Multichannel stock accuracy depends on knowing which record is authoritative for each decision. A source of truth is not simply the dashboard everyone prefers. It is a defined responsibility: which system records physical stock, which accounts for commitments, which observes supplier availability and which is allowed to change a channel’s offered quantity.
Those responsibilities may sit in one product or several. What matters is that they are explicit and coordinated. If two tools independently calculate and overwrite the same listing quantity, an accurate stock count can still produce an unreliable customer-facing result. This guide explains how to map authority, investigate disagreements and introduce a workable reconciliation routine.
Start by naming the different quantities
Write down what each number means before deciding which one is wrong. Physical units, sellable units, committed units, observed supplier availability and channel-offered quantity describe different states. A difference between them can be expected, especially while an order or update moves through the process.
For example, a warehouse may hold ten units while two are committed to orders and one is damaged. The quantity offered to new customers should reflect the business’s treatment of those states. Copying the physical count everywhere would ignore relevant information even though the count itself was correct.
Supplier-held inventory introduces another distinction. An observation from a supplier is evidence about that source at a time; it is not necessarily stock owned by or reserved for your business. Keep that evidence separate from your decision about how much to offer.
Create an authority map
| Responsibility | Question to answer | Required record |
|---|---|---|
| Physical stock | Where are receipts, losses and count adjustments recorded? | Quantity, location, reason and actor |
| Commitments | Which process recognises a unit promised to a customer? | Order reference and applicable state |
| Supplier evidence | Which source and observation time support the decision? | Exact item, value, time and uncertainty |
| Allocation | Who decides how shared availability is offered? | Rule, inputs and owner |
| Channel updates | Which system may write the resulting quantity? | Intended action and confirmed result |
| Overrides | How does an authorised manual intervention persist? | Scope, reason and expiry review |
Name the actual system and owner in each row. “The integration handles it” is not specific enough when the integration fails. The map should let an operator identify where to investigate without making another uncoordinated edit.
Give each live field a clear writer
List every system that can change quantity on a channel: inventory software, supplier tools, import routines, marketplace settings and manual operators. Establish which writer is authoritative under normal conditions and how any exceptions are coordinated.
Multiple writers are not inherently impossible, but independent writers without a shared rule can create a loop. One process lowers exposure, another restores its calculated value, and the team repeatedly sees the same discrepancy. Increasing update frequency can make that loop faster rather than more accurate.
Document the treatment of manual overrides. If an operator pauses a listing because its mapping is uncertain, the next routine update should not silently erase that decision. Confirm how the chosen tools support this requirement and test the behaviour before relying on it.
Use stable identity across the path
Every movement and observation must describe the correct item. Map the internal product and variation to the relevant channel record and supplier reference. Include attributes such as unit count and model where they change what the buyer receives.
Do not use a familiar title as the only connection between systems. Titles can change, and several variants can share similar wording. Equally, do not assume that an internal SKU has a universal meaning outside the system where it was assigned.
When a mapping changes, consider the pending work that depended on it. Open proposals, allocations and unresolved exceptions may need to be reviewed against the corrected identity. The SKU mapping guide provides a practical method for preserving that relationship.
Define how commitments affect the shared pool
If several channels draw from the same stock, a sale on one channel changes what can safely remain exposed elsewhere. Decide where that commitment is first recognised and how the revised availability reaches the other destinations. The timing and failure behaviour matter as much as the arithmetic.
Be explicit about cancellations and returns too. A cancelled order or returned parcel does not always mean a unit is immediately ready to sell. The operational state must be confirmed through the relevant process before increasing available stock.
In a simple illustrative example, eight sellable units support two channels. After three units become committed, five remain before any additional buffer or allocation policy. The channels’ displayed quantities should follow the agreed coordination model; showing five independently on both channels without considering concurrent demand can create more exposure than intended.
Choose between pooled and allocated exposure
A pooled approach aims to coordinate channels against shared availability. An allocation approach assigns portions of that availability to particular destinations. Each has operational tradeoffs, including how quickly unused capacity can move and how the business handles delayed information.
Choose the model your systems and team can maintain. Do not describe an allocation in a spreadsheet while allowing a separate tool to publish the full pool everywhere. The written policy and actual writers must agree.
Where a buffer is appropriate, document the uncertainty it addresses. It may help limit exposure during a response gap, but it does not fix double-counted commitments or wrong mappings. See the inventory buffer strategy guide for a way to record that purpose.
Make time visible when reconciling disagreements
Two values observed at different times cannot always be compared as though they describe the same moment. Record the source observation, calculation and destination confirmation times separately where the tools expose them. A single “updated” label can conceal which stage actually ran.
When investigating a discrepancy, reconstruct the sequence: the last trusted state, subsequent commitments or adjustments, the calculated target and the confirmed destination. This is more useful than repeatedly forcing the destination to match whichever dashboard was opened last.
Define an acceptable review window for your operation and what happens when it is exceeded. The window is an operating decision, not a universal guarantee. Items with uncertain supply or high consequence may need a different response from stable, low-exposure products.
Build reconciliation around causes
Group discrepancies by their likely cause: identity, missing event, stale observation, calculation, permission, failed write or competing writer. These categories help route the work to someone who can resolve it. They also stop a broad “sync issue” label from obscuring several unrelated problems.
For each discrepancy, preserve the evidence before making a correction. Record the intended state, the current confirmed state and why the chosen remedy is appropriate. If the evidence is incomplete, keep the uncertainty visible rather than choosing a number merely to close the ticket.
After correction, verify the outcome and review whether the cause can recur. A one-off manual adjustment may be necessary, but it is not a lasting fix for a process that continually reintroduces the discrepancy. Assign the underlying issue separately when needed.
Example: two systems disagree after a stock reduction
Imagine an operator reduces an eBay variation while investigating a supplier mapping. A separate integration later restores its previous target. The obvious symptom is a quantity mismatch, but the underlying issue is ownership: the integration did not recognise the hold.
The review should identify the writer, the rule it applied and the missing override control. Repeatedly reducing the listing by hand leaves the same failure available to recur. The durable remedy is a coordinated way to preserve the hold, along with confirmation that the mapping and pending proposals are reviewed before exposure resumes.
This is an illustrative operating scenario, not a reported Zelluvo customer incident. Its purpose is to show why accuracy work must include permissions and decision state, not only stock arithmetic.
Introduce a new source of truth gradually
Before migration, save the relevant mappings, rules and unresolved exceptions in a reviewable form. Identify when the old writer stops and the new one takes responsibility. Avoid an undefined overlap period in which both systems assume they are authoritative.
Start with a representative group and compare expected outcomes with confirmed results. Include an ordinary sale, a count adjustment or supplier change, an ambiguous mapping and a failed update. Test recovery as well as the successful path.
Expand only when the team can explain the sequence and handle exceptions. A rapid import proves that data moved; it does not prove that authority, commitments and destination updates now behave as intended. Keep a clear pause condition if the trial exposes uncertainty.
Where Zelluvo belongs in the map
Zelluvo publishes this guide. Its current scope centres on eBay listings, supplier mapping and local stock proposals, with approval-first live eBay stock updates. Early access permits one eBay seller account per client organisation.
That describes a particular part of an operating model, not a claim that Zelluvo is a general live multichannel inventory authority. If you need shared warehouse stock across several channels, evaluate the responsible system and its current capabilities separately. Document how any supplier-monitoring workflow will coexist with it.
For a controlled supplier-to-eBay process, discuss your authority map with Zelluvo. Bring the list of existing writers and a representative exception. That makes it possible to assess the intended role without assuming that connecting another tool will automatically improve accuracy.
Maintain the agreement as systems change
Review the authority map whenever a new channel, integration, warehouse or supplier workflow is introduced. Name a person responsible for keeping it current. An outdated ownership document can become actively misleading if operators use it to decide where to make corrections.
Track recurring discrepancies and their causes. The useful question is whether the process produces explainable, recoverable decisions. A lower count of alerts is not necessarily improvement if failures have become less visible or exceptions are being closed without confirmation.
Keep the map concise enough to use during an incident. Link to detailed procedures where necessary, but make the normal writer, override owner and recovery contact immediately visible. Operational clarity is the point of the document.
Specify what a handoff must contain
When one system passes a decision to another, document the item identity, intended value, reason and originating record. The receiving process should be able to distinguish a new instruction from a repeated delivery of the same instruction. Ask the provider how its supported integration handles retries and partial failures rather than assuming that a second delivery is harmless.
Also define the acknowledgement you need. A message accepted by an integration is not necessarily a quantity confirmed on the marketplace. During a trial, trace one ordinary change through the full path and save the evidence each stage exposes. If a stage is opaque, record that limitation and the practical method used to investigate it.
This contract is useful even when you are not developing the integration yourself. It gives support teams a common description of the expected behaviour and lets operators identify the missing step. Without it, each provider may correctly report success for its own stage while the business still has no confirmed customer-facing outcome.
Review this handoff contract whenever an integration changes.
Frequently asked questions
Must all stock data live in one application?
No. Several applications can participate if authority and coordination are explicit. The risk is contradictory ownership, not simply the number of tools. Define who controls each decision and how results are reconciled.
Which dashboard should win when quantities differ?
First establish what each number represents and when it was observed. Then follow the authority map and event sequence. Automatically choosing the newest-looking dashboard can conceal a wrong calculation or a missing commitment.
Does faster synchronisation solve stock accuracy?
It can reduce some timing gaps, but it does not repair identity errors, unclear ownership or double-counted events. Evaluate the whole process, including failure handling and destination confirmation.
What is the simplest first step?
List every system and person that can change live quantity, then assign responsibility for physical stock, commitments, supplier evidence and channel updates. Unclear or overlapping answers reveal where to investigate first.