Gift card API production access is a cutover, not a config flip, and it deserves a written runbook. The nine steps below take an integration from a green sandbox test matrix to a monitored production system that has already bought, delivered and reconciled one real code before any customer arrives.
The sandbox proved the code works. The runbook proves the environment does, and the two fail independently.
The runbook
-
Production credentials, stored properly. Production keys are issued separately from sandbox keys, usually after commercial and KYB approval. Store them in a secrets manager, never in code or config files under version control, and never reuse a sandbox secret anywhere in the production path. Note the rotation procedure now, while nobody is under pressure.
-
Base URL and environment flags. Grep the entire codebase for the sandbox hostname before switching. Hardcoded sandbox URLs hide in webhook handlers, cron jobs, health checks and test fixtures, and one survivor means part of your system quietly talks to fake stock. Environment selection should live in exactly one configuration point.
-
Webhooks re-registered for production. Register your production endpoint URLs with the provider, with new signing secrets distinct from the sandbox ones. Then verify with a real test event, not by assuming: fire whatever test delivery the provider offers and confirm your handler receives it, verifies the signature, and processes it once.
-
Fund the account, arm the alerts. Deposit a minimum viable balance, enough for early trading with a buffer, rather than a large float on day one. Set low-balance alerts before the first order, because an unfunded account turns every checkout into a failed order with a confused customer attached.
-
Re-sync the production catalogue. Sandbox catalogues are test data. Product IDs, denominations, regions and prices in production may all differ, so run a full catalogue sync against production and rebuild your mappings from it. Shipping sandbox product mappings is a classic first-day failure, and keeping the catalogue fresh afterwards is its own discipline: see catalogue sync and price drift.
-
One small real order, end to end. Buy a low-denomination code with real money. Follow it through delivery, storage and customer notification, redeem it yourself if the brand’s terms allow, and find the charge on the provider’s statement. This single order validates credentials, funding, webhooks, encryption and settlement in one pass.
-
Monitoring live before customers. Dashboards and alerts must exist before traffic does: order failure rate, provider error responses, account balance, webhook delivery lag, and catalogue sync age. An integration that fails silently on launch morning fails expensively.
-
Reconciliation from day one. The daily match of your order log against the provider’s statement starts with the first order, not in week two. Day one is when mistakes are most likely and cheapest to catch, and a reconciliation habit established early costs minutes; established late, it starts with archaeology.
-
A tested pause switch. Know exactly how to stop sales cleanly: a flag that takes products off sale, holds new orders and shows customers an honest message, without dropping orders already in flight. Test it before launch. The day you need it is not the day to design it.
What changes after the cutover
Going live is a beginning, not a finish line. The first weeks surface everything the sandbox could not: real latency, real stock movements, real settlement timing, and the occasional provider incident. Keep the sandbox environment alive and maintained, because every future change to the integration should rehearse there first, and keep the runbook updated as the system evolves. Teams that treat it as a living document get their second provider integration live in a fraction of the time; the broader sequencing of that work is covered in the integration guide.
Frequently asked questions
How do I get production access to a gift card API?
Production access is granted by the provider, normally after a commercial agreement, KYB checks and sometimes a review of your sandbox integration. You receive separate production credentials, webhook signing secrets and a production base URL, and you fund the account before ordering. Expect the approval to take days rather than minutes, so request it before your launch date depends on it.
What should I check before going live with an API?
Work through a runbook: production secrets stored outside code, no hardcoded sandbox URLs anywhere, webhooks re-registered and verified with a test event, the account funded with alerts armed, the production catalogue synced fresh, and monitoring plus reconciliation running. Then place one small real order end to end and confirm it on the provider’s statement before opening to customers.
Why did my integration work in sandbox but fail in production?
The usual causes are environmental, not logical. Sandbox credentials or hostnames survived somewhere in the code, webhook secrets were not re-registered so signatures fail, product IDs differ between the sandbox and production catalogues, or the production account is unfunded. Re-run the go-live checks in order; the first failing step almost always names the culprit.
Should I place a real order before launching a gift card store?
Yes, exactly one, at the lowest denomination available. Buy it with real money, follow it through delivery, storage and notification, redeem it yourself if the brand’s terms allow, and find the charge on the provider’s statement. That single order exercises credentials, funding, webhooks and settlement together, in production, before any customer depends on them. It is the cheapest end-to-end test the integration will ever get.
How much should I fund a gift card API account at go-live?
Start with a minimum viable balance: enough to cover expected early trading plus a buffer, not a large float on day one. The important part is not the amount but the alerting, since an unfunded account turns every checkout into a failed order. Arm low-balance alerts before the first order, then let real sales data size the working float over the following weeks.