In short
Start with a workflow that carries a meaningful consequence
- Start with a workflow that carries a meaningful consequence, and test whether changing it improves the accepted outcome.
- Six questions build the shortlist. They cover consequence, constraint, understanding of the work, information, quality and who can make the change happen.
- Compare alternatives in order: whether the work should exist at all, then simpler process changes, then conventional automation, then AI.
- Rank candidates on consequence, evidence, feasibility, dependencies and risk. A simple priority judgement with reasons is more useful than a complex score built on unsupported estimates.
Detail
Six questions to test a candidate workflow
Look at intake and triage, transactions, customer cases, quoting and proposals, and knowledge or review work. These are places to investigate, and none of them automatically needs AI. Useful signals include backlogs, slow response, rework, rising cost-to-serve and effort spent moving information between systems. Confirm the pressure is current, because an old vendor case may describe a problem management has already solved.
Then ask 6 questions of the candidate.
What consequence would improve? Name the customer, financial, service, capacity or risk outcome. Saving time is an incomplete answer until you identify whose time and what happens to it.
Where is the constraint? Follow the whole process, because automating upstream work may fill a downstream queue faster.
Is the work understood? Clear inputs, rules and exceptions support assessment. Unresolved policy disagreements need management decisions before automation.
What information is required? Check quality, access, currency and appropriate use. An AI summary of conflicting records is still unreliable.
How will quality be judged? Define acceptance and the consequences of error, and include reviewer cost and capacity in the case.
Who can make the change happen? Identify the workflow owner, the implementation capability and the benefit owner.
Commercial impact
Compare removal, simpler change and automation before AI
Compare alternatives in a sensible order. First consider whether the work should exist. A report nobody uses may need removing, and producing it faster adds nothing. Then consider simpler forms, clearer rules, fewer hand-offs or a better approval process. Existing software and conventional automation may solve the remaining task more reliably than AI.
Use AI where handling language, variation or inference creates an advantage you can defend within acceptable limits. Keep human judgement where it is required. The sequence is to remove the work first, then automate what survives, then move the time to other work. Use it as a set of questions. It does not mean that every human task should disappear.
Ranking follows the same discipline. A high-potential but uncertain idea may deserve a small experiment. A moderate-value improvement with strong evidence may be the better first implementation. Some uses need specialist work before any pilot runs.
In practice
Three illustrative candidates for process selection
Three illustrative candidates show what selection looks like.
Enquiry routing: investigate whether classification is the constraint, how errors affect service and whether the receiving teams can handle the volume.
Management reporting: establish which decisions the pack should support before automating its existing slides.
Customer eligibility decisions: repetition alone does not justify automation. Assess the affected people, the obligations, the data, the review arrangements and the specialist controls.
These examples illustrate selection. They are not client results or deployment recommendations.
What to do
Steps to select and test the strongest automation candidate
- Start from a workflow with a meaningful consequence. The most visible repetitive task is often the wrong place to begin.
- Confirm the pressure is current before building a case, because the organisation may already have solved it.
- Follow the whole process to find the constraint. Otherwise, automation upstream may simply fill a downstream queue faster.
- Test removal and simpler process change before automation, and conventional automation before AI.
- For the strongest candidate, define the question to resolve, the evidence required, the owner and the decision at the end.
Last reviewed
Put this topic against your own numbers, owners and timelines.