Why Manual Processes Keep Growing Even as Companies Go Digital

A company rolls out an ERP, moves invoicing to the cloud and puts everyone on Teams. Three years later there are more spreadsheets in circulation than before. The manual processes did not disappear. They moved.

This is not a lack of commitment or a bad software choice. It is the predictable result of how digitalisation actually happens in most SMEs: one tool at a time, each solving its own slice well, none responsible for what sits between them.

Going digital does not force the work to disappear

A process is digitised when it stops existing on paper. It is automated when recording, checking and passing information along no longer needs a person.

Those are different things, and they are easy to confuse, because the first is visible and the second is not. The PDF replaces the paper, the form replaces the printed sheet, and the job of copying data from one place to another stays exactly where it was — now with a screen in the middle.

The way digitalisation gets measured feeds the confusion. The Eurostat Digital Intensity Index counts the adoption of 12 technologies and treats a score of 4 as "basic" level. In 2024, 73% of EU SMEs reached that basic level, but only 6% hit very high digital intensity and 27% remained in the lowest band.

The index counts tools installed. It does not count whether information moves between them without someone carrying it.

Every new system adds a seam

This is the mechanism behind the growth. Each new application solves a problem inside its own boundaries and creates a new boundary with everything else.

Okta's Businesses at Work 2025 report found that the average number of apps per customer organisation reached 101, passing 100 for the first time after years of broadly flat growth. In an SME the number is much smaller, but the effect is identical: tools grow in a straight line and the contact points between them grow far faster.

The seams are where manual work shows up. Nobody hires a person to "copy data from the CRM into invoicing". That work installs itself, in two-minute slices, the day someone needs an approved quote to reach the invoicing system and discovers there is no connection between the two.

That is why integrating CRM, email and invoicing without creating duplicates usually pays back better than buying one more tool.

Manual work hides in small slices

A two-hour manual process is visible. Someone complains, someone raises it in a meeting. A ninety-second manual process repeated forty times a day complains to nobody.

A 2022 study published in Harvard Business Review measured precisely this. It tracked 137 users across three Fortune 500 companies and found an average of nearly 1,200 toggles between applications and tabs per day. Reorienting after each jump alone consumed close to four hours a week — roughly 9% of annual working time.

Note that this is not time spent doing the work. It is time spent finding your place again.

An example: the invoice that went digital and stayed manual

Take a typical SME handling 300 supplier invoices a month.

Before digitalisation: the invoice arrives on paper and someone keys the data into the accounting software. Around 6 minutes per document.

After digitalisation: the invoice arrives as a PDF by email. Someone opens the email, saves the PDF to the shared folder, copies the number, date, tax ID and amounts into the software, checks it against the purchase order and files it. Around 5 minutes per document.

One minute saved. The process now counts as "digital" and drops off the priority list. At 300 invoices a month that is still 25 hours a month — 300 hours a year — of work nobody budgeted for, because nobody calls it manual any more.

The per-document times above are the estimate we use in discovery, measured with the team before proposing anything; the total is simply 300 × 5 minutes ÷ 60. It is good for an order of magnitude, not for a slide deck without measuring your own case.

The fix here is not better accounting software. It is connecting the mailbox to the software already in place: extract the fields from the PDF, validate against the purchase order, post it, and route to a person only what failed the check. The goal is not doing the same thing faster. It is not doing it at all.

To put numbers on your own case before deciding, the method is set out in the article on how much manual processes actually cost.

Why this never shows up in a report

Three reasons, all structural.

Manual work has no owner. It is spread across ten people in small slices, and none of them has it written into their job description.

Software is bought per department. Each purchase is justified by the gain inside that department. The integration cost falls outside the buyer's calculation.

Capable people paper over the gaps. A good team compensates for missing connections without complaining. The problem never reaches management as a problem — it arrives as "we are short-staffed".

What actually stops the growth

Three practical decisions, in order of effort.

Measure processes end to end, not per tool. The right question is not "do we have an ERP?". It is "how many times is this piece of data typed by a person between the customer order and the payment?". If the answer is more than one, there is hidden manual work.

Automate the seams before buying tools. Connections between systems are almost always cheaper to fix than replacing any of those systems, and the payback arrives in weeks rather than quarters. Platforms such as n8n exist for that layer and can run on your own server, which matters when the data cannot leave company infrastructure.

Set an entry rule for new software. Before approving another subscription: what data does this tool need to receive, where does it come from, and who is going to put it there? If the answer to the last question is "a person", you have just bought a new manual process.

If you need concrete candidates to start with, the list of business processes worth automating is a reasonable starting point.

The question to ask in your next meeting

Not "are we digital?". That one already has an answer, and the answer is yes.

This one instead: take the process that generates revenue, from customer order to payment, and count how many times a piece of data is re-entered by hand along the way. That number is the real size of the manual processes your digitalisation never solved — and it is where to start.