The way to decide which office tasks to automate is not to read a list of tasks other offices automated, it is to look at your own week, because the tedious thing in your office is probably not the tedious thing in anybody else's. This page is a method rather than a list. It takes about a week of light record keeping and an hour of arithmetic, and it produces three candidates ranked by what they actually cost you, which is a far better shopping list than any vendor comparison.
Log the week, roughly, for five days
Keep a note of every administrative job you do and roughly how long it took. Rough is fine and precision is counterproductive: five minute granularity, a line per job, no categories. Do it for a full week including the day the invoices go out and the day the post is heavy, because a Tuesday is not representative. At the end you will have between forty and ninety lines, and the surprise is almost always which small job appears fifteen times rather than which big job appeared once.
Score each line on three things
How many times it happened, how long each one took, and whether the steps were the same every time. Multiply the first two for the monthly cost in minutes. Use the third as a gate rather than a score: a job that varied is not a candidate however expensive it is, because automating a varying job produces exceptions that somebody has to handle, which is the job again with extra steps. What survives the gate, sorted by monthly minutes, is your list.
Take three, and leave the rest alone
Three is the number a small office can change at once without losing track of which change caused what. Do them one at a time, a fortnight apart, and write down what each was supposed to remove so you can tell afterwards whether it did. Then stop and run the week again in six months. The tasks below the top three are not worth the disruption, and an office that automates everything on the list has spent more attention on the automating than the tasks were costing.
Questions people ask about automate office tasks
Is a week long enough to log?
For most offices yes, provided it includes the month's heavy days. If your work is genuinely monthly, log the month rather than the week; the effort is the same per day and the answer is better. What does not work is logging a quiet week and generalising from it.
What if the biggest cost is something we cannot automate?
That is a useful result, not a failure. Knowing that most of your week goes on judgement calls or interruptions tells you the answer is scheduling or delegation rather than software, and it saves you a subscription that would not have touched the problem.
Does the paper worksheet fit this?
It costs one specific slice of the week, the paper handling and the searching, in hours and money, from your own counts. If the log shows paper near the top, the worksheet turns that line into a figure you can take to whoever approves the spend.