Skip to main content
E-commerce integrations

Magento, ERP and PIM: who should manage products, prices and stock?

A price is overwritten after a correction, a product stays incomplete or stock no longer reflects what can be sold. The connector is not necessarily at fault: several tools may believe they own the same information. Before adding synchronisation between Magento, an ERP and a PIM, define who decides, who transmits and who checks.

MagentoBy Julien BaillyText updated on 5 min estimated read

Illustration of three separate groups connected by glowing channels, representing exchanges between commerce, product information and business management

Key takeaway

Assign a source of truth to each piece of data, define its transformation rules and plan for conflicts, errors and retries. There is no universal division of responsibilities between Magento, ERP and PIM: it depends on your organisation and tools.

Describe responsibilities before choosing a transport method

An ERP generally handles business management processes, a PIM organises product information and Magento supports the sales journeys. These are starting points for discussion, not a rule automatically assigning every price or stock value to one tool. Some businesses have other systems or do not use a PIM.

Magento provides APIs to exchange data with external services, including management systems. Their availability enables integration; it does not determine which team should maintain each piece of data. Business responsibilities must be understood before choosing between an API, a file or intermediate processing.

Break data down by use rather than assigning entire product records

A product combines several types of information: identifier, labels, attributes, images, sales rules and availability. They may have different owners. Saying “the PIM manages products” does not specify who creates the identifier, validates a translation or authorises publication for sale.

Also identify the contexts that change a value: store, language, customer or warehouse, depending on the chosen setup. A mapping rule should explain what each field means and what happens when a value is missing. The same column name in two tools does not guarantee the same meaning.

Write an explicit division of responsibilities, then have it approved

Here is a fictional example for discussion: the ERP assigns identifiers and management data; the PIM enriches descriptions and media; Magento applies the sales rules assigned to it and presents products. This proposal must be adapted to your business, not copied as a ready-made architecture.

For each piece of data, identify the source of truth and the tools allowed to change it. Also specify how local corrections are handled. If someone adjusts information in Magento, should the next import overwrite it, preserve the exception or flag a conflict? That decision deserves to be written down and tested.

  • Data: its meaning and identifier across tools
  • Source of truth: the team and system that validate it
  • Distribution: destinations, frequency and transformations
  • Conflict: the rule applied when a value changes on both sides
  • Check: how a failure or inconsistent value is detected
Fictional example of responsibilities, to adapt to your organisation
DataSource of truth chosen in this exampleMagento's role
Product identifierERPPreserve the mapping between tools
Descriptions and mediaPIMPresent information suited to the store
Reference priceERPApply the agreed sales rules
Physical stockERPReconcile movements with availability rules
Online orderMagentoTransmit the order and track its acceptance

Distinguish physical stock from what can be sold

The number of items in a warehouse and the quantity available to sell answer different questions. In Inventory Management, Magento distinguishes sources, stocks linked to sales channels and salable quantity. Reconciliation must account for how these elements work, not simply compare two numbers displayed at a given moment.

Ask what happens when an order arrives, when it is cancelled and when an adjustment comes from the management system. The aim is to check the consistency of the complete journey. Writing a quantity directly to the wrong place may bypass the very rules you wanted to preserve.

Plan for errors, duplicates and retries from the start

An exchange must also work when part of the system is unavailable. After a timeout, you may not know whether the destination applied the action. Resending the same order without a reconciliation identifier can create a duplicate; ignoring the failure can leave an operation incomplete.

I recommend defining identifiers, operation order and what a retry is allowed to repeat. A manual correction should also leave a useful record. These are operational decisions: simply choosing a connector does not replace them.

  • Can an operation that has already been processed be recognised?
  • Who receives the alert, and with what information?
  • Can only the failed items be retried?
  • How do we check that source and destination are consistent again?

Give each integration the access it needs

A connection to products does not necessarily justify access to every piece of data and every action. Commerce administration allows specific resources to be authorised for an integration. Permissions, credential storage and responsibility for renewing them are part of the technical scope.

Documentation for a data flow should remain usable when the provider changes. It describes its purpose, the data exchanged, the owners and how to check its status. The secrets themselves remain separate from this working documentation.

Test a complete business journey before expanding the data flow

For a first version, choose a few representative products, a normal case and known exceptions. Follow a change from its source through to its effect in the store. Then check order transmission, anomalies and recovery after a simulated interruption in a suitable environment.

Validation should involve the people who use the data: sales, catalogue, logistics or management, depending on the organisation. A technically successful synchronisation may still send information to the wrong place. Acceptance testing must therefore verify the business rule, not just the API response.

Sources and references

Documentation consulted when writing this guide. Advice and procedures should be adapted to your own context.

Related projects