Inventory not syncing between channels
“Not syncing” describes a symptom, not a cause. A channel showing an unexpected quantity can mean old evidence, a delayed or failed operation, two listings that are not really the same item, a write path that does not exist, or a manual edit. Classify the symptom first; each one needs a different response.
What “not syncing” can actually mean
- Stale observation — the channel has not been read since the stock changed. The value may be fine; the evidence is old.
- Delayed operation — the change exists but is still waiting in a queue, retry, or rate-limit backoff.
- Execution failure — expired credentials, an API error, or a timeout stopped a read or write on one channel.
- Identity problem — the two listings are not the same sellable unit, or the real pair was never linked.
- Unsupported write path — the connection is read-only or one-way, so nothing was ever going to write the correction back. Observation does not imply write capability.
- Manual divergence — someone or something edited the value directly on one channel, outside the normal path.
Sometimes nothing is broken: a buffer, cap, or allocation can make channels show different quantities on purpose. Check whether the two values are actually comparable before treating the difference as a failure.
- Expired or revoked credentials
- API error or timeout
- Rate limit or retry backoff
- Queue or retry delay
- Unconfirmed product relationship
- The read or write path does not complete normally
- The channel keeps its last known valueIt ages, but can still look like a current reading.
- Review compares uneven evidenceA mismatch, stale state, or false match becomes harder to separate.
Stale is not the same as wrong
If one channel was observed five minutes ago and another several hours ago, a difference between them does not automatically prove that either value is wrong. It proves that the evidence is not equally current.
That distinction matters because retrying a write is a different action from refreshing stale evidence. Read more on judging whether a stock number is still current.
Mismatch is not the same as failed sync
A stock mismatch is a state difference. A failed sync is an execution event. They can happen together, but one does not prove the other.
A channel can have trusted recent evidence even after a later refresh fails, and a successful refresh can still reveal a real mismatch.
From classification to a live incident
Once you know which kind of symptom you are looking at, investigate one concrete item rather than the whole catalog. Run the one-item incident checklist to capture the evidence, check identity, timestamps, operations, write capability, and policy in order, and verify the fix afterwards.
Where GNIZDO fits
GNIZDO is detection-first. It keeps each channel’s observations, their freshness, connection state, and operation results as separate facts, so you can see whether a channel is stale, whether an attempt failed, or whether current evidence really disagrees.
GNIZDO does not automatically synchronize inventory between channels.
