"Is there anything you can't see?" I asked this in 2013, staring at our brand new business intelligence platform for the first time. It was linked to every scrap of data the business generated. The spectrum of information felt as vast as the universe.
That was the problem.
We could see order patterns down to the hour. We could compare which day drove the highest average order value against any other day on record. It was extraordinary.
And it was useless. Sales started to slide, and nobody could say why.
So we asked for more, and our department head asked for more besides. I asked BI for sales by account, by day, summarised into average order value against the same week last year. One report would support a conclusion. The next would contradict it.
We were drowning in data and losing the plot with every report we pulled.
We had fallen into a trap most leaders never notice they're in. More data doesn't sharpen a decision. Past a point, it just adds noise, and noise looks exactly like insight until you've bet on it.
Before BI, our unit managers carried three or four numbers in their heads. Sales. Margin. Stock cover. That week's trend. They knew, instantly, whether a day was good or bad.
After BI, they couldn't tell you. The numbers hadn't vanished. They were buried under a thousand others.
In the late 1990s, Gerd Gigerenzer and his colleagues at the Max Planck Institute for Human Development ran a strange experiment. They asked people to guess which of two companies had the higher stock value, based on nothing but whether they recognised the name.
The laypeople who simply picked the name they knew built portfolios that beat the market index. In some comparisons, they beat professional analysts running full valuations too.
The experts had more signals. They used them, and it cost them.
More inputs feels safer. It rarely is.
The Less-Is-More Razor isn't an argument for less effort. It's a decision to commit, in advance, to the one cue that has actually predicted the right call before, and treat everything else as a tiebreaker rather than a fresh input.
Take hiring. Most scorecards run twenty competencies deep, and every one drags the decision toward the middle. Pick the one trait that has actually separated your best hires from your worst, and let the rest settle ties only.
That's what our unit managers had before BI arrived, and lost the day it did. Four numbers, held in the head, checked against reality every morning. Not a dashboard. A razor.
Look at the decision on your desk this week. Write down every input currently feeding it. Cross out every one except the single number or trait that has actually predicted the right call before, and decide on what's left.
