Inventory mismatch between platforms
Two platforms showing different numbers is not automatically a mismatch. A difference only means something when both observations describe the same sellable unit, are current enough to compare, and were supposed to show the same quantity.
This guide covers when two channel observations are eligible to compare, and what a difference can and cannot tell you.
The short version
Two channels showing different numbers is not automatically a problem. First make sure you are comparing the same item, that both numbers are recent, and that the channels were supposed to show the same quantity. A difference that survives all three checks is worth chasing; one that doesn’t is usually old data, a wrong pairing, or a buffer doing its job.
Identity: are these the same sellable unit?
Comparison belongs at the level that is actually sold: a variant, an offering, a specific listing. Do not compare two listings simply because their SKU, title, or price looks similar. Duplicate SKUs, bundles, multipacks, and variant families make look-alikes common.
If shared identity is unknown, the correct result is “not compared,” not “mismatch.” How cross-channel product identity is confirmed covers which evidence counts.
Freshness: are both observations current enough?
A quantity observed recently and a quantity observed hours ago are not equivalent evidence. If the latest read of one channel failed, its last good value is older evidence, not a current reading.
When one side has not been observed since the stock changed, the difference is bounded by stale evidence. Refresh that channel before interpreting it. How inventory data freshness works explains trusted, stale, and unverified evidence.
Policy: should the values be equal at all?
Many sellers expose different quantities on purpose. A buffer holds back units on a marketplace, a cap limits what one channel shows, reserved stock is excluded, an allocation splits physical stock between channels, or two channels sell from separate stock pools.
Compare the observed difference with the policy that was supposed to produce it. A difference that matches the policy is expected, not an error. How stock intent works explains how that relationship is declared for a confirmed item.
Observed state: interpret the values last
Only when identity is confirmed, both observations are current, and the channels are meant to agree does the reported quantity, price, or listing state become a meaningful comparison.
Comparison eligibility, step by step
-
Are both listings confirmed as the same sellable unit?
Matching SKU or title text marks a candidate, not a confirmed pair.
Yes — continue to 2.
If no or unsureNot comparable. Identity is unresolved, so there is no mismatch to interpret yet.
-
Is each observation current enough to compare?
Compare observation times, and check that the latest read of each channel completed.
Yes — continue to 3.
If noInsufficient current evidence. The older value is past evidence, not a current reading. Refresh it first.
-
Should both channels expose the same quantity?
A buffer, cap, reservation, allocation, or separate stock pool can make different values intended.
Yes — continue to 4.
If noCompare against the policy. A difference that matches the policy is expected, not a mismatch.
-
Do the current values still disagree?
Compare the quantity, price, or listing state each channel reports.
Yes — continue to 5.
If noNo mismatch. The observations agree on the evidence available.
-
Current, comparable, unexplained difference. This is a real state mismatch. It still does not say which value matches physical stock or why it happened.
What the outcome can mean
Even a real mismatch does not identify which value matches physical stock or why the difference appeared. To trace the cause on one item, run the one-item incident checklist.
Where GNIZDO fits
GNIZDO compares channel observations only for items with confirmed shared identity: confirmed by the seller, or automatically only from an exact valid GTIN or an exact SKU that is unique on each participating channel with no GTIN conflict. Ambiguous matches stay unconfirmed.
Each observation keeps its own channel and timestamp, so stale evidence is shown as stale rather than as a mismatch. GNIZDO shows the evidence of a difference; it does not decide which physical stock count is correct or which channel value should win.