The numbers were telling us something. We found a way to make them say something else.
The most dangerous evidence is not the evidence we fail to see. It is the evidence we see clearly enough to explain away.
Numbers do not walk into a meeting and announce their meaning. People interpret them, qualify them and place them inside a story. We like to believe we are asking what the evidence is telling us. Often we are asking something else.
How can this still mean the plan is right?
I learned that during an integration in a transport business. We were bringing several acquired companies into one operation, each with its own systems, habits and local ways of getting things done.
One of the big ideas was a centralised dispatch system.
Until then, local dispatchers decided which driver received which job. Their decisions were shaped by availability and geography but also by preference, history and relationships. From a distance some of it looked inefficient. Some of it looked like favouritism.
The new system would distribute work evenly, improve efficiency and give management better visibility.
It was clean, rational and fair.
So we pushed ahead.
Six months later the operation was in trouble.
Driver churn was rising sharply. Worse, the people leaving were the ones we most needed to keep. Experienced drivers who knew the difficult routes and who could be relied on when the night turned awkward.
We held an emergency meeting.
“They are resistant to change.”
“They will get used to it.”
“We need to enforce compliance.”
“They are old-school,” I said.
Four words were my whole contribution. They settled it.
The explanations were plausible. That was what made them dangerous. They fitted the story I already wanted to believe. We were modernising. They were resisting. We had the data. They had habits.
We did not ignore the numbers. We made them mean something more comfortable.
Then we carried on for another six weeks while drivers kept leaving.
Eventually I grew tired of confident explanations from people nowhere near the problem.
So I drove to one of the depots my managers had quietly labelled a problem child.
In the back office sat a dispatcher called Dave. He was entirely unimpressed by my arrival. I brought him a cup of terrible machine coffee and asked the question I should have asked six weeks earlier.
“Dave, drivers are leaving. I am being told the system is better, but something is clearly wrong. What am I missing?”
He replied without hesitation.
“You are not missing anything, Mr Vinall. You are just wrong.”
I asked him what he meant. The system was fair. It removed personal preference and gave everyone a clearer process.
Dave pointed at the screen. There was a good airport run. The system had allocated it to a new driver who had logged on twenty minutes earlier.
“Fair,” Dave said. “Stupid, but fair.”
Before the new system he would have given that job to Steve, one of his best drivers. Steve had three children and his wife had just lost her job.
If Dave looked after Steve when the good work appeared, Steve would take the dreadful run on a Friday night when nobody else wanted it. Not out of obligation to the company. Out of obligation to Dave.
“It is not the system, sir,” he said. “Steve owed me one. Now he does not.”
The favouritism I had set out to remove was the payment mechanism.
The old rota had been running an informal ledger. Good airport run now, miserable Friday night later. Nothing recorded it, no manager could see it and from above it looked exactly like bias. We had built a system for visibility and it was blind to the only thing holding the rota together.
We had not replaced inefficiency. We had replaced loyalty.
The algorithm was not wrong about everything. Local judgement did create real favouritism and the business did need to scale. But we had confused what the system could measure with everything that mattered.
So we changed it. We kept the algorithm and let dispatchers override it where loyalty, reliability and the realities of a bad shift could not be reduced to what was on the screen.
The answer was not technology or people. It was technology with people still allowed to think.
Churn stabilised over the following weeks and the depots felt different.
But I do not know how much of the recovery was mine to claim. Other things shifted too. Business recoveries rarely belong to one decision, however neatly we arrange the story afterwards.
It would be easy to end here with a satisfying number. That would repeat the original mistake. It would take evidence and make it serve the story I wanted to tell about myself.
First as the moderniser.
Then as the leader humble enough to listen.
Both versions place me comfortably at the centre.
The more honest lesson is less flattering.
I backed a system because its logic matched the kind of progress I believed in. When evidence challenged it I accepted explanations that protected the plan. Those explanations were coherent, professional and supported by people whose judgement I respected.
They were also wrong.
That is how competent people stay wrong for longer than they should. They do not reject the evidence. They fit it into a story that leaves their identity, investment and authority undisturbed.
Poor sales become proof that the market needs educating.
Complaints become proof that customers dislike change.
Good people leaving becomes proof that standards are finally being raised.
Sometimes those explanations are true. That is the difficulty.
The explanation that protects your ego does not have to be absurd. It only has to be plausible enough to prevent the harder question.
What if the plan itself is causing the problem?
One sentence from that depot has stayed with me.
“You are just wrong.”
Not misguided forever. Not incapable. Wrong about this.
The numbers had been telling us something. Dave helped me understand what they meant.
Field Note
The evidence that most needs your attention may be the evidence you have already explained.
When a result threatens a decision you championed, ask whether your interpretation is helping you understand reality or helping you remain right.
Try This
Choose one disappointing result in your business. Write down your current explanation for it.
Then ask: what would this evidence mean if the decision I defended is part of the problem?
Before commissioning another report, speak to someone close enough to the work to tell you what the system cannot see.