Journal · Thailand
How Thai fintech apps misread drop-off
Bangkok product teams know PromptPay. Their funnels often do not. A user who leaves your app to confirm a transfer in another bank app looks like drop-off. Many of those users come back and finish. Some finish without returning, because the receiving merchant already got paid.
If your session timeout is fifteen minutes and the hop takes four, you may still stitch the story. If timeout is thirty seconds, you will invent a crisis every evening around electricity bill due dates.
KYC is not a single step
National ID OCR fails in glare, with worn cards, and when a parent’s card is used for a child’s account. Users retry. A funnel that treats any exit from the OCR screen as abandonment will page the wrong squad. Watch the session: some retries succeed on the third attempt. That is recoverable friction, not drop-off.
Some retries fail because the product burned OTP attempts. That is a quality failure with a different owner. Lumping both into “KYC conversion” hides the distinction the Lab exists to teach.
7-Eleven and cash rails
Top-up and bill-pay flows that end in a barcode at a convenience store complete in the physical world. Your app session ends at “show barcode.” If you mark that as incomplete checkout, you will never celebrate the journey that actually worked for users without a bank app.
Session quality evaluation asks whether the job progressed, including offline completion you can only infer from a later “balance updated” event. Write the inference as a limitation. Do not pretend you saw the cashier scan.
What we ask Thai teams to change first
Extend session windows around known hop destinations. Add a “returned from payment app” event if the OS allows. Sample replays on the day bills are due, not only on quiet Tuesdays. Sit CX and analytics in the same replay for one hour; they already know which failures are real.
None of this requires a new vendor. It requires admitting that a Thai payments graph is not a web funnel with nicer colours.