Every Order Traced End to End for a Global Hygiene Company

The challenge

The company's order-to-cash flow crosses trading partners, SAP, a MuleSoft integration layer and downstream transport and warehouse systems. One order passes through many message types, from the inbound purchase order to the shipment notice and invoice sent back to the partner. Integration logs already reached Elasticsearch, but nothing followed one order from its first message to its last.

Teams matched messages by hand across message types that carried different business identifiers. Failures and delays usually came to light only when a partner or business user reported them. An order that silently stopped at one stage stayed invisible until someone went looking, and volume drops or failure spikes went unnoticed until they became incidents.

What we delivered

SquareShift built the solution on the client's existing Elastic environment, directly on the integration logs already indexed there. The team made the delivery ID the single key that links every stage of an order across SAP, the integration layer and downstream systems. Each message is classified by type, so every step appears in the right place in the flow.

Five Kibana dashboards show the end-to-end journey of one order, transaction health and failures, SAP processing, integration throughput and errors, and downstream logistics and partner activity. Five machine learning jobs watch message volume, failure surges, processing time, bottlenecks and the behaviour of individual interfaces and partners.

Five alerts cover stuck orders, failure spikes, abnormal processing time, volume anomalies and critical conditions. Critical conditions open incidents in ServiceNow automatically. Every change was built and tested in a development environment first, and the team tuned alert thresholds with stakeholders to balance timeliness against noise. For order and transaction monitoring on Elastic, see our Elastic search and observability services.

The outcome

Operations teams now see where any order is and where it stalled, in one place, without chasing information across systems. Machine learning and alerts surface problems earlier, rather than after partners or business users report them. Significant issues enter the existing ServiceNow process automatically, so response starts sooner.

Monitoring of the transaction flow has moved from reactive investigation to proactive detection. SquareShift documented the work and handed it over, and the client's engineers now maintain, tune and extend the dashboards, machine learning jobs and alerts themselves. The jobs and alerts run continuously without manual effort.