Third-party Integrations

Integration work is where optimistic project plans meet reality. The vendor documentation describes the happy path; your actual data has duplicates, the sandbox behaves differently from production, and the rate limit is lower than advertised. Doing this well is mostly about designing for the failure cases, because they are not edge cases — they are Tuesday.

What this covers

Reliable connections between the systems you run, built to handle the failures that integrations inevitably encounter.

• CRM, ERP and accounting system integration
• Payment and billing provider connections
• Data synchronisation and conflict resolution
• Webhook handling and event processing
• Retry, backoff and failure recovery
• Field mapping and data transformation
• Monitoring, alerting and reconciliation

Where integrations usually go wrong

Assuming delivery. Webhooks get missed, networks drop and vendors have outages. Integrations without retry logic and reconciliation lose data quietly, and you find out from a customer.

No idempotency. A retried request that creates a duplicate order is worse than the original failure. Every write needs to be safe to repeat.

Sync without conflict rules. When two systems can both edit the same record, someone must decide what happens when they disagree. If nobody decides, the last write wins and data degrades over time.

Silent failure. An integration that stops working without alerting anyone is discovered weeks later, usually during a financial close.

How we build it

1. Map the data. Which system owns which field, how records match across systems, and what happens when they conflict.
2. Design for failure. Retry policy, idempotency keys, dead-letter handling and reconciliation designed in from the start.
3. Build and verify. Implementation against sandbox and production behaviour, with tests covering the failure paths rather than only the happy one.
4. Monitor and reconcile. Alerting on failures and scheduled reconciliation that catches discrepancies before they compound.

Technologies we work with

• Platforms: Salesforce, HubSpot, NetSuite, Xero, QuickBooks
• Payments: Stripe, PayPal, Adyen, Braintree
• Patterns: REST, GraphQL, SOAP, webhooks, SFTP
• Queueing: RabbitMQ, SQS, Redis, Kafka
• Backend: Node.js, Python, PHP, Go

What you get

• Documented field mappings and ownership rules
• Integrations that recover from failure automatically
• Alerting when something needs human attention
• Reconciliation reporting you can trust at close
• Source code and infrastructure you own

Frequently asked questions

Can you integrate with a system that has no API?

Often, yes — through database access, scheduled file exchange, or automation against an existing interface. These approaches are more fragile than a proper API and we will be clear about the trade-offs before you commit.

How do you handle vendor outages?

Queue the work, retry with backoff, and alert when retries are exhausted rather than dropping the data. The integration should degrade rather than fail.

What about ongoing maintenance?

Vendors change APIs and deprecate versions, so integrations need occasional attention. Most clients keep a maintenance arrangement covering exactly this.

How long does an integration take?

A straightforward one-way connection can take a couple of weeks. Two-way synchronisation with conflict resolution across several systems is substantially more involved, and the data mapping usually takes longer than the code.

How we start

Every engagement starts with a short, fixed-scope conversation. You leave with a plan and a price whether or not you continue with us.