THE BEST SYSTEM
IS THE ONE
NOBODY NOTICES.

01

THE DEMO PROBLEM.
WE HAVE BEEN ON BOTH SIDES OF IT.

There is a kind of system that demonstrates beautifully. It has a screen with everything on it. It has colours that change. It has a view the owner has never had before and cannot stop looking at for the first fortnight. Everybody leaves the meeting pleased.

Then there is a Tuesday in February. A customer has moved a delivery, two people are out, and something has arrived short. On that Tuesday nobody opens the screen with everything on it. They do the thing that gets the day finished, and the system quietly becomes a thing that is shown to visitors.

We have built both kinds. The difference between them is not sophistication and it is not budget. It is whether the thing was designed for the day it gets shown or the day it gets used.

A system that has to be remembered has already lost. The work does not pause so you can go and update something.

02

WHAT WE TRY
TO MAKE DISAPPEAR.

The honest goal is that nobody can point to the thing we built, because it is not a destination any more. Information arrives where somebody was already going to be looking. A record gets created because work happened, not because someone remembered to record it. A number is simply correct when it is next read, and nobody can remember who assembles it, because nobody does.

That is a strange outcome to sell. There is no screen to show. The best review we get is somebody saying they are not really sure what changed, only that they stopped chasing a particular thing a few months ago.

03

THE THINGS WE HAVE LEARNED
TO TAKE OUT.

Most of what we have learned in operations is subtraction. Every field we ask someone to fill in is a small tax collected on the busiest day, and it will be paid with a guess. So we ask whether a field has ever changed a decision, and if it has not, it goes.

Alerts go the same way. A notification that fires on a normal day teaches people to ignore notifications, and then the one that matters arrives into a habit of ignoring. We would rather have four alerts a month that are all real than forty that are mostly noise.

Dashboards are the hardest to argue about, because owners ask for them specifically. The useful version is small and answers the two or three questions that get asked every week. The version with twenty tiles on it gets looked at until the novelty goes and then it becomes wallpaper, and worse, it becomes a reason not to fix the thing that made the number hard to find in the first place.

If the only way to get the number is to open something, the number is still inconvenient. We are trying to make it unavoidable instead.

04

WHERE RESTRAINT IS
THE WRONG INSTINCT.

We should be fair to the other side of this, because taking things out can become a pose. There are places where a company genuinely needs something visible, and we get this wrong by under-building about as often as we get it wrong by over-building.

An owner who has never been able to see the state of the operation without ringing three people does need a place to look, and telling them the information is distributed through the workflow is not an answer. Somebody taking over a function for the first time needs more scaffolding than the person who has run it for twenty years, and that scaffolding should not be removed on principle. A regulated or safety-critical step should be deliberately difficult, slow and obvious, because the friction is the point.

The rule we try to hold is narrower than minimalism. Remove what is paid for on the busiest day and never read. Keep what someone needs in order to decide. Those are different tests, and the second one is the one that stops a system becoming so lean that it only serves the people who already know everything.

05

HOW WE DECIDE
WHAT A SCREEN IS FOR.

Before we build anything anyone has to look at, we try to write down the question it answers and the decision that follows from the answer. If we cannot name the decision, the screen is a comfort object and we are building it because it was asked for rather than because it will be used.

That one habit has killed more of our own ideas than any amount of budget discipline. It also tends to shrink what survives. A screen that answers one question well is opened every day. A screen that answers twelve is opened the week it ships and then lives in a browser tab nobody has closed.

Name the decision first. If you cannot, you are building furniture.

06

WHY THIS MATTERS MORE
WITH AN INTELLIGENCE LAYER ON TOP.

This gets more important, not less, as companies start putting automated judgment on top of their operations. A screen that is wrong is annoying. A system that is acting on its own and is wrong is a different category of problem, and it will be wrong in exactly the places where the record was being filled in with guesses.

Which is why we do operations first and always will. Not out of caution, and not because the newer technology is unproven. Because anything built on top of a record inherits the quality of that record, and most companies have never had the chance to find out what theirs is actually like.

The work we are proudest of is the work nobody mentions any more.