It isn't worth it whenever the sums don't add up, and that happens often
An automation isn't worth it when the job comes up rarely, when the rules change faster than anyone can write them down, when the decision rests on a person's judgment, when the process is about to be changed anyway, and when nobody in the company can explain any more why it's done that way
They're five different cases, and they have one thing in common, the time it takes to build the system is time you never get back, and anyone who sells it to you all the same is selling you hours, not a result
The five cases, one at a time
Low volumes, something you do two or three times a month takes half an hour to explain and ten minutes to do, so the time to describe it, to write down its rules and to test it is already worth more than everything you would save in a year, leave it by hand and don't feel guilty about it
Rules that change every month, an automation is a rule written down, and if the real rule keeps changing the system is always a step behind, so you end up paying maintenance to chase yourself, better to wait for the process to settle and automate it afterwards
Decisions that rest on judgment, if every case is decided by looking at the context, the customer, the history, what the sales rep told you the week before, there's no rule to write down, and a machine that guesses is worse than a person who thinks about it
Processes that are about to change, if you're replacing your business software soon, or you're reorganizing, or that job may disappear altogether, there's no sense building on top of it, wait until you know what it will look like, then we'll talk
Processes nobody understands any more, when you ask why it's done this way and the answer is that it's always been done this way, you have to understand what that job is for before automating it, otherwise you carve into a program a habit that could perhaps have simply been dropped
Why a serious supplier starts here
Anyone who automates processes for a living knows the worst job isn't the difficult one, it's the one that shouldn't have been done, because it sits there, it has to be maintained, and every month it reminds whoever pays for it that it was good for nothing
Muffin Suite works on a monthly fee, which means a useless automation shows up on your invoice every month, and whoever built it sees it too, so starting from the cases where it's better to say no isn't modesty, it's the part of the trade that holds the rest up
There's one more thing that helps you decide, the damage an error does, because a job where a mistake costs little and gets fixed the next day is a good candidate, while a job where a mistake drags an angry customer or a tax problem behind it has to be handled more carefully, with checks around it, and sometimes it shouldn't be touched at all
The other side of the coin, which process is actually worth starting from, is in the guide on which process to start from, while the sum to do with your own numbers is in work out your savings
And if it takes acrobatics with the numbers to make the sums add up, the sums don't add up