workflows

Make webhook accepted does not mean processed: check the queue

Distinguish HTTP acceptance, queue entry, scenario execution and the final effect before resending a Make webhook.

Your system sent a webhook to Make and got HTTP 200, but the order has not changed. Before resending, separate the endpoint’s receipt from the scenario run and the effect in the destination.

Short answer

Under Make’s documented default response, HTTP 200 Accepted means the request was accepted into the webhook queue; it does not prove the scenario finished. Check the queue, run history and matching business ID in the destination. 400 Queue is full means the excess request was rejected. A custom Webhook response module can change the meaning of the response, so inspect that scenario’s logic.

Four pieces of evidence

Evidence What it supports Next check
Default HTTP 200 Accepted Request entered Make’s queue Is it pending or already a run?
Item in the queue Unprocessed request is still held Is a schedule or blocked run delaying it?
Run found The scenario started or attempted the item Which step finished, failed or remained incomplete?
Destination ID or receipt Evidence of a business effect Is the state correct and action already applied?

This table is a Pulse editorial synthesis. Record only the event ID, time, HTTP response and business ID needed for diagnosis. Keep secrets and customer data out of an open diagnostic sheet. A started run is not proof of a completed effect.

Why the queue may not drain

Make’s webhook documentation says scheduled webhooks accumulate requests until the scenario schedule runs. The number processed per cycle depends on Maximum number of results. If arrivals exceed processing over repeated cycles, the queue grows. Check the schedule, that setting and scenario status before changing frequency; account for plan capacity and usage.

Instant webhooks run in parallel by default. With Process data in order, the next run waits for the previous one; scenario settings say an incomplete execution can hold the sequence. Investigate the blocking run before resending queued requests.

Make documents different default responses: 200 Accepted for a queued request, 400 Queue is full for a full queue and 429 Too many requests for a rate limit. These describe platform receipt, not an order’s final state. Check any custom response module and where it runs before interpreting its status.

Decision tree before replay

  1. Did the sender receive a response? Keep the code and attempt ID. If the outcome is unknown, mark it uncertain.
  2. Was it the default Accepted? Look for the event in the queue and runs. If pending, wait or resolve the documented block; do not resend merely because the destination has not changed.
  3. Was the queue full? Check whether another attempt was accepted and review the sender’s retry policy. Recover by ID when capacity is available; do not replay an entire batch blindly.
  4. Did a run start? Match the business ID in the destination. Do not repeat a confirmed effect. Investigate an external call with an uncertain result before replaying.

Hypothetical example: A received Accepted and remains queued; B received Queue is full; C has a run and a destination receipt. Monitor A, recover B through a controlled path and do not resend C. This illustrates decisions, not Pulse measurements.

For repeat deliveries, use the duplicate-webhook guide. For an uncertain effect, reconcile missing records. For a reverted final state, inspect out-of-order runs.

Sources and limits

The queue and codes above come from Make’s current webhook documentation. Do not transfer these response meanings to another provider without its contract. The tree does not guarantee automatic recovery and has not been tested by Pulse in production.

Sources