Result

After a brief series of tight development cycles, a connector solution was released. The system accessed the e-commerce platform's data store, using order and inventory APIs to enable critical data collection.

To ensure proper synchronization, duplicate ERP and e-commerce data were centrally stored, serving as a bridge between the two systems. Specialized interfaces exposed calculations, making the underlying numbers and their rationales transparent.

Additionally, the new system was designed to continue processing transactions with automatic retries, even when connections began to fail.

Hosted on a virtual machine in a remote data center, the solution was designed to handle onboarding with minimal time required. The platform's capabilities were extensive:

• Control over fulfillment from the online shop while managing the fulfillment process in the back-office ERP
• Ability to map inventory counts across multiple warehouses, ensuring quantities match exactly
• Fine-grained control over which sales order to use as the starting point for synchronizing data
• Side-by-side analysis tool enabling global visibility of item matching
• Online payment processing and refund capture with transactions reflected in accounting
• Balancing to the penny with streamlined reports that spot even the slightest difference
• Selective control over which backend inventory items to surface in the online storefront

Impact

The completed product was a fully developed solution capable of addressing a range of integration needs. Backend processes were managed using open-source Hangfire for queuing and asynchronous execution.

To maximize robustness, the integration solution included an imprint of records before upload, with sufficient source trace data to project the future footprint in the destination system.

By running both the backend and frontend processes on the same virtual machine, the communication loop was tightened.

The final deployment of the solution established a new set of norms for ERP integration. Carefully straddling two systems clearly justifies a standalone system with split duties and split resources.

However, in practice, duties are rarely split evenly. Moreover, with an in-between platform, the underlying modules required special attention to ensure safe synchronization.

In the future, implementing suitable automation that runs at a configurable frequency to ensure reliability will become virtually unavoidable.