What you pay on top of the salary
When you ask yourself what entering data by hand costs you, the first figure that comes to mind is the salary of the person sitting in front of the business software. That's only the part you can see
Underneath there are the real hours spent on the procedure, the checks done afterwards to be sure the numbers add up, the errors to put right, the interruptions of somebody who stops for a moment and then picks up in the wrong place, the time it takes to teach the job to a new person, and the backlog that builds up in the busy periods
Taken once, a procedure that lasts a few minutes looks like nothing. Repeated a few hundred times a month, it becomes a cost line that weighs as much as a person
The sum to do, step by step
The basic calculation is simple. You need the people who do the work, the hours each of them spends on it in a day, the days a month the work comes up, and the company hourly cost, meaning what that hour really costs you all in, not the net figure that ends up in the payslip
Let's say, just to show the method, that it's two people, two hours a day each, twenty days a month. They're numbers made up here, not anybody's real data
Two times two is four hours a day, over twenty days that's eighty hours a month. Multiply those hours by your hourly cost and you have the monthly figure, times twelve and you have the yearly one
Separate the mechanical part from the part that takes thinking
Not all of that total can be automated, and anyone who tells you otherwise is selling you something
Inside those hours there is the mechanical part, opening the document, reading the same fields every time, copying them into the same screen, saving. And there is the part that takes experience, the odd case, the supplier who writes his amounts his own way, the decision you can only make by looking at the context
To gauge how much the first part weighs you don't need an analysis. Look at ten cases in a row and count how many went through smoothly without a second thought, and that proportion is roughly the part an automation can work on. On the rest it stops one step earlier, prepares the data and opens the right screen, then you decide
Errors are the line you won't find in the accounts
A wrong value costs far more than the seconds it would have taken to write it correctly, because first it has to be spotted, then corrected, and it often drags along a phone call, a credit note, one more check across everything else
The more manual steps you line up, the more points there are where something can slip, and you pay for it in the hours of people who should have been doing something else
When automating is worth it and when it isn't
A process deserves an automation when it has volume, meaning the same thing comes up often, and when the rules are clear enough to be written down in black and white
If you do something three times a month, leave it as it is, the time you'd spend explaining it is worth more than what you'd save
The final test is a plain one. What you spend on automating has to stay below what you stop spending, by a margin you can see with the naked eye. If you have to do acrobatics with the numbers to make it add up, it doesn't add up
Do the sum with your own numbers
On the savings page there's a calculator that does this multiplication for you. Put in people, hours, days and hourly cost and you see the yearly figure straight away, and the numbers stay in your browser
It's still an estimate, not a promise, but it's already a lot better than the gut feeling you have now
What it takes to tell whether your case is a fit
To say whether a process can be automated we need very little, which program you use, which steps the person goes through from start to finish, and how many operations a month it comes to. Everything else comes later