The Three-Times Rule: How to Know If Something Should Be Automated

Most business owners don't know what to automate. Not because they lack options, because they have too many. Everything feels like it could be automated. Nothing feels urgent enough to prioritize.

The three-times rule cuts through that.

If you've done something the same way more than three times, it should be automated.

That's it. One question. Binary answer.

Why Three?

The first time you do something, you're figuring it out. The second time, you're confirming the pattern. The third time, you're just executing. And executing is what systems are for.

By the third repetition, the task is understood well enough to describe. And if you can describe it, you can build it.

Waiting until the tenth or twentieth time is how months of manual work accumulate into a backlog that feels impossible to automate. The three-times rule keeps you ahead of it.

How to Apply It

Go through your week and ask: what did I do more than three times, in the same sequence of steps?

Common answers:

• Replying to the same type of inquiry with a similar response
• Copying information from one place to another
• Sending a follow-up after X days of no reply
• Creating the same document or report with different data
• Notifying someone when something happens
• Scheduling or confirming appointments

If any of those feel familiar, you have workflows waiting to be built.

The Exception

The rule has one exception: tasks that require genuine judgment every time.

Writing a custom proposal for a complex client (that's judgment. Deciding whether to take on a particular project) that's judgment. Having a difficult conversation with a team member, judgment.

Everything else is pattern. And patterns belong in systems.

The honest question most business owners avoid: how much of what I do every day actually requires my judgment versus just my presence?

The answer is usually uncomfortable. Most of it is pattern.

Start With the One That Annoys You Most

If you just ran through the three-times test and came up with a list, you don't need to automate all of it at once. Start with the task that bothers you most when you have to do it.

Not the most complex. Not the most technically interesting. The one that makes you think, every single time: I shouldn't still be doing this.

That's the workflow to build first.