At which point do you stop automating repeatable tasks? As in, what tasks aren’t worth automating? My boss wants me to automate all SQL Server database repeatable tasks.
I use a financial calculation for this. Let’s say the people currently performing the task make $100k a year, and that you’re based in the US. A fully-loaded salary in the US (including benefits and payroll taxes) runs about 40% more, so to the company, they’re worth $140k a year. We have roughly 2,000 working hours per year, so they’re worth $70/hour. So every hour they spend on a task costs the company $70.
Suppose I make the same money (because the math is easy on those numbers). If I can spend ten hours automating something, that costs $700, but it reduces the number of times I have to pay the other people to do it. If something used to take them an hour (costing $70) and now only takes them a minute, I’ve saved money pretty quickly. I might even need fewer people, which means I can deploy them to other tasks that actually require human thought.
I also look at the danger of having someone do something manually. If there’s room for mistake (like forgetting a WHERE clause in a DELETE statement), I’d prefer to automate it to remove that mistake.
If you look at those two factors – money and danger – you usually find that “automate all the tasks” is the right answer.
You also have to think about automation triggers. If some kind of task can be triggered by an automated monitoring process, then it makes all kinds of sense to automate that task. That way, much less human involvement is required. So, if you’ve got a task that can only be triggered by a human, and usually only takes that human a minute or two to perform, and usually doesn’t involve a lot of room for mistakes… that might not be worth automating.