The return question, answered honestly for a company of thirty
Davos spent last week asking what AI spending produced. Here is how to answer that for a Swiss SME without a consultant and without pretending.
The honest answer for most companies your size is that AI has saved some time and produced no measurable profit, and that this is a perfectly acceptable result at the amounts involved.
What follows is how to say that with numbers rather than with a feeling.
Why the usual calculation is dishonest
The standard method multiplies hours saved by an hourly rate and produces an impressive figure. It is wrong, for one reason: saved minutes scattered across twenty people do not become money unless somebody either leaves or does something else with them.
Twenty minutes a day saved for each of eight people is not one full-time post. It is eight people with slightly easier days, which is genuinely valuable and is not a line in your accounts.
The version that survives scrutiny
Three categories, kept separate.
Money that changed. A supplier no longer needed, a temporary post not filled, an overtime bill that fell. Rare and unambiguous. Most SMEs have zero of these in year one, and that is normal.
Capacity that moved. Work that used to be outsourced or postponed and is now done in-house. Real, and countable if you name the specific work.
Days that got easier. Everything else. Worth having, worth saying plainly, and not worth converting into francs.
What to compare it against
Your actual spending, which for most Swiss SMEs is a few hundred francs a month plus some of one person’s time. At that level the bar is low and honesty is cheap: if a couple of processes genuinely improved, the spending was justified without any arithmetic.
The trap is the opposite case — a large project with an impressive projected return and nothing in the first category after a year. That is the pattern Davos was discussing last week, and it is largely avoidable at your scale by not starting there.
What to write down
The three categories, with a date, next to the three numbers from 3 October. Repeat in April.
Two data points a year, honestly recorded, will tell you more than any projection, and they are what you will want when somebody asks this question of you rather than of a panel in Davos.