If a digital investment doesn't show in your numbers, it didn't happen
21 September 2026, 12:36
The Economist asks an honest question: how quickly do investments in artificial intelligence turn into gains for everyone, and not just for those who sell the chips? It looks to Taiwan and South Korea for clues, because there the chain between those who make the technology and those who use it is short.
It is the right question, and it holds three rungs further down too. In the Brescia version it goes like this: that €40,000 of software you put in last year — does it show up anywhere in your accounts?
Almost always the answer is an embarrassed silence. Not because the investment was wrong, but because nobody had decided beforehand where it was supposed to show up.
The boom does not trickle down on its own
The idea that technology spreads by gravity — first the big players, then the small ones, first those who make it, then those who use it — is convenient and often false. Between the investment and the benefit stands the organisation, and that is where most of the theoretical gain gets lost.
In SMEs I see it in a very specific form: tools pile up that don't talk to each other. Management software, one package for quotations, another for production, spreadsheets everywhere, and data copied by hand from one system to the next. The result is the paradox of digital that slows things down: more time, more errors, more dependence on someone who knows how to hold the whole thing together.
Many companies today are not suffering from a lack of technology. They are suffering from too much badly managed technology. The challenge is not to go more digital, it is to go digital better.
The sum nobody does before signing
A technology investment is assessed like any other investment, and the numbers are always the same three: how much margin it frees up each year, how many months it takes to pay for itself, and whether the debt that comes with it stands up.
The first is the one that always gets skipped, because it seems impossible to estimate. It isn't, if you stop reasoning by gut feeling and pick a single, measurable quantity, before signing. A few examples from real life: the hours the technical office takes to prepare a quotation; the days between order and delivery; the reject rate on a machining process; the days of stock sitting in the warehouse.
Then the starting value goes down on a sheet of paper, with the date. And a check is booked for six months later.
If after six months that number hasn't moved, the investment has produced nothing, whatever the person who sold it to you says and however much more modern the company genuinely is than before. If it has moved, you have something valuable in your hands: you know which lever worked, and you can push there instead of buying the next tool blind.
Why order matters
There is a practical reason why technology pays off so little in companies that did not get organised first: tools amplify processes, they don't correct them. If the way quotations are prepared is sloppy, new software makes it sloppy faster.
The same goes for artificial intelligence, which today is the thing everyone feels duty-bound to buy. Fed messy data, it produces messy answers, with the added drawback that they sound authoritative.
What to look at tomorrow morning
Take the last technology investment you made — software, machinery, automation, a licence, whatever it was — and answer a single question: which number in my company was supposed to change?
If there is no answer, don't repeat the mistake with the next purchase: before signing, choose the number, write it down and put the date on it. It takes five minutes and completely changes the conversation with whoever sells you the next solution.
This article was prompted by a news item published in The Economist.