← Back to blogTry free

Stripe webhook events arrive out of order: how to handle it safely

November 10, 2026

Your logs show customer.subscription.updated being processed before customer.subscription.created for the same subscription. Or a customer who upgraded ends up on the old plan in your database. Nothing is broken on Stripe’s side: Stripe documents that it does not guarantee the order in which events are delivered.

Combined with retries (see why the same event can arrive twice), this means your handler has to be correct no matter which event shows up first, second, or twice.

How it goes wrong

The fragile version

// The version that works in testing and corrupts data in production
case 'customer.subscription.updated': {
  const sub = event.data.object as Stripe.Subscription
  await db.subscriptions.update({   // throws or no-ops if "created" has not arrived yet
    where: { id: sub.id },
    data: { status: sub.status },   // may be an OLDER state than what you already stored
  })
}

The fix: treat the event as a notification, not as the data

An event payload is a snapshot taken when the event was created. By the time you process it, the object may have changed again. So use the event only to learn which object changed, then ask Stripe for the current state and write that with an upsert:

// Re-fetch the current object and upsert it. Order stops mattering.
case 'customer.subscription.created':
case 'customer.subscription.updated':
case 'customer.subscription.deleted': {
  const eventSub = event.data.object as Stripe.Subscription

  // The payload is a snapshot from when the event was created.
  // Ask Stripe for the state as it is right now.
  const current = await stripe.subscriptions.retrieve(eventSub.id)

  await db.subscriptions.upsert({
    where: { id: current.id },
    create: { id: current.id, status: current.status },
    update: { status: current.status },
  })
  break
}

Two things make this robust. First, retrieve always returns the latest state, so a late delivery of an old event writes the same current data as a fresh one. Second, upsert means it does not matter whether the “created” event has arrived yet. The cost is one extra API call per event, which is cheap next to a corrupted subscription table.

Why not just compare timestamps?

You can store event.created and ignore events older than the last one you applied. It helps, but created has one-second resolution, so two changes in the same second tie, and it does nothing for the missing-row problem. Re-fetching plus upsert covers both. Use timestamps as an extra guard, not as the only one.

Quick check: find a recently upgraded or canceled customer and compare their plan in Stripe with your database. If they differ and nothing errored, you probably have an ordering problem.

Order bugs are silent

The nasty part is that none of this raises an error. The handler returns 200, Stripe is happy, and your data is wrong for weeks. That is the same failure mode described in your Stripe webhook returns 200, but it didn’t work. If events are also being rejected before they reach your code, start with signature verification failures.

Find out if your data already drifted

The quickest check is to compare Stripe against your database directly. Run this reconciliation script once, or let Venwai do it every day with a read-only Stripe key and alert you in Slack, Discord, Telegram, or email the moment a subscription and your app disagree.

Free while it’s in beta, no card required: venwai.com.

Is your Stripe out of sync with your app?

Venwai compares Stripe with who your app actually lets in, once a day, using a read-only key. It alerts you the moment they disagree. Free while in beta, no card required.

Check my Stripe for free