Baseline it before you change it

The third move in the method is the one everyone agrees with and nobody schedules. Find the work. Write the gate. Baseline it before anything changes.

I have watched rooms skip the third because it feels like delay. It is the opposite. Without a before, every after is a story. You will get a dashboard. You will not get a fact.

What to measure

Not model quality. The path. How long the case sits, where it waits, how many people touch it, who actually says yes, and what happens when it is wrong. Do this on ten real cases, including the ugly ones, not on the example that was used to buy the licence.

If the work lives in WhatsApp, measure the WhatsApp. If the write is a disbursal, measure the break and the time to undo it. If the write is a publish, measure the correction. The unit has to be the same unit you plan to claim later.

What I got wrong

I have shipped the visible path first and promised to instrument it after it settled. It does not settle. People adapt around the new thing, the old queue remains, and you have nothing to compare except memory. Memory is how a 30% becomes a feeling.

Spend a week on the before. Write it down where the approver can see it. Then change one write. If you cannot afford the week, you cannot afford the programme — you are about to spend a year arguing about a number nobody can replay.