Lesson 25 of 30 · 9 min
Write the measurement plan first
Decide what would count as proof before the data arrives.
One primary metric, a few supporting ones
Name the single number that decides whether the plan worked, and keep the rest explicitly secondary. Plans with five equal priorities have none, and they always end up being judged by whichever metric looks best afterwards.
Write the thresholds down in advance
State what result would count as success, what would count as failure, and how long you will wait. Deciding after the fact is how noise gets promoted to insight and how a mediocre campaign becomes a success story.
Agree the reporting window before launch
Conversion lag is real: a considered purchase reported at day three looks like a disaster and at day thirty looks fine. Fix the window up front so nobody gets to pick the one that flatters their argument.
Say what you will not be able to prove
Naming the limits — no cross-device view, no offline sales, no incrementality test this quarter — protects the plan from being asked to answer questions its measurement was never able to answer.
Takeaway
A measurement plan written before launch is what stops results being reinterpreted after it.
Check yourself
No score, no signup — pick an answer to see why it's right.
Question 1
Why fix the reporting window before launch?
Question 2
What is the point of naming what you cannot prove?