When Integration Strategy Defines Product Value
The most successful custom applications aren't built in isolation. They thrive at the intersection of existing systems, pulling data from accounting platforms, pushing updates to CRMs, and bridging decades-old databases with modern interfaces. Integration architecture has moved from afterthought to core product strategy.
Consider a client portal that displays invoice status. Without Xero or MYOB integration, staff manually update records. With proper connectivity, invoice creation, payment status and outstanding balances flow automatically. The value isn't in another database of financial records—it's in seamless access to authoritative data where it already lives.
This shift changes how we architect applications. Instead of building comprehensive systems that duplicate functionality, we construct intelligent middleware that connects, transforms and presents data from multiple sources. The integration layer becomes the product.
Accounting Platform APIs: Xero and MYOB Patterns
Xero and MYOB dominate Australian accounting software, each with distinct API characteristics. Xero's OAuth 2.0 implementation supports multi-tenancy well, with separate connections per organisation. The API returns JSON with predictable structures for invoices, contacts and payments. Rate limits sit at 60 calls per minute per token, which sounds generous until you're syncing 5,000 invoices with line items.
MYOB AccountRight requires on-premise or cloud deployment consideration. The API distinguishes between company files, each needing separate authentication. Batch operations help with volume, but error handling differs from Xero. A failed line item can invalidate an entire invoice creation request, requiring granular validation before submission.
Both platforms support webhooks for real-time updates, though implementation varies. Xero webhooks fire on specific events (invoice paid, contact modified). MYOB webhooks may require polling fallbacks for certain data types. Designing for both means abstracting platform differences behind a common interface while preserving platform-specific optimisations.
- OAuth token refresh logic with credential encryption and rotation
- Idempotent sync operations to handle duplicate webhook deliveries
- Conflict resolution when local edits clash with accounting system changes
- Audit trails showing data lineage from source system to display
CRM Integration and Customer Data Synchronisation
Salesforce, HubSpot and Microsoft Dynamics maintain customer relationship data that feeds sales, support and marketing workflows. Integrating with these platforms means understanding object models—leads, contacts, opportunities, cases—and how they map to your application's domain.
A service management platform might pull Salesforce accounts to pre-populate customer details, then push completed jobs back as closed cases. Bidirectional sync introduces complexity. Which system holds truth for phone numbers? How do you handle a contact deleted in the CRM but referenced in historical job records? These aren't technical questions alone; they require business logic decisions encoded in integration rules.
Custom objects and fields add another dimension. Standard CRM APIs cover common entities, but many organisations extend their CRM with industry-specific data. Integration code must accommodate custom schemas without constant redeployment. Configuration-driven field mapping, where administrators define which CRM fields populate which application properties, provides flexibility without code changes.
Legacy System Bridges: SOAP, Direct Database and File Transfers
Not every integration target offers a modern REST API. Manufacturing systems, enterprise resource planning platforms and warehouse management software often expose SOAP services or, in some cases, direct database access. Connecting these systems requires different techniques and additional caution.
SOAP integrations mean XML parsing, WSDL interpretation and envelope construction. A parts inventory system might require specific header authentication alongside body payloads for stock level queries. Error messages arrive as fault codes needing translation. While REST patterns have largely supplanted SOAP in new development, decades of enterprise software still speak this dialect.
Direct database integration—reading from or writing to a legacy system's tables—offers speed but significant risk. Schema changes in the source system can break queries. Bypassing the legacy application's business logic might skip validation or trigger rules. When necessary, read-only views, stored procedures with defined contracts, and change data capture patterns limit exposure. The goal is controlled access, not a free-for-all on production tables.
- Scheduled file transfers (CSV, EDI) with validation and reconciliation steps
- Message queues for asynchronous processing of high-volume legacy data
- Transformation layers that normalise legacy formats to modern schemas
- Fallback mechanisms when legacy systems experience downtime
Building Resilient Integration Pipelines
Integration failures are inevitable. APIs go down, rate limits trigger, network timeouts occur. Resilient design acknowledges this reality and builds accordingly. Retry logic with exponential backoff handles transient failures. Circuit breakers prevent cascading failures when a downstream system becomes unavailable.
Dead letter queues capture failed integration attempts for manual review. A payment sync that fails after three retries lands in a queue where staff can investigate whether the issue was temporary downtime, invalid data, or a changed API contract. This visibility transforms integration from a black box into a manageable process.
Monitoring and alerting complete the picture. Tracking integration success rates, latency percentiles and error patterns reveals degradation before users notice. An alert when Xero sync latency exceeds two seconds might indicate API throttling or network issues worth investigating.
Integration Architecture as Competitive Advantage
Applications that connect well multiply their value. They eliminate double-entry, provide real-time visibility across systems, and free teams from manual reconciliation. For many organisations, the custom application's primary value lies not in its own features but in how elegantly it orchestrates existing systems.
Technical execution matters. OAuth flows, webhook handlers, retry logic, conflict resolution—these details determine whether an integration delights or frustrates. But equally important is the strategic decision of what to integrate and how tightly. Loosely coupled integrations that tolerate schema evolution outlast brittle point-to-point connections.
YS Infomatics approaches integration development as first-class product work, not auxiliary plumbing. We've connected accounting platforms, CRMs, legacy ERP systems and proprietary databases for clients across industries. The result: unified workflows where data flows reliably and businesses gain visibility they couldn't achieve with off-the-shelf software alone.