How to Choose Your First Workflow Automation Project
Use a five-factor scorecard and a short pilot to find the process your team is ready to automate.

Choose a first workflow automation project with frequent work, measurable rework, reliable inputs, manageable exceptions and a named owner. Score candidates on these five factors, then pilot one step with a manual fallback.
Key takeaways
- Measure the current workload before choosing a tool.
- Score all candidates against the same criteria.
- Check whether staff can take over manually before launching.
How can you score candidate workflows?
Use the same two-week sample for each candidate where volume allows. Rate each column from 0 to 2. A higher score means a better starting point. Agree on what 0, 1 and 2 mean before scoring.
| Factor | 0 | 1 | 2 |
|---|---|---|---|
| Frequency | Occasional | Weekly | Daily or many times per week |
| Rework | Little re-entry or correction | Visible time lost | Repeated, measurable correction or re-entry |
| Exceptions | Frequent and undefined | Some repeatable categories | Few, with an agreed manual route |
| Data | Missing or inaccessible | Available with cleanup | Reliable fields and stable identifiers |
| Ownership | No accountable owner | Owner named with limited time | Owner and backup can review results |
Compare a fictional intake queue at 8/10 (2, 2, 1, 1, 2) with supplier invoice intake at 6/10 (2, 2, 0, 1, 1). The invoice workflow may eventually save more time. The team can start learning with intake while finance defines its exception rules.
An acceptable score still depends on scope. If nobody can take over manually, the necessary access is unavailable, or a mistake is hard to undo, address that constraint before piloting.
What should the first pilot cover?
- Draw the start and end of one workflow and name its owner and backup.
- Measure weekly volume, hands-on time, exception rate and downstream corrections.
- Choose one step, such as assigning incoming requests that match a known category. Keep uncertain cases in an existing queue.
- Test routine cases, missing fields, duplicates and one exception that arrives while the owner is away.
- Agree on what counts as improvement and when to pause. Keep manual work available for comparison and recovery.
Measure the full task. A faster routing step can still increase work if reviewers spend more time correcting it. Ask the owner to compare completion time and corrections after the trial and decide whether to extend the pilot, change its scope or end it.
What should you decide after the pilot?
Review the results with the owner. Compare the same task types before and after, check whether the fallback handled every exception and list any cases the team still has to route manually. Extend the pilot when the owner can support the added volume and explain the remaining errors.
If two workflows score similarly, bring a volume estimate and a few typical exceptions to a workflow discovery discussion with Wizy. See Wizy’s discovery process.
Frequently asked questions
Should you automate the workflow that takes the most time first?
Start with frequent, measurable work that has reliable data and a clear owner. A larger workload with poorly understood exceptions usually needs process discovery before automation.
How do you measure the baseline for an automation pilot?
Record volume, hands-on time, exception rates and downstream corrections for comparable cases. Use those same measures after launch.
What makes a process unsuitable for a first automation pilot?
Resolve missing ownership, unusable inputs and hard-to-reverse errors before piloting. Check that someone can take over manually if automation pauses.
