A good automation candidate is bounded before it is automated
| Signal | Good candidate | Poor candidate |
|---|---|---|
| Frequency | Happens often enough to justify attention | Rare edge case with little recurring effort |
| Inputs | Known systems, forms, fields or documents | Ambiguous information gathered differently every time |
| Rules | Explicit decision rules or routing criteria | Depends mainly on tacit expert judgment |
| Exceptions | Exceptions can be recognized and handed to a person | The system must guess when uncertain |
| Outcome | Time, completeness, throughput or error rate can be observed | Success cannot be measured or attributed |
Useful design rule
Automate the coordination around a decision before automating the decision itself. A system can collect the context, identify gaps, route the work and prepare the record while the responsible person retains pricing, approval, negotiation or technical judgment.
01 / process group
Customer intake & enquiries
- 01Capture website enquiries into one structured record.
- 02Classify incoming email by customer, service, urgency or project type.
- 03Extract standard fields from submitted forms, PDFs or attached documents.
- 04Detect required information that is missing before internal review.
- 05Send approved acknowledgement or missing-information requests.
02 / process group
Sales & follow-up
- 06Create or update CRM records from qualified inbound enquiries.
- 07Assign opportunities using explicit territory, product or account-owner rules.
- 08Schedule approved follow-up when a proposal or quote is waiting for a response.
- 09Prepare meeting context from existing CRM, email and project information.
- 10Record structured outcomes after a sales stage changes.
03 / process group
Operations & handoffs
- 11Create a project or job record after an approved commercial handoff.
- 12Route documents into the correct project folder and naming structure.
- 13Notify the next owner when a required step is complete.
- 14Track deadlines and surface items approaching a configured threshold.
- 15Compile routine status summaries from authoritative system records.
04 / process group
Finance & administration
- 16Prepare invoice reminder drafts based on configured aging rules.
- 17Route receipts, purchase documents or invoices to the correct record.
- 18Collect recurring inputs for a monthly reporting package.
- 19Reconcile simple record mismatches for human review rather than silently changing source data.
- 20Prepare approval packets that put the supporting context beside the decision request.
05 / process group
Customer service & retention
- 21Route support requests to the right queue using known account and issue information.
- 22Send status notifications when an authoritative project state changes.
- 23Prepare renewal or maintenance reminders from existing contract dates.
- 24Collect customer feedback after a defined service milestone.
- 25Summarize recurring request themes for a human owner to review.
Keep judgment, accountability and commercial authority with people
- Final pricing and commercial terms when the decision depends on context, risk or negotiation.
- Engineering, legal, clinical, safety or other professional judgments that require qualified responsibility.
- Customer commitments about scope, delivery dates, exceptions or remedies unless a person has approved them.
- Hiring, termination, disciplinary, credit or other consequential decisions where automated judgment creates unacceptable risk.
- Ambiguous exceptions that do not fit the configured rules; route them to an owner instead of forcing a confident answer.
Prioritize by value, reliability and implementation effort
Start with three to five candidate processes and score them before choosing the first build. A simple scoring conversation is usually more useful than starting with a particular AI product or automation platform.
| Question | Score 1 | Score 5 |
|---|---|---|
| How much recurring time does it consume? | Minimal | Material recurring effort |
| How standardized are the inputs? | Highly variable | Mostly structured and predictable |
| How explicit are the rules? | Tacit / subjective | Documented and testable |
| How costly are errors? | Very high / hard to reverse | Low and easy to review or reverse |
| How measurable is the outcome? | Hard to observe | Clear time, throughput or completeness metric |
| How hard is the integration? | Many fragile systems | Few stable systems with accessible interfaces |
Then run the strongest candidate through the Automation ROI Calculator. If the process does not consume meaningful time or create another measurable business problem, automation may not be the right priority.