Contact

Where to start with process automation: a practical guide to quick ROI

A man and three boys ride a small sailboat through choppy water in a fresh breeze, a schooner in the distance. (Winslow Homer, Breezing Up (A Fair Wind), 1873–1876)
Winslow Homer: Breezing Up (A Fair Wind), 1873–1876 · National Gallery of Art, Washington · Public domain
Category
Automation
Published
Reading time
~3 min read
Author

Summary

Automate first what is frequent, rule-based, and tedious. Fix the process before you mechanise it.

What do we mean by process automation?

Process automation means having software carry out a repetitive work step instead of a person: transferring data from one system to another, sending notifications, generating documents, or requesting approvals. The goal is not to replace people, but to free their attention for work that genuinely requires human judgement.

Even more important than speed is reliability: software never forgets a step, never mistypes a number, and follows the exact same procedure every single time.

How to spot good candidates

Spend a week observing where time actually goes. Strong candidates for automation share a few unmistakable traits.

  • Frequent: happens multiple times a day or week.
  • Rule-based: you can clearly state what should happen and when, with few exceptions.
  • Built on digital input: the data already exists in a system or file.
  • Involves copying: someone re-enters the same information in multiple places.
  • Error-prone: a typo takes significant effort to untangle later.
  • Causes bottlenecks: work stalls waiting for someone to hand something over.

Simplify first, automate second

Automating a bad process just makes it go wrong faster. Before mechanising anything, map out how the workflow operates today step by step, and ask: what part of this is actually necessary?

You will often find that an approval loop no longer serves a purpose, a report goes unread, or two steps can be merged into one. Whatever can be eliminated does not need to be automated. Mechanising the streamlined process that remains is both simpler and cheaper.

Which tool for the job?

Common approaches to process automation
ApproachCore ideaWhen it fits
Built-in features of existing softwareRules, templates, and notifications inside the tools you already useWhen the workflow stays within a single system
Off-the-shelf integration platformsPre-built connectors between popular software toolsFor simple, standard data handoffs
Custom API integrationTailored connection built on the systems' programming interfaces (APIs)For complex rules, high data volumes, or business-critical workflows
Robotic process automation (RPA)Software clicks and types on the user interface like a personWhen a legacy system has no programming interface
AI-powered processingInterpreting free text and unstructured documentsWhere inputs are unstructured and human review is in place

Always start with the simplest tool that solves the problem. Many companies are surprised by how much their existing software can already do once properly configured. Custom development is warranted when the workflow is business-critical, the rules are complex, or the limits of off-the-shelf tools get in the way of daily operations.

Steps to implementation

  1. Pick a single workflow with a clear owner and a measurable outcome.
  2. Record the baseline: how much time it takes and how many errors occur.
  3. Simplify the workflow and write down its rules, including exceptions.
  4. Automate the main path; leave rare exceptions to people.
  5. Set up error alerts: if the automation stalls, someone must be notified.
  6. Measure again, and only then move on to the next workflow.

Automated workflows need maintenance too: when connected systems change, the integration must follow. Make sure every automation is documented and understood by more than one person.

Key takeaways

  • The best first candidates are frequent, rule-based tasks built around copying data.
  • Simplify the process before you mechanise it.
  • Choose the simplest tool that reliably solves the problem.
  • Automate the main path and leave rare exceptions to human judgement.
  • Without error alerts and documentation, automation becomes an operational risk.

Frequently Asked Questions

What is the difference between automation and artificial intelligence?

Traditional automation follows pre-written rules and does the exact same thing every time. AI helps where inputs vary (such as free text, documents, or images), though its output requires verification. The two complement each other well.

What is RPA, and when should we use it?

RPA (robotic process automation) is software that operates other programs through their user interface by mimicking human clicks and keystrokes. It is useful when a legacy system has no API. Where an API exists, direct integration is almost always more durable.

How do we know whether it paid off?

By measuring before and after: time spent on the process, error counts, and turnaround time. Without a baseline measurement, you will only have impressions of return on investment.

Facing a similar decision? We are happy to talk it through.

Let's talk