System integrations
En Elumad diseñamos integraciones a medida entre vuestros sistemas. Con validaciones, logs y reintentos, no con scripts frágiles.
What is a custom integration?
A custom integration is the designed connection between two or more systems (ERP, CRM, eCommerce, logistics, accounting) so data flows automatically, validated and traceable. No Excel files, no manual copy-paste, no out-of-sync data.
La clave no es "tener conectados los sistemas", sino que los datos fluyan de forma fiable, trazada y sin intervención manual.
Typical features
Each integration is designed around your real systems, not generic connectors that never quite fit.
Data synchronisation
Real-time or batch data, depending on the use case. The system that owns each data field is documented and there's no ambiguity about who syncs to whom.
- One-way or two-way sync based on business logic.
- Real-time (webhooks/events) or scheduled batch (cron/jobs).
- Conflict resolution when two systems modify the same data.
- Delta sync: only changes, not the whole table each time.
APIs and webhooks
We develop adapters for third-party APIs and expose our own endpoints when the system needs it. Compatible with REST, SOAP, GraphQL and legacy formats.
- API clients for external systems (ERP, logistics, payment gateways).
- Inbound webhooks: receive events and process them resiliently.
- Secure authentication: OAuth2, API keys, JWT per provider.
- Rate limiting and HTTP error handling for unreliable APIs.
Data transformation and mapping
Every system speaks its own language. We map fields, normalise formats and transform data so information arrives clean at the destination.
- Field mapping between systems with documented rules.
- Format normalisation: dates, currencies, product codes.
- Data enrichment before syncing to destination.
- Integrity validation before writing (not after failing).
Monitoring and alerts
An integration doesn't end at go-live. We add logs, metrics and alerts to detect issues before they affect operations.
- Structured logs for every sync (what, when, result).
- Automatic alerts when a sync fails or is delayed.
- Status dashboard: last sync, recent errors, volume processed.
- Retries with exponential backoff for transient errors.
Data migrations
When historical data needs to move from one system to another, we do it in a controlled way: in phases, with validations and without unnecessary downtime.
- Data analysis and cleaning before migration.
- Phased migration to minimise risk.
- Cross-validation between source and destination.
- Planned rollback if something doesn't add up.
What we deliver
Systems and flow discovery
We map which systems are involved, what data flows between them and who owns each data field. Without this step, any integration is built on sand.
Definition of each data's source
We document which system is the source of truth for each field. Prevents silent conflicts and duplicates that appear months later.
Sync model design
Real-time, batch or hybrid depending on the case. We define frequency, direction and conflict logic before writing code.
Adapters and connectors
API clients, file readers, event listeners or direct database access depending on what each system allows. No fragile shortcuts.
Field transformation and mapping
Data rarely arrives in the format the destination expects. We map, normalise and validate before syncing.
Source and destination validations
We check integrity before writing. If data is invalid, the integration rejects it, logs the error and alerts. No corrupt data written.
Structured logs and traceability
Every sync leaves a record: what was processed, when, with what result. If something fails at 3 AM, it can be diagnosed the next morning.
Retries and fault tolerance
External systems fail. We add retries with backoff, message queues where needed, and error handling so a temporary failure doesn't become data loss.
Operational alerts
When a sync repeatedly fails or is delayed beyond a threshold, the team gets an alert. You hear about it before your customer does.
Technical documentation
Which systems are connected, how data flows, what happens if something fails and how to restart a process. Documented so it doesn't depend on one person.
QA + production go-live
We test with real data (or a replica) before going to production. We verify volumes, edge cases and behaviour under errors.
How we work
Discovery
We map the systems, data flows and current pain points. We identify the source of each data field and friction points before proposing anything.
Design
We define the sync model, the necessary transformations and the resilience mechanisms. No ambiguities before building.
Development
We build the adapters, transformation logic, validations and logging system. In phases, with real deliverables from the start.
QA + Validation
We test with real data: volumes, edge cases, simulated failures. We validate that data arrives correctly at the destination before production.
Go-live + Monitor
We activate in production in a controlled way. We monitor the first few days to catch any unexpected behaviour and respond quickly.
Common integrations
These are the most frequent connectors. If your system isn't listed here, we'll analyse it anyway.
Systems we commonly connect
-
ERP ↔ eCommerce
Stock, prices and catalogue from ERP to the store. Orders and payments from the store to the ERP. No double entry, no overselling.
-
ERP ↔ CRM
Customers, opportunities and invoices synced. The sales team sees real data without depending on manual exports.
-
ERP ↔ Logistics
Fulfilled orders are automatically sent to the logistics provider. Tracking comes back to the ERP without manual intervention.
-
ERP ↔ Accounting
Invoices, payments and accounting entries generated automatically. Monthly close without reviewing Excel row by row.
-
Legacy and internal systems
We connect proprietary systems, internal databases or flat files. Any system that holds relevant data can be integrated.
Our approach
We define the source of each data field
Before integrating, we document which system owns each field. Without this clarity, two-way integrations generate conflicts that are hard to debug.
Resilient integrations by default
Validations, logs and retries are not optional. If an external system fails momentarily, the integration handles it without losing data or generating duplicates.
What we guarantee in every integration
- Documented data source with no ambiguity.
- Validations before writing to the destination.
- Structured logs with full traceability.
- Automatic alerts on failures or delays.
- Retries with backoff for transient errors.
Real-world examples
Common situations we solve with custom integrations.
Out-of-sync stock between ERP and store
A customer buys an out-of-stock product because ERP stock doesn't reach the store. With the integration, stock updates in real time and overselling disappears.
Orders that don't reach the ERP
The warehouse team receives orders by email and manually enters them into the ERP. The integration automates the flow: the store creates the order directly in the ERP with all the correct data.
Duplicate invoices from double entry
The admin team enters sales into both the eCommerce and the accounting software. The integration automatically generates the invoice in accounting when payment is confirmed in the store.
Frequently asked questions
What we're asked most often before starting an integration project.
Related services
Tired of copying data from one system to another?
We propose a phased integration (MVP → scale), with clear deliverables and no fragile syncs.
Talk to an expert