An interface specification is the detailed technical document that defines a single system-to-system data exchange within an enterprise technology architecture — specifying exactly what data is sent, from where, to where, in what format, when, through what mechanism, and with what error handling. It is the primary build and test reference for one interface. Where the integration architecture document defines the structure of all interfaces collectively, the interface specification provides the implementation-level detail for each individual interface — at a level of precision that an integration developer can build from without requiring further clarification.
Interface Specification Content
| Section | Content |
|---|---|
| Interface identifier and name | Unique ID, human-readable name, source system, target system, data domain |
| Trigger mechanism | Scheduled batch (with exact schedule), event-triggered, API call, manual initiation — with the triggering condition defined precisely |
| Source schema | Source system, table or API endpoint, field list with data types, and any source-side filtering or pre-extraction logic |
| Target schema | Target system, table or API endpoint, field list with data types, and mandatory field requirements |
| Transformation rules | Field-level mapping from source to target, including lookup tables, concatenation logic, default values, and conditional transformations |
| Volume and frequency | Expected record count per run, maximum volume at peak, run frequency, and data latency requirement |
| Error handling | Partial failure behaviour (load records that pass, reject those that fail, or reject the full batch), error notification mechanism, retry logic, and manual resolution procedure |
| Acceptance criteria | Measurable criteria used to confirm the interface is working correctly in testing — record count reconciliation, value reconciliation, timing threshold |
For Oracle EPM Data Management (formerly FDMEE) interfaces, the interface specification translates directly into the FDMEE source system configuration, import format definition, mapping rules, and data load rule — making the specification the de facto configuration guide for the integration developer.
Common Gaps and Failure Modes
The specific failure that most consistently causes integration testing to consume disproportionate programme time is interface specifications that define the transformation rule at a category level — “map source accounts to target accounts using the account mapping table” — without defining the mapping table itself. The integration developer must then discover, through trial and error in testing, which source values map to which target values. Each unmapped value produces a load failure that requires investigation and resolution. Writing interface specifications that include the complete mapping table — or a confirmed reference to a separately maintained mapping document — eliminates this class of testing delay.
How Loop Wise Solutions Produces This
Loop Wise Solutions produces interface specifications for every integration in scope during the detailed design phase, using a standardised template that enforces completeness across all required sections. Specifications are reviewed in a structured walkthrough with both the technical build team and the business process owner for the data domain — confirming that the technical definition matches the business intent before build begins. A specification that has not been walked through with both audiences is not complete, regardless of how detailed it appears on paper.
