A Shopify storefront scales to peak traffic without much help. The integration behind it often does not. Inventory syncs fall behind, the same unit sells twice on two channels, orders queue up waiting for the ERP, and the warehouse works from yesterday’s picture.
A Shopify ERP integration has to be designed for the busiest hour of the year, not the average one. This article covers where these integrations break under load and the design choices that keep inventory, orders and fulfillment in step.
Where peak exposes the integration
| Symptom at peak | Usual cause |
|---|---|
| Oversells and cancellations | Inventory synced on a schedule, so channels sell stock that is already gone |
| Orders reach the ERP late | Orders pulled in batches, or pushed one at a time into an ERP that cannot keep up |
| Missing or duplicated orders | Webhooks not handled idempotently; retries during slowdowns |
| Integration stalls | API rate limits hit by per-item calls that were fine at normal volume |
| Fulfillment status lags | Status updates queued behind order traffic |
Most of these are invisible at normal volumes, which is why they are discovered on the day that matters most.
Treat inventory as a ledger, not a number
The most important design decision is where available-to-sell inventory lives and how changes reach every channel.
Syncing a stock number from the ERP every fifteen minutes works until demand outpaces the interval. At peak, a unit sold on one channel can still appear available on another until the next sync.
A more robust model treats inventory as a stream of events – sales, returns, receipts, adjustments, reservations – applied to a single authoritative ledger that every channel reads from. A sale anywhere decrements availability everywhere at once. Safety stock buffers protect against the remaining latency for fast-selling items.
Orders: accept fast, process steadily
Shopify will accept orders far faster than most ERPs can create them. The integration should absorb that difference:
- Receive order events immediately and put them on a durable queue.
- Create orders in the ERP at a rate the ERP can sustain, with retries and backoff.
- Make order creation idempotent, keyed on the Shopify order ID, so retries never create duplicates.
- Monitor queue depth and age, and alert before the backlog affects fulfillment.
A queue that grows during a flash sale and drains over the following hour is a healthy system. An integration that fails when the ERP slows down is not.
Respect the API limits
Shopify’s Admin GraphQL API uses cost-based rate limiting, and webhooks arrive with the retries and duplicates that come with at-least-once delivery. At peak:
- Batch inventory updates rather than sending one call per item.
- Use bulk operations for large catalog and order volumes.
- Track the cost budget and back off before limits are hit, not after.
- Process webhooks idempotently and reconcile against the API afterwards.
Rate limits and API capabilities vary by Shopify plan and API version. Check the client’s plan before designing around them.
Test for the peak you expect
Load testing the integration is as important as load testing the storefront:
- Replay a recorded day of real order traffic at multiples of normal volume.
- Slow the ERP down deliberately and confirm the queue absorbs the difference.
- Inject duplicate and out-of-order webhooks and check nothing is processed twice.
- Reconcile orders and inventory across systems after the test.
Run the test early enough to fix what it finds.
From our case studies: one inventory source of truth
This example is from eProxim’s published case studies, anonymized to the industry.
An omnichannel retailer sold the same catalog across its own website, several marketplaces and physical stores. Each channel kept its own view of stock, reconciled to the ERP by syncs every fifteen minutes, through a tangle of point-to-point connections. At peak, units sold on one marketplace still appeared available on another, producing oversells, cancellations and marketplace penalties.
eProxim replaced the point-to-point connections with a middleware layer that ingests stock, order and fulfillment events from every channel, the ERP and the warehouses, and maintains a single real-time inventory ledger. The order lifecycle runs as events, with retries and reconciliation so no message is lost under load.
The result: 99.9% inventory accuracy across every channel, stock changes propagating in under a second instead of every fifteen minutes, and a peak sale at four times normal volume absorbed without a rebuild. Adding a marketplace or warehouse became configuration rather than a new integration. Read the full e-commerce case study.
Related reading
- Idempotent webhooks: duplicates, retries and out-of-order events
- NetSuite Salesforce integration: keeping CRM and ERP in agreement
- Integration reconciliation reports: proving two systems agree
Working with eProxim
eProxim does not run a Shopify practice. We work alongside Shopify agencies and partners, and build the layer between Shopify and the back office: ERP integration for inventory, pricing, orders and fulfillment status, multi-store catalog synchronization, and the event-driven architecture that holds up at peak.
Bring us the Shopify back-office work you would rather not hire for. See our Shopify partner page, or start a partner conversation.