Automation is not an end in itself, and not every firm gets the same out of it. These four signals reliably say there is time to be won — and the three after them say you are better off waiting.
1. The same task, more than a hundred times a month
The main criterion is not size but repetition. A three-person firm processing four hundred purchase invoices the same way every month has more to gain than a fifteen-person firm where every file is approached differently.
Practical test: if you can describe a task in the form “first this, then that, except when…” and that description fits on half a page, it is automatable. If you need three pages and an exception crops up every other sentence, the answer is usually no.
2. Your staff retype data that already exists digitally
This is the clearest signal, and surprisingly often present. An amount from a PDF into a posting screen. An address from an email into a client record. Hours from one programme into another.
Every time a human moves data between two screens, an integration or a workflow is in principle waiting to be built. That work adds nothing: the information already existed, it was simply in the wrong place.
3. There is a peak in your month or quarter you see coming every time
The week before the VAT return. The first ten days of the month for reporting. Autumn for annual accounts. If you can draw those peaks on a calendar, they are predictable — and therefore largely movable to software that knows no peaks.
In this category the gain often is not even measured in hours but in stress. A firm whose files are complete four days earlier works fundamentally more calmly, even if total time spent stays the same.
4. You catch yourself saying: “only colleague X knows that”
Knowledge that lives in one head is both a risk and a brake. Whoever goes on holiday takes the follow-up of thirty files with them. Whoever leaves takes it for good.
Automating forces you to make that knowledge explicit: which documents belong to which file type, which supplier goes to which account, when to chase. That is in fact the most underrated benefit of an automation project — even the parts you ultimately do not automate finally end up written down somewhere.
And when to wait
If your processes change every quarter
Building a workflow on a process still in motion costs money twice: once to build, once to rebuild. Let the process settle first.
If you are mid-migration to a new package
Wait until you know which accounting package you will be working in next year. Integrations are package-specific; carrying them over to a new one usually means building again.
If the trigger is “the competition is doing it too”
Automating because you feel you must rarely delivers anything. Automating because you can point to a concrete task that costs too much time, almost always does. If you cannot say in one sentence which work you want to see disappear, the conversation is not ripe yet.
Rule of thumb: start with the task your staff complain about most. That is almost always simultaneously the most repetitive, the most error-prone and the most automatable. It also builds support — the first workflow has to make somebody's day noticeably better, otherwise there will not be a second.
Where to start
Take one task. Count how often it occurs per month. Estimate how many minutes it costs. Multiply. If that comes to more than half a working day per month, it is worth investigating.
That is exactly the exercise I do with you during a scan, for every task that qualifies — so you can choose which one goes first.
Want to do that exercise together? See how a project runs or request a free workflow scan.