Stepwise Outbound from Kingdee to Wangdiantong Self-Circulating Returns: A Sync Solution for Outbound Warehouse Adjustments

What This Strategy Solves

Pushing outbound warehouse adjustment documents from Kingdee Cosmic to Wangdiantong sounds trivial, but once a business scales, problems pile up: dual-maintained encoding mappings, mismatched inventory definitions, missing batch and expiry fields, confusing adjustment directions — within three months the two systems stop reconciling and the business team starts chasing IT for answers.

On one real engagement, the client used Kingdee Cosmic for finance and supply chain master records, and Wangdiantong for store and outbound warehouse operations. The "stepwise outbound" pattern means Kingdee pushes adjustment lines to Wangdiantong progressively; "self-circulating returns" means that once the outbound warehouse work is done in Wangdiantong, the feedback data flows back into Kingdee to close the loop. The strategy's core value is handling both directions in a single integration so the entire outbound adjustment process stays reconciled.

Data Flow and Field Mapping

The overall direction is: Kingdee Cosmic (source) → Qeasy Data Integration Platform (middle layer) → Wangdiantong (target). After the outbound action, Wangdiantong sends the actual results back through a return channel into Qeasy, then into Kingdee.

The key fields, refined repeatedly in field work, are roughly as follows:

For encoding mapping, a robust approach is to use Qeasy's "Centralized Encoding Mapping Management", keeping materials, warehouses, and customers in one mapping table that all downstream strategies reference — instead of duplicating mappings in each strategy. A phased header-and-body landing is also common: land the header first into Wangdiantong to capture the internal document number, then push the body line by line so the external warehouse code travels with each row.

How to Configure on Qeasy

When we build this strategy on a client site, the key configuration points are:

1. Source access: pick Kingdee Cosmic as the source. Use its open API to pull documents incrementally by document number, with a filter limited to the outbound warehouse adjustment business type — never pull other transfer orders by mistake.
2. Target write: on the Wangdiantong side, use its document write API. First create the outbound adjustment header, then iterate over line items.
3. Return channel: spin up a separate strategy to handle "self-circulating returns". Its trigger is a status change indicating the outbound warehouse work is complete, and it writes actual quantities, batches, and the receiver back into Kingdee.
4. Field converters: in Qeasy's converters, normalize units, date formats, and enumerations. For example, Kingdee quantity fields often carry a unit conversion coefficient, while Wangdiantong only accepts base units — this must be standardized.
5. Exception handling: failed line items go into a retry queue. After exceeding the retry threshold, they suspend and trigger manual intervention, so dirty data never lands in the destination.

Implementation Steps

Phased rollout is the key to stability. We typically follow three steps:

• Incremental starting point: on day one, initialize a checkpoint such as "outbound warehouse adjustment documents open in the past 7 days" and run a full trigger once to set the watermark. Then poll the source every 15 minutes, pulling only documents with a modification time later than the watermark.
• Full trigger: freeze historical data first; do not run full and incremental in parallel — the watermark will get confused. Switch to incremental only after the full run completes.
• Schedule frequency: outbound runs every 15 minutes; return runs event-driven (triggered by status change). Do not poll on a schedule for the return direction, otherwise the system will repeatedly write back before warehouse work is done.

The "incremental and full dual-track" pattern is common among Qeasy customers: incremental handles daily traffic, full runs on reconciliation days to verify numbers on both sides.

Pitfalls and Lessons Learned

• External warehouse codes not unified: Kingdee's external warehouse organization code and Wangdiantong's warehouse code are two separate systems. The typical mistake is to pour Kingdee's ID straight into Wangdiantong, causing warehouse misidentification. Mapping is mandatory — do not cut corners.
• Batch field direction inconsistency: Kingdee places the batch field on the header, Wangdiantong on the body. A common error is to sync only the header, leaving downstream batch information empty. The safe approach is to expand batches by line item.
• Return direction pushed backwards: if the self-circulating return is misconfigured as a sync direction, Wangdiantong's actual results will be treated as new outbound instructions and pushed back into Wangdiantong, creating an infinite loop. Clearly label source and target on the return strategy.
• Inventory definition conflicts: Kingdee's instant inventory and Wangdiantong's available-to-promise inventory use different algorithms. Do not subtract numbers directly during cross-system reconciliation — align by business dimension first.
• Missing state machine: outbound warehouse adjustment has states like "pending shipment, in transit, signed, returned". If you only trigger the return at the final state, intermediate state queries will miss data. Recommend writing a state log on every status change for traceability and clearer reconciliation.

When It Fits and When It Does Not

Fits: retail or manufacturing enterprises with external warehouses, consignment warehouses, or third-party fulfillment who need bidirectional sync of inventory and documents between ERP and store operations systems, with daily volumes from hundreds to tens of thousands. Does not fit: pure internal scenarios where both ends are financial systems or where the document is only generated on one side; nor businesses requiring second-level real-time inventory — this solution runs at minute-level latency by design.