By Janson Wang · Updated October 8, 2026 · 25 min read
To investigate supplier quality drift after a restock, connect each after-sales event to the actual fulfilled quantity, then to its documented arrival source.
Keep mixed stock and missing allocation evidence visible as UNKNOWN. A confirmed arrival link tells you which goods to investigate; it does not prove a factory defect.
Quick Answer
Work backward from the affected order line. Identify its actual shipment, reconcile the quantities allocated from each arrival, and attach the after-sales event to the supported allocation.
Preserve unknown quantities and candidate arrivals where records break. Compare only equivalent shipped-unit groups with a stated observation window.
Use the result to request specific inspection evidence for suspected supplier quality drift.
Keep these distinctions in the record
Arrival versus production lot. A receiving event describes what entered your operation; a factory lot needs its own source.
Order versus fulfillment. Trace the quantity actually dispatched, including split shipments and stock from several arrivals.
Event versus affected unit. A complaint, refund and replacement can concern the same original goods.
Confirmed versus unknown. A date-based shortlist helps request records but cannot establish allocation.
Association versus cause. Traceability narrows the investigation. Inspection evidence still has to explain the reported difference.
If a complaint raises a potential safety risk, pause affected shipments and isolate potentially affected stock while a qualified lead assesses the next action. Do not wait for a complete trace.
Table of Contents
Start with the reported unit
Name arrivals without inventing lots
Map actual fulfillment quantities
Keep mixed stock and gaps visible
Link after-sales events once
Compare only matched exposure
Test suspected supplier quality drift
Close the trace or keep a breakpoint
Questions about the trace
Final thoughts
What you need to decide Record to inspect Useful output
Which goods reached this order? Fulfillment line and actual stock allocation A quantity-linked arrival, or an explicit gap
Which after-sales units belong there? Event line, item identity and shipment reference A supported event link, or an unresolved quantity
What should be checked next? Reported difference, agreed reference and inspection evidence A targeted request with an owner
Can arrivals be compared? Eligible shipped units, linked events and observation dates Comparable counts, or a stated limit
Scope: A proposed manual tracing workflow. It is not an inspection verdict, an automated platform feature or a supplier-performance dataset.
In a May 2026 Shopify Community post, MiloDeMilo described a return cluster on one SKU and scattered receiving, fulfillment and returns records.
“It just lived in three different places and nobody was looking across all of them at once.”
— MiloDeMilo, Shopify Community , 11 May 2026.
That is the author’s account, not an ASG customer case or a measure of how often this happens.
Inspection firm Insight Quality describes quality drift as deterioration after initially acceptable production. Keep that diagnosis provisional while you trace the reported goods.
1. Start with the reported unit
Open a real after-sales event before naming a suspect arrival.
Find the order line, variant, quantity and actual shipment behind the report. Keep the customer’s stated problem separate from an inspection finding.
Preserve the records as they existed when you checked them. If the shipment or physical item cannot be identified, record that break before attempting to assign an arrival.
I would begin with the event that prompted the investigation, then expand the scope only as the records justify it.
A support ticket may cover an entire order. Your trace needs the item and quantity discussed within it. Record whether the customer reported a fault, requested a refund or physically returned goods.
Keep the original event reference. Save the ticket, return or transaction ID and the source export. A later summary should point back to it.
Check the exact product. Use the ordered SKU and variant, then compare any identifier visible on the reported item. Keep a discrepancy in the notes.
Find the fulfilled line. An order line can have several actual shipments. Identify the shipment concerned before following its stock source.
Separate event time from record time. Keep when the event happened and when you obtained the evidence. Add a timezone when the source supplies one.
Name the missing fact. Ask who can retrieve the dispatch record, item identifier or quantity detail. Leave a responsible person and next check time.
A tracking number may locate a parcel. It does not establish which physical units were packed unless a linked packing or allocation record supports that connection.
Packaging is an illustration; it does not identify an inbound receipt or prove a customer case.
The existing QC evidence guide distinguishes order identity from unit identity. Use that distinction when a parcel contains several identical items or substitution is possible.
If the customer cannot identify the shipment, keep the report at its known level. Do not create a precise fulfillment link merely because one shipment date looks plausible.
Preserve the first export before correcting labels or merging tickets. In the working copy, note which source changed your interpretation. Keep the same record IDs across saved versions.
Start with one identifiable event line. Carry an unknown shipment forward as a missing link, not as a guessed batch assignment.
2. Name arrivals without inventing lots
Give each receiving event a stable reference that leads back to its source record.
Record the arrival’s SKU, variant, supplier, received quantity, quantity unit and location. Store a supplier production lot separately, and only when a label or document supports it.
Several arrivals can share a production lot; an arrival can also contain several lots. Neither relationship follows automatically from a receiving date.
Use the receiving system’s event ID where available. If none exists, assign a local reference and mark it as locally created in the source notes.
That local reference helps you retrieve a record. It must not be presented as a manufacturer’s lot code.
A GS1 draft discussion of event traceability from April 2022 describes connecting receiving and shipping records through object identity and consistent source/destination details. This is a general tracing principle, not a current implementation standard or evidence that ASG uses the described system.
Field group Where to obtain it Boundary
Arrival ID and received time Receiving record or a labelled local index linked to that record An ID you create is a filing reference, not proof of a production run.
SKU, variant and quantity unit Item master, order specification and counted receipt Keep packs, pieces and variants separate until a documented conversion supports comparison.
Received quantity and location Counted receipt and the relevant stock-location record Received, ordered and currently available quantities answer different questions.
Supplier and production lot Supplier documents and item or carton labels Store the lot as UNKNOWN when it is absent or its meaning cannot be established.
Reference version and evidence Approved specification or other confirmed reference, plus original files A current listing image does not automatically establish what was agreed for an earlier arrival.
Basis: Identity and record-integrity requirements in the linked QC evidence guide. The separate arrival ID and production-lot fields are part of this proposed tracing design.
A warehouse scene, not proof of a receipt-to-unit allocation.
Retain both supplier and receiving references when they differ. Ask the sender what its code identifies: a production run, a shipment, a purchase reference or something else.
Do not merge records simply because the cartons share a delivery date. Keep distinct receiving events where the source records distinguish them.
Check the unit of measure before doing any arithmetic. A carton quantity and a piece quantity cannot be added without a confirmed pack conversion.
If a product is sold as a bundle, identify whether the quantity refers to bundles or components. Leave the component-to-bundle relationship unresolved until the packing definition supports it.
Customer returns and internal transfers should keep their own receiving history. Label the inbound kind so those movements are not mistaken for fresh supplier restocks.
In Shopify’s transfer-report definitions , received units include both accepted and rejected stock. Transfer-shipped quantity describes movement out of the origin location. Keep these transfer measures separate from customer dispatch and from an acceptance decision.
The aim is a retrievable history, not a new code pasted over missing evidence.
Keep the receiving event and supplier lot in separate fields. Request the meaning of an unfamiliar code before using it to group goods.
3. Map actual fulfillment quantities
The useful link includes a quantity and the record that supports it.
Create a fulfillment record for each actual shipment line. Then create allocation records for the portions supplied by each arrival.
Keep an unresolved remainder where the source cannot be established. Across the same product and quantity unit, confirmed allocations plus unresolved shipped quantity must reconcile to the actual fulfilled quantity.
Use actual stock movements, picking records, packing records or verifiable item identifiers. A planned allocation is a lead to check against what physically left.
Record both the original order line and the actual fulfillment line. If the source lacks a suitable shipment-line ID, create a local one linked to the dispatch evidence.
When an order ships in parts, each actual shipment gets its own fulfillment record. Preserve the common order-line reference so the parts can be checked together.
Shopify’s split-fulfillment guidance allows an order’s fulfillment to be separated by item and quantity. Use those lines to locate shipment records, then verify actual dispatch; a platform status or label alone does not establish which stock left.
When a shipped line uses several arrivals, create a separate allocation for each supported portion. Keep the quantity on the allocation, not just a list of arrival names.
The right question is how many shipped units the record connects to each arrival.
For a fulfillment line with a known shipped quantity, use this reconciliation:
Actual fulfilled quantity = confirmed allocated quantities + unresolved shipped quantity.
An unresolved remainder may be calculated by subtraction when the fulfilled total and confirmed portions are known. Label it as a reconciled difference. The calculation establishes a quantity gap, not its source.
If those allocations exceed the fulfilled total, stop the calculation. Check duplicate exports, unit conversions and planned versus actual movements before adding more links.
Also reconcile allocations drawing on the same arrival against their supporting actual outbound movements. Separate fulfillment balances must not reuse the same movement quantity twice.
Proposed manual record links. Every confirmed portion still needs supporting evidence.
1. Use the blank tracing CSV
Download the blank trace template (CSV) . It contains five starter row types and no business data. Copy the row type you need, then fill it from your records.
This is a manual working sheet. Its fields are proposed here; they are not a claim that Shopify provides these batch fields or fills them automatically.
Row type One row represents What quantity means
INBOUND A recorded receipt of a defined SKU and variant Units received at that event
FULFILLMENT One actual shipment line Units actually dispatched on that line
ALLOCATION One supported or unresolved portion of a fulfillment line Units attributed to that arrival, or the unresolved remainder
AFTER_SALE One service-event line on its stated goods and quantity basis; its link rule depends on event type Units reported, refunded, returned or cancelled, or replacement units sent; follow the event rules below
EVENT_LINK One portion of an after-sales event linked to an allocation, or left unresolved Units covered by that link
Template design: Filter by row type before totaling quantity. These rows describe different stages and must never be added into a single total.
Fill the common identity, product, quantity-unit, quantity-status and source fields first. Then complete only the fields needed for that row type; these are manual worksheet instructions.
INBOUND: Check its arrival ID, inbound kind, receiving time, supplier and location. Enter only documented production-lot references.
FULFILLMENT: Check its ID, order and order-line references, shipment reference and actual dispatch evidence. Add delivered_at when comparing delivered exposure.
ALLOCATION: Check its ID, parent fulfillment, supported arrival or UNKNOWN, quantity, link_status and evidence. Add candidate IDs and completeness only when needed.
AFTER_SALE: Check its event ID, event_type, occurred_at and event-specific fulfillment reference. Apply the rules for replacement shipments and cancellations before dispatch .
EVENT_LINK: Check its after-sales reference, allocation or UNKNOWN, quantity, link status and supporting evidence. Apply both source-confirmation checks and the per-allocation cap.
For multiple documented lots in one receipt, retain their individual source references in scope_note; do not infer unsupported lot-level quantities. quantity_status describes the quantity’s basis, not acceptance or rejection.
Fill owner, next_action and next_check_at when an evidence request is open. Use scope_note for comparison definitions and inapplicability; keep customer reasons, inspection findings and reference versions separate.
Give every filled row a unique record_id. For an INBOUND row, repeat it in inbound_event_id; for FULFILLMENT, in fulfillment_line_id.
Apply the same rule to allocation_id and after_sale_event_id. Those fields then hold the referenced record’s ID on linked rows.
Keep outside system IDs in the appropriate order fields or source_record_ref. An event ID in this sheet identifies an event line, even when the source ticket covers several items.
Use quantity_status to distinguish RECORDED, RECONCILED_DIFFERENCE and UNKNOWN. Leave the numeric quantity blank when it is unknown. A blank is not a measured zero.
Use UNKNOWN for an unavailable identifier. Leave fields that do not apply to that row type empty. Remove unused starter rows before analysis.
Reconcile each fulfillment line before grouping it by arrival. A balanced quantity with an unknown source remains an unresolved allocation.
But isn’t this too much tracking?
It can be. When an order uses unmixed stock and existing records already reliably link the affected quantities to their source, a full 36-column worksheet can duplicate work.
Use those records and fill only the rows and fields needed to answer the current question. Preserve their references and quantity basis, without copying a reliable link just to complete every column.
Add allocation bridges and UNKNOWN handling when split shipments, mixed stock or an unclear after-sales source leave a gap. Choose the extra work according to the evidence missing from this event.
Proposed blank worksheet record types; no customer or performance data.
4. Keep mixed stock and gaps visible
An allocation can be known even when the returned unit’s share of it is not.
Separate the stock-to-shipment link from the event-to-stock link. Packing records might establish how much of each arrival entered a parcel.
Without item-level evidence, they may not identify which arrival supplied the unit later returned. Keep that event’s source unresolved, even when the shipment allocation itself reconciles completely.
Record situation What you can retain What stays unresolved
Several arrivals supplied one shipment Separate supported allocation rows and their quantities The returned unit’s source if its identity cannot distinguish those allocations
One order line shipped in parts Separate fulfillment lines linked to the original order line An event’s shipment if the report does not identify the relevant parcel
Old and new stock were mixed Any reliable unit, lot or movement evidence that survived A single arrival assignment when only the SKU and dates remain
Only part of the allocation is documented Confirmed portions plus one unresolved remainder The source of the remaining shipped units
The visible history may be incomplete Known candidate arrivals and the missing date or stock scope Whether the candidate set contains every possible source
Decision aid: These are record conditions, not customer examples. Confirmation depends on the actual evidence for each link.
Keep confirmed links separate from candidate arrivals and missing evidence.
A response in the cited community thread suggests looking at time clusters. That can help you choose which receiving records to request first.
It cannot establish which units were picked. A stock-rotation policy, including FIFO, is also different from a record showing what happened on this shipment.
Shopify’s inventory-adjustment guidance describes a history of the person, time and reason behind quantity changes. Use it to investigate a discrepancy. That history alone cannot identify the arrival that supplied a particular returned unit.
Use CONFIRMED only when the linked record supports the stated quantity. Use CANDIDATE when you have a reasoned shortlist. Use UNLINKED when you cannot support a useful source assignment.
For a candidate allocation, put UNKNOWN in the arrival ID field. List the possible arrival IDs separately in candidate_inbound_event_ids, separated by a vertical bar.
Keep the unresolved quantity on one row. Do not repeat that whole quantity under every candidate; that would multiply the apparent exposure.
Set candidate_scope_complete to YES only when the relevant stock history supports a complete set. Use NO for a known gap and UNKNOWN when completeness cannot be assessed.
If the list is incomplete, an arrival missing from it has not been cleared. Ask for the opening stock, earlier receipt or transfer record needed to define the full range.
In scope_note, state why each candidate remains possible and what could distinguish it. Keep the original shortlist when new evidence narrows it.
Keep one unresolved quantity and its candidate set. Resolve a link only when new evidence supports that quantity’s source.
Where does your SKU’s trace stop?
Bring the receiving reference and the fulfillment record you can retrieve. Ask ASG which supporting records and checks can be included for your product and order.
Discuss one SKU’s traceability →
Confirm the service scope and available evidence before arranging work.
5. Link after-sales events once
Keep the service history complete without counting the same goods repeatedly.
Create an AFTER_SALE row for each genuine service-event line. Where shipment links apply, use EVENT_LINK rows for supported portions and unresolved quantities.
Preserve unresolved portions separately. Different events can concern the same units, so an event count is not a count of distinct affected goods.
Establish that overlap before calculating a unit-based measure.
First remove duplicate copies of the same source event. A ticket imported into two exports is still one event, even if the exports use different row numbers.
Then retain genuinely different actions. A complaint followed by a refund and a physical return describes a sequence. Deleting the later actions would lose useful evidence.
Shopify’s returns and exchanges guidance separates creating a return from receiving and processing the goods. Exchange items need their own fulfillment. Preserve those distinct events when completing the worksheet.
Event type Quantity to record Counting boundary
Complaint Units actually covered by the reported issue A report is not an inspected defect.
Refund Documented units associated with the refund, if known A money amount alone does not establish a physical quantity or receipt of returned goods.
Return requested Units requested for return Requested units are not yet received returns.
Return received Units actually received back and identified Keep condition and inspection findings separate from the customer’s reason.
Cancellation before dispatch Units cancelled before dispatch Keep outside shipped-unit exposure and physical-return counts.
Replacement sent Replacement units documented in the service action Trace their actual fulfillment separately; do not assign the original fault to their stock source.
Carrier return Units documented as returned by the delivery process Keep separate from customer-initiated returns and inspect the actual reason.
Basis: The existing return-measurement guide separates event types and unresolved causes. These worksheet labels are proposed categories, not claims about Shopify dashboard definitions.
Handling scene only; not an identified return, refund or replacement event.
Shopify’s refund guidance allows a refund without a physical return. It also distinguishes a staff refund note from a return reason. Preserve the actual event and reason record; do not turn refund notes into proof that goods were received back.
The return-measurement and cause guide explains why the number and the event behind it need separate checks. Here, apply that boundary to each linked quantity.
Shopify’s return-line model references the fulfilled line and distinguishes quantities processed, eligible for refund and already refunded. That reference can help locate the original shipment; it does not supply an inbound-receipt or lot assignment.
Here, RETURN_RECEIVED denotes Return received and REPLACEMENT_SENT denotes Replacement sent. These event labels and link rules are conventions of this manual template, not a platform schema.
For complaints, item-linked refunds and returns concerning original dispatched goods, fulfillment_line_id identifies the original shipment line. If that applicable link cannot be established, keep the order references and enter UNKNOWN.
Separate known shipment portions into event lines sharing the original case reference. Do not divide an uncertain quantity between possible shipments just to fill the map.
For an EVENT_LINK row, reference the AFTER_SALE record ID and the supported ALLOCATION record ID. Its quantity is the part of that event covered by the link.
Check that the allocation belongs to that event’s fulfillment line. The SKU, variant and quantity unit must also match before the link is confirmed.
For each event and allocation, the combined confirmed EVENT_LINK quantity must not exceed that allocation’s quantity. Portions within one event must not overlap.
Confirmed portions plus the single unresolved remainder must reconcile to that event’s recorded quantity. If the event quantity is unknown, leave the balance uncalculated; an unknown allocation quantity also leaves its cap unverified.
Confirm an arrival only when both the event-to-allocation link and the allocation-to-arrival link support that quantity. A confirmed link to an allocation whose arrival is UNKNOWN does not confirm the arrival.
If a returned unit could belong to several allocations in a known parcel, keep one EVENT_LINK with allocation_id = UNKNOWN and the known fulfillment_line_id.
Record candidate arrivals and their scope without copying the event quantity under every candidate. Keep the parcel’s supported allocations unchanged.
Use affected_unit_group_id when evidence establishes that events concern the same physical goods. Preserve that evidence reference; a shared order number is not enough by itself.
For multi-unit events, a shared group label does not prove which subsets overlap. If overlap cannot be established, keep the event totals separate and mark the distinct affected-unit count UNKNOWN.
Apply quantity caps within the event being traced. Across different events, count distinct goods only with supporting unit or subset evidence; do not cap the sum of complaints, refunds and return receipts as if each action represented different goods.
Keep reported_reason in the customer’s terms. Put a checked observation in inspection_finding, linked to its record. An empty inspection field does not mean the complaint was disproved.
For REPLACEMENT_SENT, record the replacement product, quantity unit and quantity. Link fulfillment_line_id to its own replacement FULFILLMENT, with its own ALLOCATION records.
In scope_note, use the note key related_original_after_sale_event_id for the related original AFTER_SALE ID, or UNKNOWN, alongside the source case reference. This uses the existing note field and adds no CSV column.
Any EVENT_LINK for the replacement follows its own allocations. It does not attribute the original complaint to the replacement stock.
For a cancellation before dispatch, retain the cancelled order-line quantity and references. Leave fulfillment_line_id and allocation references empty, explain their inapplicability in scope_note, and create no EVENT_LINK for those unshipped units.
Exclude that cancellation from dispatched-unit and physical-return measures. UNKNOWN denotes an applicable but unproved link; it does not replace an inapplicable field.
Reconcile event-link quantities, then check overlap between events. Do not add every service action together and call the result defective units.
6. Compare only matched exposure
Receiving goods and exposing those goods to customer use are different events.
Define the shipped-unit group and follow-up period before comparing arrivals. The denominator must contain the same kind of units that could enter the numerator.
Report missing allocation, unknown delivery dates and unresolved returns alongside the calculation. If the groups have not had comparable observation, present counts and limits instead of ranking their return rates.
Do not divide returns by the quantity purchased or received. Unsold stock has not had the same chance to produce a customer return.
Shopify’s sales-report definitions separate physical returns from reversed quantities, which also cover refunds, cancellations and edits. Its returned-quantity rate uses ordered quantity as the denominator. The matched, mature delivered-unit measure below uses a different denominator and window; label it separately.
For this delivered-unit comparison, fix one data cutoff C and one follow-up length W in scope_note, alongside SKU, variant, quantity unit and included shipments. Use the same C, W and time basis for both groups.
Choose that period for the actual return terms and question you are studying. Apply the same choice to both groups; do not select different windows after seeing the results.
A documented delivery at time d is mature only when d + W ≤ C. Recent deliveries whose follow-up period has not ended remain immature observations.
Count a physical return only when its RETURN_RECEIVED event occurred between d and d + W and its record was available at C. Do not add later returns to the older group’s numerator while stopping the newer group at W.
For this event, occurred_at is when the returned goods were actually received; recorded_at preserves when the evidence was obtained. A return-request measure needs its own name and endpoint, so do not substitute request or export time for physical receipt.
Keep late, immature and date-unknown records separately identified. Leave records with unestablished necessary dates outside the timed share, and report their scope.
Missing delivery dates leave maturity UNKNOWN. Keep those units visible rather than treating the absence of a return as a completed observation.
Measure Include Keep separate
Received quantity Counted units entering a receiving event Unsold, transferred or held stock; this is not the return-rate denominator
Eligible exposure Actual delivered units of the chosen product and unit basis, with documented d and d + W no later than C Recent deliveries, missing dates and excluded shipment types
Arrival-linked denominator Eligible quantities with confirmed allocations to that arrival Allocation-unknown quantities, shown alongside
Physical-return numerator Distinct units from that same mature delivered group, received back within W after documented delivery and recorded by C, with both links to the arrival confirmed Requests, refunds without physical receipt, duplicate actions and unresolved links
Unknowns The number of unresolved units where it can be calculated, plus the missing records No assumed zero and no proportional allocation to candidate arrivals
Proposed delivered-unit comparison. The existing return guide supports separate event types and clocks; the same C, W and mature delivery cohort govern both sides of this worksheet’s ratio.
Visible stock is not a denominator of matured delivered units.
For a fully documented subset, the observed physical-return share is:
Distinct units from the eligible group physically received back within W after documented delivery, recorded by C and confirmed through both links to the arrival ÷ mature delivered units in that group confirmed to the same arrival.
State whether the result covers the full chosen group or only its linked subset. Where the denominator is zero or unknown, do not calculate a percentage.
If a returned unit could belong to several arrivals, leave it outside an arrival-specific numerator and show it in the unresolved count. Do not distribute it by purchase share.
A known-subset calculation can still omit unresolved returns. That limit must sit beside the result; it prevents the subset from being presented as the arrival’s full return rate.
Keep variants, packaging versions, destinations and routes visible when comparing groups. Differences can guide the next check; they do not establish the cause.
Document how replacements and previously returned stock are treated. A fresh shipment of the same physical unit should not quietly become another unique manufactured unit.
Name the numerator, denominator and observation window. If missing links could change the comparison, keep the conclusion provisional.
7. Test suspected supplier quality drift
Use the trace to ask a narrower inspection question.
A confirmed arrival association tells you where to request evidence. It does not establish when a defect arose or who caused it.
Compare the reported difference against the reference actually agreed for those goods. If that baseline is missing, describe the observation and request confirmation before declaring nonconformance or material substitution.
The wrong question is whether a matched arrival has proved supplier fault. Ask which observation, reference and inspection record could support or challenge that conclusion.
Send the receiving or inspection contact a bounded request: the SKU, arrival reference, linked fulfillment quantities, reported symptom and evidence already available.
Keep the customer’s original reason alongside your request. A description such as a visible finish difference should not silently become a claim about the material inside.
Ask which specification version, approved image, packaging requirement or agreed sample record applied to that arrival. If a physical reference exists, confirm its identity and condition.
This does not assume anyone offers a sealed or retained-sample service. Confirm the actual reference arrangement for the order.
Photos can support what is visible in the identified item and recorded view. They do not establish hidden composition, long-term performance or the condition of every unit in an arrival.
A warehouse illustration, not evidence of inspection coverage or supplier defect.
Where the reported issue needs functional or material testing, ask for the relevant method, scope and qualified interpretation. Do not turn an untested observation into a pass or fail label.
Use the guide to inspection stages to place a follow-up check at the appropriate stage. For the record itself, the Supplier Verification & QC Proof Guide provides a broader evidence checklist.
Existing QC, reinspection and release records handle the check and its disposition. Attach their references to this trace rather than copying those forms into it.
If there is potential safety or regulatory risk, stop affected shipments, isolate potentially affected stock and escalate to a qualified safety or compliance lead.
Keep uncertain scope visible. A commercial release signature cannot settle safety obligations or clear stock whose risk has not been assessed.
Request evidence for a defined question. Keep arrival association, inspection finding and responsibility as separate conclusions.
8. Close the trace or keep a breakpoint
At the next check, update the missing link rather than restart the whole investigation.
Reopen the same event and compare the new evidence with the saved record. Confirm identities, quantities and source references before changing a link.
Close completed evidence requests individually. If a fact cannot be recovered, record the precise tracing limit, the remaining owner and the next decision.
A finished trace does not itself close the quality or customer case.
Question still open Person to identify Evidence to request
Which receipt supplied these units? The operator holding actual allocation or dispatch records A quantity-linked movement, pick or packing record
Which parcel or unit was returned? The support or returns contact handling the event The event line, identifiers and receiving-condition record
What reference applied? The person who confirmed the order requirements The applicable version and confirmation record
What did the check establish? The person responsible for the check The method, scope, result and linked inspection evidence
What can move next? The authorised handling or release owner The relevant disposition record; safety escalation where applicable
Proposed responsibility prompts. Confirm the actual people and service scope; the table does not assign ASG or a supplier responsibilities by default.
Check new records against the original IDs. A fresh export with different row numbers should not create new events or replace the history of what was unknown.
Save the prior worksheet version before changing a working row. Record the new source and reason for the change in scope_note. Analyze only the current version of each record.
Resolve each filled, non-UNKNOWN worksheet reference to a record of the correct type. Keep external system references distinct from these local row references.
When new evidence identifies part of an unresolved quantity, update the confirmed portion and reduce or remove the old unresolved portion in the same working update. Do not retain both as additional goods.
Recalculate the affected fulfillment and event balances, allocation caps and shared-arrival outbound totals. Recheck related rows if a corrected SKU, variant or unit of measure changes the grouping.
Use owner, next_action and next_check_at to keep the open request concrete. Choose the next check from the actual response commitment and business deadline; this guide supplies no fixed interval.
Record the tracing status on the AFTER_SALE row. Define its scope in scope_note, so a completed request is not confused with a complete trace.
Use OPEN while evidence is being pursued, TRACE_COMPLETE when the defined links are supported, or TRACE_LIMIT_RECORDED when a documented gap cannot presently be resolved.
TRACE_LIMIT_RECORDED means the limit is explicit. It does not mean the stock, supplier or customer case has been cleared.
If no allocation record exists, say so. Keep the candidate scope, unresolved quantity and reason the record could not be recovered. Identify who must decide the next handling or customer action.
Finish with a short summary: confirmed arrival links; candidate or unlinked quantities; evidence requested or received; observation limits; and the owner of each remaining decision.
That summary is useful even when the trace ends at UNKNOWN. It tells the next person what is known and what document would change the answer.
Close the data task at its supported scope. Keep unresolved quality, safety and customer decisions with named owners.
A sorting scene is not evidence that a particular stock trace is complete.
9. Questions about the trace
Can I trace an arrival from only a SKU and dates?
You can build a shortlist of records to request. A confirmed link still needs actual allocation evidence or reliable item identification. Keep the arrival assignment UNKNOWN if the evidence cannot distinguish the candidates.
Is an arrival event the same as a supplier production lot?
No. An arrival records receipt into your operation. A supplier production lot describes a production grouping supported by its own label or document. Keep the identifiers separate, including when several arrivals share a lot.
What if a returned unit came from a mixed parcel?
Retain the parcel’s supported allocations. If the returned unit cannot be distinguished among them, keep its EVENT_LINK unresolved. Do not divide that unit among candidate arrivals to make the rows look complete.
Can a refund count as a physical return?
Only a record of goods actually received back supports a physical-return count. Keep the refund event separately, including its documented unit quantity when available. A refund amount alone cannot supply that quantity.
How long should tracing a return take?
There is no verified turnaround time in the evidence used here. Timing depends on which records exist and who can retrieve them. Assign the next check from actual commitments, and preserve any missing link between checks.
Does a confirmed arrival link prove a supplier defect?
No; it supports the source assignment within the documented scope. The reported issue still needs an appropriate reference and inspection evidence. Keep picking, packaging, transport, description and other explanations open where the records warrant them.
Final thoughts
For suspected supplier quality drift, start with the records behind the affected goods. Connect the event, actual shipment and documented arrival quantity before drawing a cause conclusion.
Take one open event and complete the trace as far as the evidence permits. Leave a clear breakpoint wherever it stops.
The useful handover names the linked quantities, the unknowns and the next evidence request. Keep the source documents behind every confirmed link.
Turn the missing link into a clear request
Share a redacted trace summary with ASG: the product, confirmed links, unresolved quantities and records you need. Ask for the proposed scope before arranging any checks or handling.
Discuss the evidence you need →
Available records and service arrangements need confirmation for the specific order.
External Sources
ASG Data Note
The worksheet and quantity rules are a proposed manual method. They contain no ASG order data, customer results or sample performance figures.
ASG’s linked guides supply background on record identity, inspection and return-measurement boundaries. They do not verify the records or capabilities available for a particular order.
This article makes no assumption that ASG or Shopify maintains the complete mapping described here. Confirm the actual records, responsibilities and service scope with the relevant parties.