Mindset Data Intelligence Lab

Everything Is an Experiment.

We do not act on impulse alone.
Before we act, we make a hypothesis. After we act, we check the result.
We move forward by testing what is true.

Experiment rule 1

Draw the line first

Before starting, write one sentence that defines success and failure. Without that line, we only did something.

Experiment rule 2

Check honestly at the end

Do not just ask what happened. Ask, "Compared with xxxxx, how did it do?" Look at the result plainly, not conveniently.

Numbers only mean something
when we compare them.

"We sold 1 million yen" -- is that good or bad?
Without something to compare it against, there is no answer.

How we use our heads
5 Habits

1

Treat everything as an experiment with a hypothesis

Before doing it, put the purpose into words.
2

Set the line for success and failure before you begin

No retroactive "that was pretty good." The criteria come first.
3

Numbers matter when they are compared

Not just "it went up," but "compared with what, and by how much?"
4

"We tried it" is only halfway

Check the answer, form the next hypothesis, and then the loop is complete.
5

Assume AI answers need checking

The smarter the tool, the more we need the habit of not swallowing its output whole.

When we hand work to AI, whether it is a big strategy or a production task, we always pass along both what correct means and how we will verify it.

Our work is shifting from "producing the right answer"
to creating good questions and fair scoring criteria.

AI can bring us marketing knowledge.
Deciding where to aim, and whether the result can be trusted -- that is our work.