Essay · 18 January 2026
SKAN ladders are an operations problem
A conversion-value scheme is a living contract between product, paid media, and the person who updates the MMP.
Teams treat the SKAdNetwork ladder as a clever puzzle they solve once. Then product ships a new paywall, the revenue bands shift, and paid social optimises toward an outdated bit. The postback still looks official. The meaning has drifted.
We ask labs to name an owner. Not “growth,” a person. That person keeps a dated map: which in-app event writes which bits, which lock window you chose, and which revenue band is now wrong. When product changes the paywall, the map gets a new date before the store review lands.
Coarse values cannot encode a full funnel. Choose. First-order margin bands usually beat vanity events. Tutorial completion is a product health metric; putting it in SKAN teaches the ad network to buy curious people, not paying ones.
Android is not a holiday from this discipline. OEM graphs and Play referrers have their own decay. If your iOS scheme is documented and your Android scheme lives in a Slack thread, the board pack will quietly prefer the prettier column. Resist that.