← Back to the blog

AI & automation

Before you automate a small business task, answer these five questions

The best automation candidate is a task you already understand, with an owner, a review step, and a way to keep working if the tool stops.

Illustrative scene of Rikki reviewing a workflow on a laptop with a colleague

“Can we automate this?” is a good question, but it is rarely the first one. A tool can repeat a clear task and save time. It can also repeat confusion faster, move incomplete information downstream, or make it harder for the team to understand what happened.

Before choosing software or AI, walk through one real task with the person doing it and answer these five questions.

1. Is the task repeatable enough to describe?

Write down what starts the task, the information it needs, the normal steps, and what counts as a complete result. If the answer changes dramatically each time, you may need a clearer process or human judgment before automation.

2. Is the information trustworthy and appropriate to use?

Where does the input come from? Who can access it? Is it current and complete? Decide what information a tool may handle and what must stay out of it. If the task involves customer or employee data, review privacy, access, and any obligations that apply to your business before connecting systems.

3. Who owns the outcome?

Assign a person to review the workflow, handle exceptions, and decide when a result is ready to use. An AI draft, for example, can help prepare an internal summary; a person still needs to check its facts and decide whether it should be shared.

4. How will you test it?

Use normal examples and awkward ones: missing information, a late change, a duplicate request, or a result that looks plausible but is wrong. Compare the new process with the current way of working. Define what you want to improve, such as fewer repeated entries or faster routing, and watch for errors or new work created by the tool.

5. What happens when the tool is unavailable?

Important work needs a documented fallback. The team should know how to finish the task manually, where to record what happened, and when to escalate. A fallback is part of a dependable process, not an admission that the technology failed.

Example: A service team might automate the routing of routine incoming requests. It can test several request types, assign a person to review exceptions, and keep a manual inbox process for outages. Whether that saves time depends on the team's actual volume, quality of intake, and how often requests need judgment.

Try this this week

Choose one repetitive task. If you cannot name its trigger, owner, review point, and fallback in a few sentences, document the manual workflow first. You will have a better basis for deciding whether automation is worth the effort.

Business Progress Solutions can help you improve the process, test a useful tool, and train the people who will use it. Read our AI approach or explore a task together.

LET’S TALK

What would make work easier?

You do not need a finished plan to start a useful conversation.

Talk it through with us