Skip to content
Start a project

How to find a process worth automating

Systems Guide
How to find a process worth automating

How to audit a process on paper before choosing a tool.

It is tempting to start an automation project by choosing a tool. A better start is a pencil and a sheet of paper. Many processes are not worth automating, and many that are need to be simplified first. Automating a muddled process only produces a faster muddle. This guide shows how to find the processes that deserve attention and how to audit them before you spend on software.

Signs a process is a candidate

Look for work that has some of these traits.

  • It repeats often, on a regular rhythm or whenever an event occurs
  • It follows rules that can be written down, even if there are exceptions
  • People copy the same information from one place to another
  • It passes between several people or teams, with waiting in between
  • Mistakes are common and fixing them takes time
  • It depends on one person who knows how it works

Work that is rare, highly varied or based on judgement is usually a poor fit. If every case is different, a tool will either not fit or will need so many exceptions that it is not worth building.

Look at handoffs and error rates

The costly parts of a process are rarely the tasks themselves. They are the handoffs, where work waits in someone’s inbox, and the points where errors creep in. Ask the people doing the work where things get stuck, where they re-enter data and where they check someone else’s work. Their answers are more reliable than a guess from outside.

You do not need exact figures to start. A rough count of how often the task happens and how long it takes is enough to tell a process that costs an hour a month from one that costs several hours a week. If you do not know, measure for two weeks before deciding.

Audit the process on paper

Before looking at any tool, write the process down as it really runs, not as the manual says it should.

  • Name the trigger: what starts the process?
  • List each step in order and who does it
  • Note what information each step needs and where it comes from
  • Mark every handoff and every wait
  • Record what goes wrong, and how often, in the words of the people involved
  • Note the end result and who depends on it

A whiteboard, a shared document or sticky notes all work. Walk through it with the people who do the job. They will correct the diagram within minutes, and the corrected version is the one that matters.

Simplify before you automate

Once the process is on paper, look for steps you can remove. Are there approvals nobody ever rejects? Reports nobody reads? Data entered twice for historical reasons? Removing a step costs nothing and is often the biggest saving. What remains is the part worth automating.

Estimate whether it pays

Compare the cost of the current way against the cost of the change. Count staff time, errors and delays on one side. On the other, count the build or subscription, set-up, training and upkeep. Be realistic about upkeep: automations break when the systems around them change, and someone has to look after them.

Also consider risk. An automation that touches money, customer records or compliance needs testing and a way to see when it has failed. Start with a lower-risk process to learn how it behaves.

Choose a tool afterwards

Now you can choose a tool, and the choice is easier because you know what it must do. The options generally fall into a few groups.

  • Features already inside software you pay for, such as rules, templates or scheduled tasks
  • No-code connectors that link systems you already use
  • Lightweight scripts for a specific repeated task
  • Custom software, when the process is central to the business and off-the-shelf tools do not fit

Start with the simplest option that covers the need. A built-in feature you already own beats a new subscription, and a connector beats custom code until the process outgrows it.

Pilot, then extend

Run the change on a small slice first: one team, one type of request or one week of work. Compare the result against what you wrote down at the start. Ask the people involved what changed and what broke. Fix what you learn before extending the change to everyone.

It also helps to name what you will not measure. Not every benefit shows up as saved hours. Fewer errors, faster replies to customers and less dependence on one person are real gains too, and worth noting in the audit even when you cannot put a number on them.

Keep a human in the loop where it matters

Not every step should be hands-off. Where a wrong result is expensive, keep a person reviewing, and make the automation flag unusual cases rather than guess. Document how the process works now, who owns it and what to do when it fails.

The goal is not to automate as much as possible. It is to remove tedious, error-prone work so people can spend their time on the parts that need them.

Planning a data migration without losing records
Systems

Planning a data migration without losing records

A step-by-step approach from inventory to rollback plan.

What to prepare before you hire a software agency
Software

What to prepare before you hire a software agency

What a useful brief contains and which questions to put to a studio.

Have a question or project in mind?

Tell us what you’re trying to achieve and we’ll help you figure out the best way to build, grow or streamline it.