Which tasks should an SME automate first?

Why the first automation sets the tone

A company's first automation matters more for what it teaches than for what it saves. If it goes well and the team feels ownership, the second one proposes itself. If it goes badly, "automation" becomes the word attached to a project that complicated everything.

So choose the first one for adoption, not for maximum impact. The most expensive task in the business is usually also the most complex, the one with the most exceptions, and the worst place to start.

Three criteria for prioritising

First, frequency. A ten-minute task done five times a day weighs more than a two-hour task done monthly, and it is far easier to automate because it is better defined.

Second, how much an error hurts. Some repetitive tasks have no consequence when they go wrong; in others a mis-copied figure reaches the customer. The second kind pays off even at equal time.

Third, how many exceptions it has. If the task is always done the same way, it is a candidate. If every case needs a different decision, do not automate it: what is missing there is usually accessible information, not a workflow.

What to leave for later

Anything involving judgement about a commercial relationship. A discount, a deadline exception, the reply to an unhappy customer. Not because it is technically impossible, but because the value of that exchange is that a person handles it.

Also tasks only one person knows how to do and nobody documented. Understanding them comes first, and that work often reveals the process can be simplified before any tool is needed.

How to tell whether it paid off

Before starting, write down two numbers: how many times a week the task happens and how long it takes. They are estimates and that is fine. Without that baseline there is no way to discuss the result with data.

Two months in, the question is not only how many hours came back. It is what was done with them. If the team used them to handle more enquiries or to follow up where they previously could not, it paid off. If nobody notices a difference, you picked the wrong task.