Bidirectional CRM Sync: Salesforce ↔ Administrate
A 15-minute, event-driven pipeline that keeps accounts, contacts and opportunities perfectly aligned between Salesforce and Administrate — with 100% workflow health across all active entities.
The challenge
Administrate (training & event management platform) and Salesforce both held CRM data that had to stay in sync. Records were managed manually, duplicated across systems, and there was no reliable way to keep accounts, contacts and opportunities aligned in both directions.
The solution
We designed and delivered a bidirectional integration running on a self-hosted n8n instance, with an orchestrator that processes a sync queue every 15 minutes.
- Salesforce → Administrate: accounts, contacts, opportunities and line items flow as batch upserts matched by external ID.
- Administrate → Salesforce: account and contact profiles flow back as composite batches.
- 11 workflows (orchestrator, source/target pipelines, canonical transform, finalizer, error notifier) + 10 data tables (integrations, mappings, sync queue, sync state, sync runs, sync errors, execution log, tenants…).
- Loop prevention (PATCH by Salesforce Id) and conflict resolution (last-write-wins) built in.
- Dual-channel alerting on failures via Slack and Telegram.
Results
- 100% workflow health across all active entities (accounts, contacts, opportunities, line items, profiles).
- Handles pagination of 2,000+ record responses and batches of 200 records per run.
- Designed for a worst case of ~350k updates/day, with ~70% filtered out at the source.
Geschäftsprobleme
- Data lived in two systems (Salesforce + Administrate) with no reliable sync
- Manual data entry caused duplicates and drift between CRM records
- High projected volume (~350k updates/day worst case) needed a scalable, incremental design
Wesentliche Vorteile
- Bidirectional, incremental sync keeps both systems in agreement automatically
- Batch upserts matched by external ID — no duplicate records
- Loop prevention and conflict resolution built into the pipeline
- Failures surface instantly in Slack and Telegram via a dedicated error notifier
Anwendungsdetails
- Self-hosted n8n with 11 shared + tenant-specific workflows
- 10 n8n data tables configure entities, mappings, queue, state and alerts
- Orchestrator runs every 15 minutes with 200-record batches
- Event & registration sync pre-built and ready to activate