12 March 2026 · Funnels

Where paywalls leak between impression and receipt

Most “IAP funnels” start too late. If your first step is “checkout started,” you have already agreed to ignore three quieter exits.

In-app purchase analytics inherits a web habit: name a checkout event and call the rest a funnel. Stores do not work like hosted checkout. The person sees your paywall, may or may not see the system sheet, may or may not authenticate, and may receive a receipt hours later. Collapsing that into one conversion rate is how teams argue past each other.

Leak one: impression without sheet

The paywall rendered. The store sheet never appeared. Causes are ordinary: a client kill-switch, a missing product identifier after a catalog publish, a user on a restricted Apple ID, a Play device without a billing account. If you only log “purchase tap,” you will blame copy for an identifier mismatch. Log impression, tap, and sheet-presented as three lines. In Cohort Signal Lab we treat a tap-without-sheet rate above your historical band as an engineering page, not a growth experiment.

Leak two: sheet without attempt

The system sheet opened and closed. No payment method was attempted. This is often product: price shock, family-share confusion, or a trial that already ran on another device. It is also sometimes a Face ID cancel. Do not average them. A short post-sheet survey is less useful than a timestamp against whether the same Apple ID already held an introductory offer — a fact sitting in the receipt history if you persisted it.

Leak three: attempt without receipt

Payment was attempted; no receipt arrived in your listener. Pending states on Play, Ask to Buy, and network drops live here. In-app purchase analytics that marks these as “failed conversion” will later look foolish when the receipt lands. Hold a pending bucket with an ageing rule. Do not celebrate or mourn until the bucket clears or expires on a written clock.

Leak four: receipt after a billing holiday

Thailand’s banking calendar, like others, creates quiet days when card retries pile up. A Monday spike in paid is not necessarily a paywall win from Friday. If your window is “week of tap,” you will attribute a holiday catch-up to whatever campaign was on screen. Align the paid clock with the store’s successful charge, and keep the tap clock separate. Two clocks, two sentences.

None of this requires a new vendor. It requires the humility to admit that “checkout started” was always a web metaphor wearing a mobile jacket. If you want the worksheet we use to mark these four leaks, it ships with Cohort Signal Lab. For a shorter desk version, see Store Receipt Literacy.