Nobody's job to remember

A test result belongs to nobody, and that is why it dies.

Watch where one goes after the meeting. It is presented on a slide. The slide goes into a deck. The deck goes into a folder with the quarter's name on it. Three months later someone proposes a test, and it is the same test with a different name, and it runs again.

Nobody lied and nobody was lazy. The structure has a person who owns running tests and no person who owns remembering them. Running a test is a job with a name, a budget line and a slot in the weekly meeting. Remembering one is a favour, done by whoever happened to be in the room, until they leave.

I have seen this from the platform side across a great many accounts, and it holds at every size. The teams that ran the most tests were rarely the teams that knew the most. They were the teams with the most slides. The learning had been paid for, sometimes more than once, and it had not been kept anywhere it could be found.

The sunk cost here is the knowledge, not the money. The knowledge was the only thing the money was for, and it was left in a folder because no job description says otherwise.

The fix is small and unglamorous. One log, in one place. One person whose name is on it. A line for every test, written the day it ends, saying what was tried, what happened and what it means. And one rule, which is the part that works: no test starts until the log has been read.

That rule turns a hopeful proposal into a conversation about what is already known, and most teams have never had that conversation, because nobody was ever asked to hold the memory.

A test nobody is responsible for remembering is not an experiment. It is an expense.