Method
Lifecycle Analytics for Mobile Users, as we actually teach it.
This page is the spine of every programme. If a vendor pitch contradicts it, we keep the spine. The programme list is how you practise the same sequence with faculty in the room.
1. Name the human, not the install
An install is a store event. A user is a person who may share a device, reset an ID, or refuse tracking after a prompt. Lifecycle Analytics for Mobile Users starts by writing who is allowed into a cohort and who is parked in a holding set. Holding sets are not failures; they are honesty.
2. Inventory events with three shelves
Trusted, tolerated, forbidden. Trusted events may define a window. Tolerated events may annotate. Forbidden events — autoplay completes, background fetches, pre-permission pings — may not decide that someone “activated”. Teams skip this shelf exercise because it is unglamorous. Their later charts then argue with their privacy notices.
3. Draw windows on paper
Activation, first habit, lapse, and possible return. Each window has a clock, an allowed event, and an owner on the product team. If two windows share the same event, one of them is decoration. Paper comes before SQL so that a warehouse join cannot launder a weak definition.
4. Write the limitation you will not fix this quarter
GB teams in particular should name consent gaps, kids’ accounts, and store-level attribution delays. A map without a limitation gets rejected in class. The same standard belongs in a board pack, even when nobody asked for it.
5. Defend the map out loud
Someone who did not draw it must be able to restate the windows. If they cannot, the schema is still private knowledge. That is the failure mode we designed Cohort Cartography to interrupt.