I couldn't move the executives with intuition. Only by turning it into numbers could we have a conversation.
I knew it intuitively. But I couldn't take action until I turned it into numbers.
I knew that our company had extraordinary losses. Fraudulent uncollected debts and price discounts due to quality claims. I had the numbers for both in my head. There was no way the management didn't know about them.
But there is a gap between "knowing" and "being able to act."
You cannot move people with intuition alone.
If you just say, "I feel like SG&A expenses are a bit heavy," the executives won't know what to change. Only when you present it in a form like, "This expense item is costing this much per year, so we will cut it next term," can you have a conversation.
Presenting numbers is not for your own sake, but to move the team.
Lining up the data
I deciphered the daily cash flow statement and had it aggregated mechanically. It deviates slightly from accrual accounting, but I wanted to reliably line up the facts of cash actually leaving the company.
I will deliberately hide the detailed numbers, but "extraordinary losses" were the main cause of the final deficit. Fraudulent uncollected debts and claim discounts from a large project in the previous term. The two numbers I had in my head were confirmed on the screen.
Confirmation that "it was as I thought"
Looking at the numbers lined up by expense item, the first thing I thought was, "It's just as I thought."
About 60% is related to personnel costs. SG&A expenses are determined by "people." Fixed costs are stable, and I didn't see any major waste. There were no surprises. But that was a good thing.
The confirmation that "my intuition was correct" becomes the confidence to move forward. Conversely, if it had been way off, I would have had to doubt my own management sensors.
Presenting numbers is also a way of testing yourself.
Raising an AI that doesn't sugarcoat things
I often hear people say lately that "AI has started to understand me." But what I am doing has the opposite goal.
I am accumulating my daily thoughts and judgments in a tool called Obsidian. It is a "second brain" where I verbalize and store my management statements, ways of thinking about issues, and the basis for my decision-making.
The AI that runs within it knows me well. That is why it doesn't sugarcoat things.
It doesn't say, "You are right." If I am wrong, it points it out. It poses questions from angles I have overlooked.
It was the same this time. To me, cowering before the deficit numbers, the AI asked back, "Is that a structural deficit? Or is it a wound from a specific factor?" The operating profit was positive. If you add back the extraordinary losses, it effectively exceeds break-even. It is not a hemorrhaging deficit. The core business is running, and it was just hit by a specific wound. Yes, the numbers talked back.
Why does this happen? Because that is how I am training them.
An AI that acts to avoid displeasing its master is useless in management.
From numbers that judge the past to numbers that steer the future.
I have identified the nature of the wounds. Now it is time to make those wounds visible in real time.
If it ends with a one-time analysis, we will just return to relying on intuition. That is why I started a system to track funds on a weekly basis. I consolidate the daily cash flow every week and link it to the monthly cash flow statement. From numbers that judge the past to numbers that steer the future.
The wounds we have sustained already have names. For fraudulent non-payment, we use credit management. For low gross margins, we use project selection and cost management. For complaints, we use quality control. One by one, I am turning them into levers.
A tool cannot exceed the quality of the person using it.
DW
*Next time, the story of how I built an evaluation system from scratch. I will present it internally next month.*