Support Funnel App Analytics, as Identity Nodecore teaches it, starts with a suspicion: product analytics loves events that fire when a UI is touched. A person opening an article, reaching 80% scroll, or dismissing a coachmark looks like progress. On a support desk, those same events often sit one tap before “contact us”. Drop-off is not the gap between step A and step B on a vendor funnel chart. Drop-off is the moment the person’s intent leaves your named path.
In The Funnel Signal Method we print the help tree and mark every node that cannot deep-link into a status screen, a cancel flow, or a proof upload. Those nodes are exits even if the analytics export calls them completions. The vanity move is to report “72% finished the article”. The reading-room move is to ask how many of those finishers opened a ticket on the same intent within twenty-four hours.
Joining those two worlds is ugly. Ticket systems and product events rarely share a stable user key. When they do not, we still write the question on the Atlas: “unknown join”. Faculty would rather grade an honest unknown than a blended rate that hides the join. Several Bangkok desks we sat with had beautiful article funnels and no way to see repeat contacts by article ID. Their drop-off work began with a tagging discipline on the desk, not a new dashboard tile.
Zero-result search is another exit that vanity charts omit. If help search returns nothing, the person did not “drop off” from an article they never saw. They left a stage you may not have drawn. Log it separately. If the vendor cannot export it, the box stays empty and the leadership brief says so on page one.
None of this makes a prettier weekly slide. It does make the argument smaller and sharper: this article, this intent, this missing deep link. That is the unit of Support Funnel App Analytics we will keep grading.