product feasibility checklist?
Use it while an AI product idea is still being judged, before the whole product is built, and answer for one idea at a time. It checks the problem, data, approach, failure, people, delivery and approvals, so the weak area shows up while a test is cheap.
When is it right to tick a statement on the checklist?
Only when the statement is true today and you could show why. A plan or a hope is not a tick. This keeps the result candid, because the decision guide reads your ticks section by section, so a thin section stands out instead of hiding behind a high total.
How do I turn my ticks into a decision?
The decision guide points to a bounded first build, a test of the weakest sections, or a stop. Thin problem, data or ownership sections mean stop building and return to discovery, because a better model cannot repair those gaps later.
Can the checklist tell me whether my AI product will succeed?
No. It is a guide for a conversation, not a score, and it tests no model or data. It shows where the idea is thin. If the problem section is weak, the Product Discovery Canvas helps, and the AI Project Brief Builder turns your answers into a brief.
Testing a real AI product idea?
Describe what you are weighing in a few lines. A senior practitioner replies with a suggested first conversation.