A hosted layer where integrations live instead of being scattered across the systems they connect.
An integration platform as a service is a hosted environment for building, running and monitoring integrations between applications. It supplies connectors, transformation, routing, scheduling, error handling and monitoring, so each integration is configured on a shared platform rather than built separately.
Point-to-point integrations are cheap individually and expensive collectively. Each has its own error handling, its own logging, its own schedule and its own author, and the number of them grows faster than the number of systems. The platform's value is not the connectors, it is having one place to look when something has not arrived.
Pre-built adapters for common applications, handling authentication, pagination and rate limits you would otherwise write.
Mapping between source and target structures, configured and versioned rather than coded in each endpoint.
Multi-step flows with conditional routing, retries and compensating actions when a later step fails.
Run history, failures and latency across every integration in one view — the single largest practical benefit.
Environments, deployment promotion and change control, so integrations are managed like other software.
A business adopts an iPaaS platform to connect its ERP, CRM, storefront and warehouse system. The licence looks reasonable and the connector library covers all four.
Two costs emerge that were not in the business case. The first is volume: pricing is per message, and a stock-level sync running every fifteen minutes across twelve thousand items generates far more messages than anyone estimated. Restructuring to send only changed items cuts this by over ninety percent, but it required rework.
The second is that connectors handle the connection, not the mapping. Deciding that this system's customer type maps to that system's account category, and what to do when a value has no equivalent, is business analysis that no connector performs. Teams that budget for the platform and not for the mapping work consistently find the project costs more than planned — the platform removed the plumbing, which was never the expensive part.
| Term | What it means |
|---|---|
| Connector | A pre-built adapter to a specific application, which is where most of the time saving comes from. |
| Flow | A configured integration between systems, with its mapping and logic. |
| Mapping | The field-level translation between two systems' data structures. |
| Orchestration | Coordinating several steps across systems into one logical process. |
| Consumption pricing | Charging by message volume or flow execution, which is where costs surprise people. |
Verify the specific systems and the specific operations you need. A connector that exists may not cover your call.
Many platforms charge per message or per task. Model it at real volume before committing.
Transformation belongs there; business rules usually do not. They become invisible to the systems that own them.
Flows built in one platform rarely port to another. Factor the exit into the decision.
| Metric | What it tells you | What distorts it |
|---|---|---|
| Message volume against plan | Consumption pricing makes this a cost line, not just a technical measure. | |
| Flow failure rate | Integration health, by flow rather than overall. | |
| Time to build a new integration | Whether the platform is delivering the speed it was bought for. | |
| Flows built but unused | Abandoned integrations still consuming licence and attention. |
We build and support the systems this page describes. Start with a free working proof of concept — no cost, no commitment.