AbjakHow-to
The Earns Its Keep test: score an AI idea before you spend a penny
Once you've found a few jobs that drag, the next question is which one to fix first. Gut feel tends to pick the most exciting idea. The test below picks the one most likely to pay.
Four questions, scored 1 to 5
On every question, 5 is good. Add them up for a score out of 20.
| Question | 1 | 3 | 5 |
|---|---|---|---|
| Impact: what does this drag cost us a year? | Under £2k | £10k to £25k | Over £50k |
| Frequency: how often does it happen? | A few times a year | Weekly | Many times a day |
| Feasibility: how ready is the information? | In heads and on paper | In systems, a bit messy | Digital, tidy, one place |
| Ease: how fast could a fix be running? | Months: a custom build | Weeks: connect a few tools | Days: off the shelf |
What the total tells you
- 15 to 20: start here.
- 11 to 14: fix something first, such as the data or the process, then come back.
- 10 or under: park it for now.
High impact with low feasibility isn't a no. It's a not yet.
Three ideas, scored honestly
Here is how three ideas from one example business might score:
- Chasing unpaid invoices: 3 + 4 + 5 + 5 = 17. Start here. It's probably already in your accounts software.
- Preparing quotes: 4 + 5 + 3 + 3 = 15. The big prize. Worth doing properly.
- Predicting which customers will leave: 4 + 3 + 2 + 2 = 11. Not yet. Sort the data first.
Climb one rung at a time
Most of the value for a small business sits on the first two rungs: assist (you ask, it helps) and automate (when this happens, do that). The next rungs, augment (it spots and suggests, you decide) and act (it does the job within guardrails), are worth reaching only once the rung below is proven.
Five questions before you buy anything
- Does it do my job, on my real, messy examples?
- Does it plug into what we already use?
- Where does our data go, and is it used for training?
- What does it do when it's not sure?
- Can we leave, and take our data with us?
The demo always works. The demo has never met your data.
If an idea scores well and the numbers are big, it's worth doing properly. That's exactly where Discovery & Diagnosis starts.
Sources
From Abjak's “AI That Earns Its Keep” talk and workshop materials, University of Liverpool Management School, 2026. Example scores are illustrations, not client results.