Integration is often the highest-value, hardest work in eCommerce. We design data flows and reliability — not just API wiring — across Adobe Commerce, Shopify, and the systems behind them.
The hardest, most valuable work in eCommerce is rarely the storefront. It's the integration layer between the eCommerce platform and the systems that actually run the business — ERP, PIM, CRM, OMS, 3PL, payment, shipping, tax, and the long tail of internal services nobody outside the ops team has heard of. When integrations are wrong, orders fail silently, inventory drifts, finance can't reconcile, and customer service spends its days chasing ghosts.
We design and build integrations across Adobe Commerce and Shopify, in both directions, with a bias toward reliability, observability, and clean ownership rather than fastest-API-wiring-wins. For broader service context see our services; for the platforms themselves see our platform notes. Integration architecture is also one of the most impactful uses of our technical audit — most integration problems are visible long before they become outages.
Designing a new integration, or fixing one that's been quietly losing orders? Start with an architecture review before more code is written.
Get an Architecture ReviewThe categories we integrate most often, with the platforms within each that we've actually shipped against.
The systems that own pricing, inventory, customers, and orders for most established businesses. Bidirectional integration with token-based auth, idempotent writes, replayable failures, and a clear model for which side owns each piece of master data.
Customer sync, lead capture, order events, and lifecycle triggers connected to marketing and sales workflows. The integration boundary kept narrow so CRM changes don't break commerce — and vice versa.
Akeneo, Salsify, Plytix, inRiver, or homegrown PIMs as the source of truth for product data. Attribute mapping, asset sync, channel-specific overrides, and clean workflows that don't require a developer for every new attribute.
Stripe, Adyen, Braintree, PayPal, regional gateways, and direct acquirer integrations. Refunds, partial refunds, captures, recurring billing, and the reconciliation tooling finance actually needs at month end.
FedEx, UPS, DHL, Canada Post, regional carriers, and 3PLs (ShipBob, ShipStation, Shipwire, custom WMS). Rates at checkout, label generation, fulfilment status, returns, and the multi-warehouse routing logic that keeps fulfilment honest.
Internal warranty systems, loyalty engines, B2B portals, dealer locators, custom OMS, and the long tail of services unique to your business. Designed and built to the same reliability standards as the off-the-shelf integrations.
The data flows that show up on almost every project — and the ones where most integration outages happen.
Stock levels kept accurate across the eCommerce platform, the ERP or WMS, and any 3PL. Webhook-driven where possible, with reconciliation passes to catch drift and clear escalation paths when systems disagree.
Attributes, descriptions, media, pricing, and category structure flowing from PIM or ERP into the storefront. Channel-specific overrides preserved, deletes handled deliberately rather than silently, and bulk updates that don't melt the platform.
Orders pushed to ERP or OMS at placement, with idempotent retries and dead-letter handling so a momentary ERP outage doesn't turn into lost orders. Status events flow back to the storefront for customer-facing tracking.
Customer creation, updates, address sync, customer groups, and credit limits flowing between commerce and ERP/CRM. B2B customer hierarchies handled properly rather than flattened on the way through.
Tracking numbers, shipment events, and delivery confirmations from the carrier or 3PL back to the storefront and the customer. Partial shipments, splits, and returns modelled correctly rather than papered over.
Post-purchase registration flows, warranty terms, serial number tracking, and downstream service workflows. Often the workflow that connects commerce to the support and field-service organisation.
Live pricing and availability driven from the ERP for B2B, contract pricing, and dynamic pricing scenarios. Cached intelligently so the storefront stays fast without lying to customers about stock or price.
Inventory of systems, master-data ownership decisions, integration topology, and reliability requirements. Output is an architecture document and risk register signed off before code is written.
Field-by-field mapping between systems, including transformations, defaults, error handling, and conflict resolution rules. The artefact your operations team can actually use to debug a stuck record at 2am.
Iterative build with each flow shipped end-to-end before the next starts. Queues, retries, idempotency, and observability built in from the first commit rather than added later.
Failure-mode testing — what happens when the ERP is down, the gateway times out, the webhook is duplicated, the payload is malformed. Each scenario explicitly handled and tested rather than left to chance.
Production cutover with monitoring active from minute zero, alerts routed to the right humans, and a runbook for the first week. We stay on standby through the first end-of-day reconciliation.
Version migrations on either side, schema changes, new flows, and capacity planning as the business grows. Smooth handoff to your team, or an ongoing support relationship.
Both, depending on the data. Inventory and order placement are usually real-time or near real-time so the storefront and ERP don't drift. Product data, large pricing updates, and historical reporting are often better as scheduled batches because they're high-volume and not customer-visible. We pick per data flow based on freshness requirements, volume, and ERP load tolerance — not as a blanket policy.
Every integration has a queue, retry policy, dead-letter handling, and alerting. We log every outbound and inbound message with correlation IDs, expose a small admin surface for replaying failed messages, and route alerts to whatever the team already uses — email, Slack, Teams, PagerDuty. Silent integration failure is the most common eCommerce outage. We design specifically against it.
Both have well-documented APIs but very different shapes. SAP integrations usually run through SAP's REST/OData services or a middleware layer (Boomi, MuleSoft, or custom). NetSuite uses SuiteTalk REST and SOAP, with token-based auth and SuiteScript on the NetSuite side for custom logic. The right design depends on which system owns the master data for products, pricing, inventory, and customers — and that decision shapes everything that follows.
Direct point-to-point integrations are simpler when there are two or three systems. Middleware (Azure Logic Apps / Service Bus, MuleSoft, Boomi, or a custom service) earns its keep when there are many systems, when transformations are complex, or when you need a queue and replay layer between systems. We don't push middleware as a default — it's a real architectural choice with cost and operational implications.
That's a real question and we answer it upfront. Options range from full handoff with documentation and runbooks, to ongoing managed support with us monitoring and responding, to a hybrid where you run day-to-day and we handle ERP version migrations and breaking changes. We document everything either way — credentials, mapping, schema, error scenarios — so the integration isn't a black box dependent on one engineer's memory.
Headless splits the integration question — back-office integration (ERP, PIM, OMS) still happens at the eCommerce platform layer, but customer-facing integrations (search, personalisation, content) often connect directly from the storefront to those services. We design the boundary so that secrets, business logic, and reliable writes stay server-side, while storefront calls stay reads-only and cacheable.
A simple, well-documented integration with one well-behaved API on each side — say, a payment gateway or a single-warehouse 3PL — is usually a few weeks of work. Full ERP integration with bidirectional product, pricing, inventory, customer, and order flows is typically 2–4 months and runs into ongoing operational costs. We estimate after a discovery and architecture review, not from a feature list.
Bring the systems list and the symptoms. The first 30 minutes are on us. We work across Adobe Commerce, Shopify, and migrations between them — see our replatforming approach if you're moving platforms and integrations together.
Book a Free Consultation