Before choosing a tool, look at the workflow: what repeats, which rules apply and where human involvement is still needed.
Start with the work already being done
Think of a task that involves copying information between tools, checking records or preparing the same report. Describe who performs it, which data it uses and what it needs to deliver.
This description helps separate the problem from the technology. A tool only makes sense when it fits the workflow and the actual conditions of the work.
Look for repetition, rules and exceptions
A predictable workflow provides a starting point for assessing automation. Identify its frequency, decision rules and the cases that require human judgment.
For example, transferring an order with defined fields can follow clear rules. Deciding how to handle an incomplete order requires guidance for that exception. Automation needs to account for both paths.
Check the conditions before promising a solution
Is the information accessible? Do the systems allow integration? Who confirms the result? Does the process depend on someone’s personal access or a step that still changes frequently?
These questions help identify dependencies that must be resolved before implementation. The assessment should also cover how a failure will be detected and who will be responsible for following it up.
Choose an initial deliverable that can be verified
Instead of automating the entire operation at once, define a single stage. Specify the input data, the expected result and how it will be checked.
An initial deliverable could be a standardized dataset for review, the transfer of a set of records or flags for discrepancies. Validation should use examples that cover both normal situations and exceptions.
Connect the improvement to the business objective
The purpose may be to reduce repetitive steps, make checks easier or improve process consistency. Define what will be observed to assess the improvement.
If the workflow does not yet have clear rules, organizing it may be the first task. Assessment, defining criteria and automation are all part of the same conversation.