"What can we automate?"
It sounds like the obvious starting point. Teams look for repetitive work. Data entry. Approvals. Document processing. Notifications. They find tasks that take time and look for ways to make them faster.
That can create value. But it can also result in businesses automating individual steps without questioning why those steps exist in the first place.
There is a better question to ask first.
What decision is this process ultimately trying to support?
A process is more than a list of tasks
A process might look like a sequence of steps. Receive information. Check it. Enter it somewhere. Compare it against something else. Send it to another person. Wait for approval. Record the result.
Viewed individually, those steps look like separate automation opportunities. But zoom out and something different comes into focus. All of those steps may exist because eventually someone needs to decide: approve or reject, proceed or investigate, act or escalate.
Understanding that decision gives the automation a purpose. Without it, you are making a sequence of steps faster without necessarily improving the outcome those steps were designed to produce.
Start at the end and work backwards
Ask what decision needs to be made. Then ask what the person making that decision actually needs to know. What information supports it? Where does that information currently live? What work is required to prepare it? Which parts are repetitive? Which parts follow clear enough rules that a system could handle them? Which parts require judgment or context that only a person can provide?
Now the system can be designed around what actually matters. The technology follows the decision and the process. Not the other way around.
A simple example
Imagine a company approving a new supplier. The current process involves receiving documents, entering information manually, checking compliance requirements, searching internal systems, reviewing previous records, requesting missing information, and eventually preparing everything for someone to review and approve.
A basic automation project might ask: how do we automate document entry? Useful, but narrow.
A better starting point is: what does the person approving this supplier actually need to know?
Perhaps they need compliance status, any missing documentation, previous supplier history, relevant risks, and a summary of how this supplier meets internal requirements.
With that clarity, the system can be designed differently. AI extracts and organises incoming information. Integrations retrieve existing records automatically. Clear rules handle straightforward checks. Automation routes cases appropriately. The decision maker receives what is relevant and is only drawn into what genuinely deserves their attention.
The technology handles the preparation. The person handles the judgment.
Good automation removes work, not responsibility
Some simple decisions with clear rules and low consequence can reasonably be handled automatically. But important decisions involving judgment, unusual circumstances, accountability, or significant consequences may still require a person.
The objective should not be to eliminate as much human involvement as possible. It should be to ask honestly where human involvement actually adds value. Let technology handle repetitive preparation where it can do so reliably. Let people focus on judgment, context, relationships, and responsibility.
Technology and people are not in competition here. A well-designed system makes that clearer, not less so.
Do not automate a bad process
If a process contains unnecessary approvals, disconnected systems, duplicated work, or steps that exist only out of habit, automating those steps will make a poorly designed process move faster. The problems remain. They simply happen more quickly.
Before building automation, understand the process. Ask why each step exists. Remove what does not need to be there. Simplify what can be simplified. Then automate where technology genuinely adds value.
Do not automate complexity that should have been removed first.
About Wave Group
Wave Group approaches automation by starting where the article started: with the business problem and the decision it needs to support.
Wave Tech brings together custom software, AI systems, data infrastructure, integrations, and automation into systems that work around the actual decision rather than isolated tasks.
Wave Advisory brings a different layer. Understanding the decision itself. Understanding the context the organisation operates in. Understanding where human judgment belongs and where it can be supported rather than replaced.
The combination matters because building automation is only part of the challenge. The harder question is often what should actually be automated and why.
"What can we automate?" is not a bad question. It is simply not always the right place to start.
The best automation does not begin by looking for work to remove. It begins by understanding the decision that work exists to support.