Paywall drops without vanity metrics
This note was written after a House Table with an independent language app. They had shipped a more beautiful paywall, watched trial starts rise, and very nearly celebrated in a channel called #wins. The second invoice told a colder story. We were given permission to write the method, not the brand.
A paywall is a door. Counting how many people touch the handle is not the same as counting who lives in the house. Trial starts are a handle count. They are useful the way a queue is useful: they tell you the door is visible. They do not tell you the rooms are worth the rent.
Hold the people still
The only honest paywall read is a cohort of the same people. Those who saw wall A, those who saw wall B, followed until the second successful charge, after refunds, after family-share quirks, after the store’s cut if you are reporting what the company actually keeps. If your analytics event fires on “purchase initiated” and not on “purchase succeeded,” you are reading a hope.
Teams skip this because the second charge takes a month. Product culture is impatient. App Analytics that cannot wait a month is not analytics; it is interior design. Ship the wall if you must, but write the Signal Note as a provisional claim with a dated review. Put the date in the channel, not in a private doc.
What “drop” even means
People say “paywall drop” as if the wall were a feature flag with a single number attached. In practice you are changing copy, price, trial length, and often the moment the wall appears. If you change three things, you do not have an experiment. You have a mood. Experiment Hygiene exists in our catalogue because we grew tired of calling moods science.
If you can only change one thing, change the moment, not the gradient. A wall that appears after a moment of competence converts differently from a wall that appears after a moment of confusion. Both can raise trial starts. Only one tends to raise the second charge.
Store fees belong in the sentence
A 30% store fee is not a footnote. If your “successful” wall recruits annual plans, the cash arrives differently from monthly plans, and the refund window is a different animal. A Signal Note that reports gross trial starts and never names net proceeds is a costume. We say so even when it makes the slide less pretty.
Revenue Leak Mapping is the longer desk for this work. The flagship only gives it a week. That is a limitation, not a teaser. If monetisation is the whole job, book the longer desk or hire someone whose entire week looks like week six.