Business process automation: what to automate first and what pays back fastest

Business automation usually starts from the wrong end: a company commissions a good-looking dashboard while the money keeps leaking out of how enquiries are handled. Here is the test for choosing the first process, and the three that turn out to be worth it almost every time.

When a company first starts talking about business process automation, the choice is almost always made on visibility rather than on money. People want a dashboard with charts, an AI assistant, an app. That is understandable: those things can be shown to others. The trouble is that losses rarely sit where things look impressive.

Below is the test for choosing the first process, and the three places where automation pays back fastest in almost any company, whatever the industry.

Where to start with business process automation

A process is worth automating first if it is frequent, done by hand, and expensive when it goes wrong. Lose any one of the three multipliers and the return drops sharply.

Frequent but cheap to get wrong — sending welcome emails, for example. Pleasant to automate, but the manual work costs little.
Expensive to get wrong but rare — the annual stocktake. Once a year it is easier to do it carefully by hand than to build a system.
Frequent and expensive, but no longer manual — there is nothing left to automate; it already runs itself.

When all three come together, though — someone doing the same thing ten times a day, by hand, where every mistake costs a customer or a payment — that is where automation returns the investment fastest.

A practical way to find that process: ask your staff what they do every day and hate doing. People name the places where routine eats their time with remarkable accuracy.

One: taking and routing enquiries

An enquiry arrives by messenger, by email, by phone and through the form on the site. Then somebody has to notice it, reply, write it into a spreadsheet and pass it to the right person. Every step is manual, repeats dozens of times a day, and the cost of an error is a lost customer who simply went to whoever answered sooner.

What automation gives you: enquiries from every channel land in one place, the customer gets an answer immediately, and the person responsible is notified rather than finding out from a group chat. There is no magic in it — what disappears is the gap in which the customer was waiting and drifting away.

One caveat: automating intake only makes sense if somebody actually works the enquiries afterwards. If they already sit unanswered, a system will simply stack them more neatly.

Two: reconciling data between systems

Nearly every company has two or more places holding the same data: the accounting system and the website, the CRM and a spreadsheet, the warehouse and a marketplace. They get synchronised, manually or half-automatically. And they drift apart.

What makes this process particular is that the drift gives no signal. Nothing crashes, the logs are clean, nobody complains — part of your range simply does not sell, and part of your orders arrive for goods you do not have. We covered one such case separately: the catalogue on the site and the catalogue in the accounting system had drifted so far that an entire section stayed invisible to customers for six months.

Automation here is not only the data exchange but the regular reconciliation: a report that states plainly how many items diverged and in which direction. Without reconciliation, any exchange starts lying sooner or later.

Three: reporting to the owner

The third place that nearly always pays off is recurring reports. Usually a person assembles them: exports the data, pulls it into a spreadsheet, works out the percentages, sends it round. That takes several hours a week, and it is the most expensive routine on this list, because the person doing it is rarely the cheapest employee you have.

The main gain, though, is not the hours saved. A hand-made report arrives late and shows only what its author chose to include. An automated one arrives on time and is equally honest about the good and the bad.

A report that cannot tell you what is missing from it looks credible no matter how much data it loses.

That is why any automated report deserves a reconciliation line: how many records were in the source and how many reached the result. It is one line, and it separates a report you can trust from a nice-looking picture.

What not to automate first

Some things look like obvious candidates but return less than they cost at the start of the journey.

A process that is still changing. If the way of working is revised every month, automation will freeze a draft and then have to be rebuilt.
A process with no owner. If nobody is accountable for the outcome, a system will not create accountability — it will create one more screen nobody looks at.
A process with little manual work that merely feels unpleasant. Irritation is a poor basis for a budget.
Everything at once. Automating a whole company in a single pass almost always overruns and loses its point by the time it launches.

How to check your own setup

Four questions give a fairly accurate picture of where to start.

How long does it take from a customer getting in touch to the first reply? More than an hour during working hours means start with enquiries.
Are there two systems holding the same data, and when were they last reconciled in full? No answer to the second half means start with reconciliation.
How many person-hours a week go into assembling reports, and who exactly assembles them? If it is the owner or a senior specialist, start with reporting.
Which process do your staff call the most tedious? It is almost certainly costing you money already — it has simply never been counted.

After that a simple rule applies: automate what repeats every day first, and only then what looks good in a presentation. The first approach pays back in weeks; the second, it depends.