Archive for September, 2026

20th September
2026
written by Therese

A business becomes obsessed with efficiency and cost reduction. It looks at the hotel doorman and asks: what is his function?

The obvious answer is: he opens the door.

Once you’ve defined the job that way, the business case is trivial. An automatic door can perform the same function for much less money. Replace the doorman, record the salary saving, and declare the optimization successful.

But opening the door was only the doorman’s notional function.

He was also:

hailing taxis,
providing security,
discouraging vagrants,
recognizing regular customers,
and signalling that this was a high-status hotel.

Those functions are harder to quantify, so the efficiency calculation effectively values them at zero. Yet collectively they may allow the hotel to charge more for a room.

The hotel saves the doorman’s salary, so the decision initially looks successful. But it has also lost the doorman’s less measurable functions: security and deterrence, help getting taxis, recognition of guests, human welcome, and the status signal created by having a doorman.

The result is that the hotel can become less desirable and potentially less valuable. The crucial asymmetry is that the salary saving is immediate, attributable, and easy to put in a spreadsheet, while the lost value is diffuse and much harder to attribute to removing the doorman.

And the consultant who recommended the saving may be long gone by the time those second-order effects become visible.

Lets just take another example. What about the use of AI? It seems like a saving on paper, but what does it signal to the outside world that you are replacing your workers with AI – like the doorman and the automatic door? Are your workers also providing other benefits that are not present in the spreadsheet?

  • You are currently browsing the Complexitymaze blog archives for September, 2026.