Practical guide
Simplify before you automate
Follow a practical sequence for understanding and improving a workflow before deciding what, if anything, to automate.
- Author and publisher
- Nineteen Software
- Published
- Reviewed
Automation preserves the process it is given
Automation can make a clear, repeatable workflow easier to operate. It can also make confusing rules, unnecessary steps, inaccessible experiences, and poorly handled exceptions happen faster and at a larger scale. Improve the process first so any later technology supports work the business actually wants to keep.
The sequence below does not require specialized software. A shared document, paper, or whiteboard is enough if the people responsible for the work can see and challenge the same description.
1. Define the desired outcome
State what should be true when the process works well. Describe the result for the business and affected people without naming an automation, application, or feature. Record how the business will recognize a useful improvement using evidence it can actually observe.
2. Identify users and owners
List who starts the work, provides information, makes decisions, performs the steps, receives the result, supports the process, and owns future changes. An owner needs enough authority to resolve unclear rules and accept the ongoing responsibility of any chosen solution.
3. Observe the current workflow
Follow representative normal work from beginning to end. Then examine common exceptions. Use what people actually do rather than relying only on a policy or diagram. Note inputs, outputs, tools, approvals, wait states, duplicate entry, rework, accessibility barriers, and informal workarounds.
4. Map steps and handoffs
Put each action and decision in order. Show where information changes format or ownership, where a person waits, and where the path branches. Record what triggers the work and what counts as complete. Keep exceptions visible rather than forcing them into an ideal path.
5. Describe the friction
For each material difficulty, record how often it occurs, who is affected, what delay or rework follows, what errors or risks can result, and what workaround is used. Use verified business evidence where available. Do not create a savings or return estimate simply to justify technology.
6. Remove unnecessary work
Challenge steps that neither create a useful outcome nor manage a necessary constraint. Remove duplicate capture, combine compatible decisions, and stop producing information nobody uses. Preserve controls that serve a real legal, privacy, security, safety, accessibility, or operating need.
7. Clarify decisions and exceptions
Define the information each decision requires, who owns it, and what should happen when information is missing or the normal rule does not apply. An automation that cannot handle exceptions safely still needs an explicit human path, recovery behavior, and accountable owner.
8. Standardize the improved process
Write the smallest shared description people can follow and maintain. Agree on the authoritative data, inputs, outputs, roles, measures, failure handling, documentation, training, privacy and security responsibilities, accessibility needs, and review trigger. Test the revised process manually before automating it where practical.
9. Assess whether technology is needed
Ask whether the improved process can be sustained with clearer ownership or better use of an existing tool. If a gap remains, compare configuration or purchase, integration, automation, and custom-building. Include implementation effort, recurring cost, data portability, reliability, maintenance, support, training, and exit strategy—not only the initial feature match.
Process-mapping worksheet
Use this worksheet privately on screen or in print. It sends no answers to Nineteen Software.
| Prompt | Your working notes |
|---|---|
| Desired outcome | What should be true for the business and the people affected? |
| Trigger and completion | What starts the workflow, and what verifiable result ends it? |
| Users and owner | Who performs, decides, receives, supports, and owns the process? |
| Inputs and outputs | What information or material enters and leaves, and which source is authoritative? |
| Current steps | What happens in order during a representative normal case? |
| Decisions and approvals | What information is required, who decides, and which approval manages a necessary risk? |
| Handoffs and wait states | Where does ownership change, work pause, or follow-up depend on memory? |
| Duplicate entry and rework | What is copied, corrected, reconciled, or repeated, and why? |
| Exceptions and errors | Which non-standard cases matter, how are they detected, and who recovers them? |
| Accessibility and experience | Can affected people understand and complete their part across relevant devices and access needs? |
| Privacy and security | What data and permissions are necessary, what obligations apply, and what should not be collected or exposed? |
| Reporting and evidence | What information supports a real decision, and how will its quality be checked? |
| Maintenance and support | Who will update rules, monitor failures, train people, document changes, and retire the solution? |
| Simplifications to test | Which step, rule, handoff, duplicate, or workaround can be removed or clarified before technology changes? |
| Remaining capability gap | After simplification, what need still cannot be met responsibly? |
Decide from the improved process
A stable manual process may be the right result. An existing product or a configuration change may close the remaining gap. Integration may remove duplicate transfer between sound systems, and automation may handle clear, repeatable rules with appropriate exception paths. Custom software deserves investigation only when the need and lifecycle responsibility justify it.
If you need help assessing the workflow, request a free consultation. The assessment provides verbal initial options only and does not guarantee scheduling, a proposal, or delivery.