Every shipping integrations roadmap eventually asks the same question: do we stay on EDI or go REST? The honest answer is "yes."
What to keep on EDIFACT, what to move to REST, and how to run both in parallel without breaking partners.
Anything where the partner graph is >200 counterparties and the messages are mature (IFTMIN, IFTMBF, CODECO, COPRAR). The cost of migrating each partner is much higher than the marginal gain.
Also: anywhere the regulatory requirement is EDI-specific. Some customs authorities haven't moved, and the compliance risk of fighting that is nonzero.
New partners. Internal systems. High-frequency, low-schema-stability flows (visibility, IoT telemetry, event streams). Anywhere you need bidirectional real-time.
One of the underappreciated wins of REST is the developer-onboarding cost — a new partner can be live in hours instead of weeks, which matters more than the technical merits most of the time.
Strangler pattern. Build the canonical domain model once. Adapt EDI inbound to the canonical model at ingress. Adapt canonical to EDI outbound at egress. REST speaks the canonical model directly.
This gives you two wins. Partners don't see any disruption. And when you eventually deprecate an EDI flow, it's an adapter change, not a domain change.
One email a fortnight. Field notes + case studies + early access to webinars.