Automation often appears near the top of the priority list for people building products, running teams, or growing businesses. The appeal is obvious. Whenever we encounter a repetitive task, our first instinct is often to ask whether it can be automated.
Over time, I have found that the more important question is not what can we automate, but what should we automate.
Not every repetitive process justifies an automation effort. In many cases, building automation too early introduces complexity, maintenance costs, and operational overhead that end up outweighing the original problem. I have seen teams spend weeks automating workflows that only occurred a few times per month while much larger customer problems remained unresolved.
Automation creates value when it removes repetitive work, reduces errors, improves consistency, or frees people to focus on higher-value activities. If a process consumes hours every week, requires constant manual intervention, or directly impacts customer experience, it is usually a strong candidate for automation.
On the other hand, when a process is still evolving, when the workflow is not yet well understood, or when the frequency is low, keeping it manual is often the better choice. Automating a poorly designed process simply allows an organization to make the same mistakes faster.
This is especially true in the early stages of a product or business. Before automating, it is often worth spending time understanding how the work actually gets done. Talking to customers, handling requests manually, and experiencing operational pain points firsthand can reveal insights that no workflow diagram or requirements document will capture.
The rise of AI has made automation more accessible than ever. Today, it is possible to connect systems, classify information, generate content, answer customer questions, and orchestrate complex workflows with relatively little code. Yet the underlying principle remains unchanged. The fact that something can be automated does not automatically mean it should be.
One question I often ask myself is simple: if this automation disappeared tomorrow, how much pain would it cause? If the answer is “not much,” then it may not be solving a meaningful problem. If the answer is “the team loses hours every day,” “customers are impacted,” or “operations slow down significantly,” then there is probably real value to be unlocked.
The most valuable automations I have seen are not necessarily the most sophisticated. In many cases, they are small improvements that remove friction from everyday work: automatically generating reports, synchronizing data between systems, eliminating duplicate data entry, or streamlining administrative tasks that add little business value.
Automation should be viewed as a tool for amplifying value, not as a goal in itself. Before building any automation, it is worth understanding where the actual bottleneck exists and whether solving it will create measurable benefits for customers, employees, or the business.
Sometimes the right decision is to automate. Other times, the right decision is to wait, continue learning, and allow the process to mature. Knowing the difference is often what separates useful automation from unnecessary complexity.