Symptoms
How this shows up in operations
If you are reading this page, you have probably noticed some of the following symptoms in your operation:
- Inventory levels between Shopify and the ERP diverge weekly or daily
- Refunds appear in Shopify but never make it to the ERP for revenue recognition
- Multi-currency orders show wrong amounts after FX conversion
- Order webhooks fail silently and accumulate in dead-letter queues nobody monitors
- Bundles and kits explode inventory math and break order routing
- Multi-warehouse routing makes the wrong call and ships from the wrong location
None of these alone is conclusive — every operation has bad weeks. The diagnostic question is whether the symptoms are recurring, growing, and resistant to one-off fixes. If yes, you are likely looking at one of the root causes below rather than a tactical problem.
Root causes
Root causes
Four root causes account for the majority of cases we see. They are not mutually exclusive; most operators have two or three running at once.
No defined source of truth. The most common root cause is no explicit decision about which system owns inventory and order data. Both systems try to write; conflicts accumulate. The fix is documenting the source of truth at kickoff, not after divergence.
Missing error-queue design. Integrations fail at the edges (refunds, partial fulfillments, gift cards). Without a dead-letter queue and daily review process, failed flows pile up silently until finance close reveals the gap.
Catalog drift. SKU master data in Shopify drifts from the ERP over time as merchandising and ops teams make independent changes. Without governance, the integration progressively breaks.
iPaaS treated as a black box. Celigo flows configured by partners and never owned in-house become invisible. When something breaks, nobody knows how to debug.
Identifying the root cause is the leverage point. Symptoms can be patched indefinitely without making progress; root causes, once addressed, fix multiple symptoms at once.
Solutions
How specialists fix this
Vetted specialists in the network typically pursue these approaches, in roughly this order:
1. Lock the source of truth in writing. Document which system owns inventory, orders, customers, and refunds. Inventory almost always belongs to the ERP or WMS, not Shopify, once you cross two locations or 200 orders a day. Make the decision explicit; review it quarterly.
2. Build a dead-letter queue with alerts. Every integration flow needs a failure path. Failed orders, refunds, and inventory updates land in a queue with daily Slack alerts. Operators review and resolve within 24 hours.
This single change prevents the slow-drift category of failures.
3. Establish SKU master governance. Designate a SKU master owner (usually merchandising or ops, not engineering). Document the SKU lifecycle: how new SKUs are created, how variants are configured, how retired SKUs are handled. Review the SKU master quarterly for drift.
4. Take ownership of iPaaS flows. Even when a partner implements Celigo or Boomi flows, your team should be able to read them. Spend a half-day during implementation walking through every flow with the partner. Document each one in a simple internal wiki.
The order matters because the first two solutions often unlock the rest. Skipping them in favor of tactical patches is the most common path to repeated problems.
Sequencing
Sequencing the fix
Operators often try to fix these problems in the wrong order. The instinct is to start with whichever symptom hurts most this week, which produces tactical patches that do not stick.
A more durable sequence: stabilize the highest-impact symptom enough to buy thinking time, then attack the most upstream root cause (usually a missing source of truth, a missing process, or a missing owner), then layer the remaining solutions on top of the now-stable foundation.
Skipping the stabilization step leaves the team firefighting; skipping the root-cause step guarantees the problem returns in a different shape within a quarter.
A vetted specialist's first deliverable is usually this sequencing plan rather than any specific fix — because the sequence is where most operators lose months of progress.
Measurement
What to measure once you have fixed this
Once the root causes are addressed, set up the measurements that will catch the same problem if it returns.
The right metrics differ by situation but tend to share three properties: they are leading indicators rather than lagging ones, they are visible weekly rather than monthly, and they have explicit thresholds that trigger investigation.
For most operations problems the leading indicators are workflow-level (cycle time, accuracy, exception rate) rather than financial — by the time finance sees the issue, the operational damage has already been done.
The brands that stay out of this cycle for years are the ones that built the right measurements once and treated the weekly review as non-negotiable.
When to hire
When to bring in outside help
Hire a specialist when symptoms recur (not one-off bugs), when finance loses trust in the data, when integration changes are needed and your partner is unresponsive, or when you are evaluating an ERP change and want a stack assessment first.
The scoping call is free. We route requests to one or two vetted specialists whose case studies match the situation.
Within one business day, you have introductions and an opinionated recommendation about whether the situation needs a project engagement or a smaller-scope assessment first.