All field notes
ERP Integration

Celigo, Boomi or Custom: Choosing Integration Middleware

The honest version of a decision that is usually made on a vendor demo and regretted about fourteen months later.

360 Expert Solutions3 min read

iPaaS platforms sell speed to first integration. Custom middleware sells control. The mistake is evaluating them on how fast the first flow ships, when the cost that actually bites is what happens at the fortieth change request.

Integration middleware is chosen badly more often than almost any other component of a commerce stack, and for an understandable reason: the evaluation measures the wrong thing. Everyone demos the first flow. Almost nobody models the fortieth change request.

Below is how we actually advise on this, including the cases where we recommend clients do not hire us to build something custom.

What you are really buying

An iPaaS is not primarily software. It is a maintained library of connectors, a hosted runtime, and — most valuably — somebody else’s responsibility when NetSuite changes an API. Custom middleware is the opposite trade: complete control of behaviour, complete ownership of consequences.

Comparison of Celigo, Boomi and custom middleware across seven dimensions. Time to first integration: Celigo days, Boomi one to two weeks, custom three to six weeks. Prebuilt NetSuite and e-commerce connectors: Celigo strong and purpose-built, Boomi broad but more general, custom none. Handling unusual business logic: Celigo constrained by the template, Boomi flexible through a visual builder, custom unconstrained. Cost shape: Celigo and Boomi are recurring subscription growing with volume and endpoints; custom is higher upfront then hosting only. Who fixes a breaking API change: vendor for both iPaaS options, you for custom. Observability: good out of the box for both iPaaS, whatever you build for custom. Team skills needed: integration configuration for iPaaS, software engineering for custom.
DimensionCeligoBoomiCustom
Time to first flowDays — templated NetSuite/commerce flows1–2 weeks — visual build3–6 weeks
Prebuilt connectorsPurpose-built for NetSuite + commerceVery broad, more general-purposeNone — you write them
Unusual business logicConstrained by template shapeFlexible within the visual builderUnconstrained
Cost shapeSubscription, grows with volume + endpointsSubscription, grows with connectionsHigh upfront, then hosting
Who fixes a breaking API changeVendorVendorYou
ObservabilityGood out of the boxGood out of the boxWhatever you build
Skills requiredIntegration configurationIntegration configurationSoftware engineering + on-call
There is no winning column. There is a column that matches your logic complexity and your appetite for ownership.

The cost curve nobody models

Subscription pricing looks expensive against a fixed build cost, and teams often reason that custom "pays for itself in year two". Sometimes it does. But the comparison usually omits the running cost of custom: on-call, dependency upgrades, the engineer who understands it leaving, and the week absorbed each time a platform ships a breaking change.

Conversely, iPaaS costs scale with exactly the thing you are trying to grow. Pricing tied to endpoints, flows or transaction volume means a successful year produces an invoice increase, and by year three the "cheap" option can be the expensive one.

When we recommend iPaaS

  • Your flows are recognisable: orders down, fulfilments and tracking up, items and inventory across, customers synced. This is precisely what the templates exist for.
  • You have no engineering team, or the one you have should be working on the storefront rather than on plumbing.
  • You need it working this quarter. A templated flow genuinely can be live in days, and that is not marketing.
  • Your ERP is NetSuite and your platform is mainstream. Connector maturity is highest exactly here — Celigo in particular is built around this pairing.

When we recommend custom

  • Your business logic is the product. Complex allocation, bespoke pricing resolution, multi-warehouse rules that do not resemble anyone else’s — fighting a template here costs more than writing the thing.
  • You are at a volume where transaction-based pricing has become the dominant line item.
  • You need latency the platform cannot offer. Most iPaaS flows are scheduled or near-real-time; if you genuinely need sub-second, you need your own path.
  • You already run engineering with on-call and deployment discipline. If that culture does not exist, custom middleware becomes an unmaintained liability within a year.

Questions to ask any vendor

  1. Show me a failed flow

    Not the happy path. What does the error look like, who is alerted, and how is it replayed after a fix?

  2. How are changes promoted?

    Is there a real dev-to-production path with version control, or does someone edit the live flow?

  3. What happens at ten times this volume?

    Ask for the pricing, not the reassurance.

  4. What is not covered by the connector?

    Every connector has gaps. A vendor who admits them readily is describing a product they know well.

Key takeaways

  • Evaluate the fortieth change request, not the first flow — that is where the cost actually lands.
  • iPaaS buys you someone else’s responsibility for API changes; custom buys you unconstrained logic. Both are real.
  • Model three years including maintenance headcount; assume 15–20% of build cost annually for custom.
  • Standard flows plus NetSuite plus a mainstream platform is the strongest case for iPaaS.
  • A deliberate hybrid is fine. An accidental one is the worst of both.
CeligoBoomiiPaaSMiddlewareNetSuiteIntegration

360 Expert Solutions

E-commerce & ERP Integration Team

We build and scale B2B and DTC commerce on Shopify Plus, BigCommerce and WooCommerce, wired into NetSuite and Oracle. These articles are written by the engineers and architects who deliver those projects.

Building something like this?

We architect B2B commerce and ERP integrations end to end.

ERP Integration

Real-Time Inventory Between NetSuite and Shopify Plus B2B

Most NetSuite-to-Shopify integrations poll on a timer. For B2B that is too slow: a single wholesale order can move a thousand units, and the next buyer sees stock that no longer exists. The fix is event-driven sync with a reconciliation pass behind it.

Read
B2B Commerce

Customer-Specific Pricing That Survives 40,000 SKUs

The instinct is to precompute every customer-product price. With 800 accounts and 40,000 SKUs that is 32 million rows that go stale the moment a contract changes. The alternative is to store rules and resolve them — with caching designed around how buyers actually browse.

Read
AI & Automation

MCP Servers for Commerce Operations, Not Demos

Most commerce AI demos are a chat box over a product catalogue. The useful version connects an agent to the systems your operations team already uses, through a typed tool interface with real permissions. That is what MCP is for.

Read
Book