0 — not ready
The process is unclear, rare, highly variable, or depends on inputs that are not reliably available.
Automation decision guide
Which business process should you automate first?
The best first automation is usually not the task that feels most annoying. It is a repeatable process with a clear trigger, usable inputs, observable outcomes, and failure modes that can be controlled.
How to use the scorecard
Choose one process with a named start and finish. Score each dimension from 0 to 2 using representative cases, then review the pattern rather than treating the total as an automatic approval.
1 — partly ready
The recurring path is visible, but rules, access, sample data, or exception handling still need discovery.
2 — strong candidate
The trigger, inputs, rules, outcome, owner, and review points can be described and tested with real examples.
Value and safety must be assessed together
High volume alone does not make a safe automation. A useful first release keeps risky decisions visible, logs what happened, and gives a person a clear way to review or recover failed cases.
Automate movement before judgement
Collection, validation, routing, drafting, and reporting are often easier to control than an irreversible decision.
Design the exception path
Define what happens when data is missing, an integration fails, confidence is low, or a case does not match the normal rules.
Keep an accountable owner
Name who reviews alerts, approves sensitive outcomes, and decides when a changed process requires the automation to be updated.
Automation priority scorecard
Score each dimension 0, 1, or 2. Add a note with the evidence behind the score; the note is more useful than the number when the workflow is scoped.
- Frequency and volume
- How often does the same workflow run, and how much repeated handling does it create?
- Trigger and outcome
- Can you name the event that starts the process and the observable result that finishes it?
- Rule clarity
- Can the normal decisions be written as explicit rules rather than individual judgement?
- Input quality
- Are the required fields, documents, permissions, and representative examples consistently available?
- Exception rate
- How many cases leave the normal path, and can those cases be detected and routed to a person?
- Error reversibility
- Can a wrong output be stopped, reviewed, corrected, and replayed before it causes material harm?
- Ownership
- Is one person accountable for the process, its rules, alerts, and post-launch changes?
Interpret the result
Mostly 2s, with reversible errors and a clear owner
A prototype using representative cases is a sensible next step.
Strong value, but weak rules or poor sample data
Scope discovery first: map the process and collect edge cases before choosing the implementation.
High error cost with no reliable approval or recovery point
Do not automate the final decision yet; automate preparation and keep a human checkpoint.
Low frequency and many unique exceptions
Improve the manual workflow or templates before funding custom automation.
Turn the worksheet into a delivery scope
The guide is a planning aid, not an estimate. The related service page explains the normal deliverables, process, and baseline boundaries for this kind of engagement.
Explore workflow automation →Discuss a similar project
Describe the problem, current situation, and outcome you need. I will suggest a sensible first step.
Send a project enquiry