Skip to content
AppDues
Log in
Contact salesSign up

App Store outage on January 24, 2020: what happened and how to respond faster

On January 24, 2020, Apple's verifyReceipt endpoint failed for roughly five hours. In-app purchases could be completed, but developers could not validate them, so paid content remained locked. This page reconstructs the timeline, explains why detection and mitigation lagged, and turns the lessons into a practical readiness checklist for teams that depend on a third-party receipt endpoint.

What users experienced

The core problem was a validation gap, not a checkout failure. A user could finish a purchase inside an app, yet the developer's server could not confirm that purchase with Apple. Until validation succeeded, the app had no reliable basis to unlock the paid feature. The incident therefore looked successful at the point of payment and broken at the point of entitlement.

Why the response lagged

Two weaknesses compounded each other. First, monitoring data existed but did not page anyone: alerts were tuned mainly for latency spikes, while this incident produced elevated 5xx errors. Second, there was no ready-made fallback. Retries helped in normal operation — one server-side retry plus SDK retries — but they could not fix a sustained upstream failure. A mitigation had to be written during the incident, and it only helped users of one entitlement approach. The practical lesson is to treat detection and fallback as separate systems: an alert that fires on error-code anomalies, and a pre-built degraded mode that does not need to be invented under pressure.

Readiness checklist for receipt-dependent apps

Use this checklist before you need it: 1) Alert on anomalous upstream error codes and on semantically invalid success responses, not only on latency. 2) Page a named on-call person, with escalation if there is no answer within minutes. 3) Pre-build a fallback validation path, such as local receipt validation when the upstream returns a 500. 4) Keep a replay queue so receipts that failed during an outage can be re-sent after recovery. 5) Decide in advance who may grant temporary access, how it expires, and how it is reconciled later. 6) Publish a status update early, even if the update is only that the scope is still being confirmed.

Recovery is more than the endpoint coming back

Recovery was staged, not instant. The endpoint began working again at 10:55 AM PST, but restoring users took longer: failed receipts were re-sent from server logs starting at 12:57 PM PST, new purchases from the outage window were replayed by 2:27 PM PST, and all receipts including renewals were re-imported by 8:47 PM PST. For your own incident plan, separate three milestones — upstream recovery, entitlement restoration, and full reconciliation — and report each one separately so users are not told the incident is over while their access is still pending.

Incident timeline, January 24, 2020
Time (PST)Event
4:11 AMverifyReceipt begins failing and timing out intermittently; about 45% of requests fail with 50x errors.
6:20 AMFailure mode changes to HTTP 200 with an unusable response: a webpage redirecting to iTunes.
7:36 AMIssue discovered independently via social media and internal monitoring.
8:05 AMScope recognised: endpoint has been fully down since 6:22 AM.
8:12 AMPublic alert posted and full monitoring begins.
8:39 AMWork starts on temporary entitlements for anyone posting a receipt.
10:20 AMServer-side temporary-entitlement mitigation deployed.
10:55 AMverifyReceipt begins functioning again.
12:57 PMRe-sending receipts Apple could not validate during the outage.
2:27 PMReplay of new purchases from the outage window completed.
8:47 PMAll receipts, including renewals, re-imported; issue closed on status page.

Times and events as reported in the original incident account; purchase-impact figure is that account's estimate.

Frequently asked questions

Why did users pay but not get access?

Purchases could be completed in the app, but paid content stayed locked because the purchase could not be independently validated. That mismatch generated support tickets, negative reviews, and lost developer time.

What made detection harder than a normal outage?

The failure changed shape. It began as intermittent 500 errors, then became HTTP 200 responses containing an unusable webpage redirecting to iTunes. Monitoring only for error codes or only for latency would miss one of those phases.

How was access restored after the endpoint recovered?

A temporary entitlement unlock was built and deployed at 10:20 AM PST for users relying on that system. It did not cover everyone, so receipts from the outage were later replayed and re-imported, finishing at 8:47 PM PST.

What should teams fix before the next upstream outage?

Alert on anomalous error codes, not just slow responses, and prepare a fallback validation path before an incident. The planned fallback was local receipt validation when Apple returned a 500, giving fuller tracking than temporary entitlements.