Test payments in sandbox

Run a payment flow in the sandbox to confirm checkout, webhooks and ledger entries. Use the test card 4242 4242 4242 4242, complete a one-time $49 payment, verify the payment intent reaches succeeded state, confirm the webhook posted invoice.paid to our staging endpoint, and check our database shows a settled transaction with payment_id and user_id. Finish this by Wednesday and report any mismatch to me.

Before I hand this to QA, make the sandbox test clear and repeatable: put the success card number at the top, list the exact steps to create a payment intent, where to watch for the webhook, and three failure cases to try (declined card, insufficient funds, 3D Secure required). Flag anything that could make a reviewer hesitate, like missing test secrets or unclear DB column mappings.

I just ran the full checkout in test and everything looked green, but after swapping to live keys cards are being declined for customers who previously charged in test. Support is pushing for answers and Finance wants refunds held. I don’t know whether our payment method settings or fraud rules changed. What is the most likely diagnosis and the fastest set of checks to unblock live testing without risking real charges?

Every release we validate flows in sandbox and still hit surprises in production: different decline codes, missing webhooks, and unexpected fraud hold. I own product and payments. Which routine should I change so sandbox runs are predictive of live behavior, and what one operational control would make production surprises rare?