The AI reality check

01 — What problem are we actually trying to solve?

Can you clearly describe the business problem without using the words AI, automation or innovation?

Reality check: If the starting point is “we need an AI strategy”, you may be starting with the technology rather than the problem.

Your score:
🟢 Yes, confidently 🟠 Partly 🔴 Not yet

02 — What gets measurably better if we solve it?

Revenue, cost, time, risk, customer experience, staff capacity. What specifically should improve, and by how much?

Reality check: If you can’t define what success looks like, it’ll be difficult to know whether your investment has worked.

Your score:
🟢 Yes, confidently 🟠 Partly 🔴 Not yet

03 — Does this actually need AI?

Could a process change, better integration, conventional automation or simpler software solve the problem more effectively?

Reality check: AI should earn its place in the solution, not be the starting assumption.

Your score:
🟢 Yes, confidently 🟠 Partly 🔴 Not yet

04 — Do we have the data and systems to make it work?

Where will the information come from? Is it accessible, reliable and permitted to be used? What existing systems will need to connect?

Reality check: Great AI built on inaccessible, incomplete or poor-quality data is unlikely to deliver great results.

Your score:
🟢 Yes, confidently 🟠 Partly 🔴 Not yet

05 — What happens when the AI gets it wrong?

Who checks its output? What’s the impact of an error? When does a human need to intervene? Can decisions be understood and audited?

Reality check: Don’t just design for when AI works. Design for when it doesn’t.

Your score:
🟢 Yes, confidently 🟠 Partly 🔴 Not yet

06 — Will people actually use it?

Does it fit naturally into how people already work and solve something they genuinely struggle with?

Reality check: A technically impressive solution creates no value if people don’t adopt it.

Your score:
🟢 Yes, confidently 🟠 Partly 🔴 Not yet

07 — Can we prove the value before making a major investment?

What’s the smallest experiment you could run to test the biggest assumption and demonstrate measurable value?

Reality check: You shouldn’t need a six-month transformation programme to find out whether an idea has potential.

Your score:
🟢 Yes, confidently 🟠 Partly 🔴 Not yet