Most organisations do not have an idea shortage

Once people begin looking for automation opportunities, the list grows quickly. Reporting, approvals, document handling, data entry, customer communication, knowledge search and internal service requests all appear to be candidates.

A long list can feel like progress, but it creates a new problem: every item competes for attention without a consistent way to compare value, feasibility and risk. The loudest request or most interesting technology can become the first project.

Prioritisation is the discipline that turns possibility into a credible implementation decision.

Describe the operating problem precisely

Avoid opportunity statements such as “use AI in customer service” or “automate finance”. They describe a technology or department, not a workflow.

A useful statement identifies the trigger, current handling, consequence and intended improvement. For example: “Supplier invoices arrive through several channels, require manual field entry and are often returned for missing approval information. We want a consistent intake and approval process with visible exceptions.”

Precision makes the opportunity testable and prevents the solution from expanding before the problem is understood.

Establish enough evidence

The first assessment does not need perfect data, but it should use more than opinion. Review representative cases, frequency, handling steps, waiting time, error patterns, systems involved and the people who own the decisions.

Separate productive judgement from avoidable handling. A task taking time does not mean all of it should be removed. The goal is often to return capacity by eliminating re-entry, coordination and context gathering while preserving valuable human work.

  • How often does the workflow occur?
  • Where does work wait or return for correction?
  • Which steps require genuine judgement?
  • What systems and permissions are involved?
  • What happens when the normal path does not apply?
  • Which measure would show that the change worked?

Compare opportunities across five lenses

A simple scoring conversation is more useful than false precision. Use the same questions for each candidate and document the evidence behind the judgement.

  • Operational value: would improvement release capacity, shorten a material cycle or improve a decision?
  • Evidence: is the friction recurring and observable?
  • Feasibility: can the required systems, data and decisions be accessed?
  • Risk: what is the consequence of error, delay or inappropriate access?
  • Adoption: are the owner and affected team ready to change the way the work is performed?

Select a bounded first move

The best first implementation is not always the highest theoretical value. It is a meaningful opportunity with clear ownership, manageable dependencies and a result that can be observed within a focused delivery cycle.

A bounded scope should still represent real work. Automating a demonstration path while leaving every exception manual proves little. Include the most important normal cases, a safe exception path, operating controls and the measures needed for review.

Use decision gates instead of a large program

Separate understanding, prioritisation, implementation and measurement. Each stage should produce a decision about whether to continue, adjust or stop.

This reduces commitment to weak assumptions and keeps technology choices proportional to the problem. It also allows an organisation to build confidence and internal capability without launching a broad transformation program.

  • Discovery: is there a meaningful problem worth assessing?
  • Assessment: are the findings material and validated?
  • Prioritisation: is the first improvement valuable and buildable?
  • Implementation: does it work safely in the real environment?
  • Review: did the operating measure change enough to extend?

Keep the backlog, but do not let it run the strategy

Capture ideas discovered along the way, including dependencies and evidence gaps. Then review them through an ongoing operating rhythm rather than treating every suggestion as an active project.

A measured first improvement changes what the organisation knows. It may reveal that another opportunity is more valuable, that a platform constraint needs attention or that the expected capacity was not present. Good prioritisation remains open to that evidence.

Practical progress comes from choosing one well-understood improvement, implementing it with the right controls and using the result to make the next decision.