Understanding Automation
Automation is the practice of having software do the predictable parts of a process so people can do the parts that need judgement. Done well, it is unglamorous, invisible, and saves hours every week.
Most business processes contain a mixture of judgement and repetition. Deciding whether a new client is a good fit is judgement. Copying their details from an email into three separate systems is repetition.
Automation targets the repetition. It rarely removes a job; it removes the tedious portion of many jobs, which is where most of the errors and most of the frustration live.
The trap is automating a bad process. If the underlying workflow is muddled, automating it simply produces muddle faster and harder to see.
What is genuinely worth automating
The best candidates share four traits: the task is repetitive, the rules are clear, the volume is meaningful, and mistakes are costly or annoying.
Onboarding a new employee is a good example. It involves a known set of accounts, permissions, and equipment steps, it happens regularly, and missing a step leaves someone unable to work on their first morning — or worse, leaves an account active after they leave.
Joining and leaving are the two highest-value workflows to automate in most businesses, because both are error-prone and both have security consequences.
Where to look for the time
Ask people what they do every week that feels mechanical. The honest answers are usually consistent across a team, and they do not require any technical insight to identify.
Watch for double entry: the same information typed into more than one system. Each additional copy is both wasted time and a place for the two records to disagree.
Look for chasing — reminders to clients, follow-ups on outstanding documents, status updates. These are ideal automation targets because the trigger and the action are both predictable.
And look for anything that has a checklist. A written checklist is often a process someone has already documented but not yet automated.
The infrastructure underneath
Automation depends on systems being able to talk to each other reliably. That means stable networking, consistent identities, and ideally a single directory so accounts are created in one place and flow outward.
It also depends on data being in reasonable shape. Automating on top of duplicated, inconsistent records produces confident wrong answers, which are worse than slow right ones.
This is why automation projects so often begin with tidying: consolidating accounts, removing leavers, and agreeing what the authoritative record for a customer actually is.
Starting safely
Automate one small workflow end to end before attempting anything broad. You learn the failure modes cheaply and you build the habit of checking what the automation did.
Always include a way to see what happened. An automation that runs silently is an automation nobody trusts, and untrusted automation gets bypassed.
Prefer keeping a human in the loop for anything consequential at first. Approval steps are easy to remove later and much harder to add back once something has gone wrong.
Understanding Automation — common questions
Will automation replace our staff?
In practice it removes tasks rather than roles. The usual outcome is that the same team handles more work, with fewer errors, and spends more time on things that need a person. That is the honest expectation to set internally.
Do we need to replace our systems to automate?
Not necessarily. Many of the most useful automations connect the systems you already own. Replacements are only warranted when a system cannot exchange data at all.
How do we avoid creating a fragile mess?
Keep it visible and documented. Each automation should have a named owner, a description of what it does, and a way to see its recent activity. Unmanaged automations become invisible dependencies surprisingly quickly.
More in this series
Understanding Ransomware
What locks your files, who pays, and why backups decide the outcome.
Read the explainer UnderstandingUnderstanding Shadow IT
The software your team uses that nobody approved — and why it matters.
Read the explainer UnderstandingUnderstanding Business Continuity
How long could you be closed before the business is actually in trouble?
Read the explainer