A List Isn't a Program
// the backlog
Last session you left with a ranked roadmap. It was ranked on how urgent each finding felt.
That is the right way to end an expert review and the wrong way to run a program. Gut-feel urgency is honest about one page you looked at closely for an hour. It says nothing about which of your forty pages deserved that hour, and nothing about whether you can prove you were right.
A CRO program is three things a list doesn't have:
| a queue | Everything waiting, in an order you can defend to someone who disagrees. |
| a model | A repeatable way of producing that order, so it isn't re-argued every Monday. |
| a cadence | A loop that keeps running when nobody is excited about it. |
the three gates
Every item in a real backlog has passed three questions, in this order. Skip one and you get motion without progress.
| 1 // which page? | Traffic is the ceiling on what any win is worth. A 40% lift on a page nobody visits is worth nothing. |
| 2 // what on it? | A finding is not a test. "The size guide is buried" is an observation; a hypothesis names a change, an expected effect, and the evidence behind it. |
| 3 // can we learn? | Can this page produce enough data to tell a real win from noise? Usually the answer is no — and knowing that changes what you do, not whether you do it. |
Gate three is the one almost everyone skips. It's also the one that separates a CRO program from a redesign with extra steps.
today
- walk a real store snapshot and pick the page worth working on
- argue with the app that picked a different one
- turn findings into hypotheses, and score them
- work out what you can actually prove, and what you should just ship
- leave with a backlog in an order you can defend