Case study · Scout
Scout: importing large supplier catalogues without starting over at the slightest error
On Scout, suppliers need to put entire catalogues online from files whose format varies from one supplier to the next. Entering product records one by one is not realistic, and an import must be able to fail in part without everything having to be redone. We built a complete import flow: column mapping, reusable settings, validation, background processing, review of changes, resumption and correction of only the failed rows.
Scout is a B2B procurement platform for property organisations that manage properties made up of apartments or units. It connects these organisations with product suppliers and service providers, who present their offering there and receive their orders. The platform is built with Medusa, for the commerce foundation, and Mercur, for the multi-supplier dimension. For a supplier, joining Scout first means publishing its catalogue there, then keeping it up to date. This catalogue usually already exists elsewhere, as a file exported from its own tools, with its own column names, units and data-entry habits. In the supplier workspace, we designed a flow that accepts these files as they are and takes them all the way to the published catalogue. The supplier stays in control at every step: they see what is going to change, they can stop, and they can fix a problem without rerunning the whole file.
Starting from the supplier’s file rather than an imposed template
Every supplier exports its products in its own way. A column is called “Désignation” at one supplier and “Libellé article” at another; a value is expressed one way in the file and must be understood differently on the platform. Requiring each supplier to rework its export to match a single template would have shifted the entire preparation effort onto its side, with every new delivery.
The flow therefore begins with uploading the CSV file to the supplier workspace. The supplier then indicates which Scout field each of its columns corresponds to, and how certain values should be converted. The original file does not need to be rewritten: it is the mapping that links the way the supplier describes its products to the way the platform presents them to buyers.
Settings saved for subsequent deliveries
A catalogue is not imported just once. Prices change, new items arrive, others disappear, and the supplier sends back a file built on the same model. Redoing the entire mapping with each upload would be tedious and error-prone.
The supplier can therefore save its configuration as a reusable setting. On a subsequent import, it finds its account’s settings and starts from the appropriate one. If its own tool changes the structure of the export, it corrects the existing setting instead of creating a new one. And when a setting no longer matches any file, it can delete it to prevent it from being reused by mistake.
Checking before loading, and loading into the right account
Before products join the catalogue, the file goes through a validation step. The supplier thus sees problems at this stage rather than after publication. The flow also includes a step that classifies products in Scout’s taxonomy. Depending on the mechanism used, a product may be classified automatically, flagged for review, rejected or left unmatched.
On a platform where many suppliers work side by side, an import must also stay within its scope. An uploaded file is attached to the account of the supplier that sent it and cannot be imported for another. Likewise, the import is refused as long as the supplier has no stock location in its name: the quantities loaded cannot end up in another account’s warehouse. These refusals halt the flow with a clear reason, instead of letting through wrongly attached data.
Background processing with visible progress
A large catalogue cannot be processed in a single click. If the import depended on a page being left open, closing a tab or losing the connection would be enough to compromise it. Once launched, processing therefore continues in the background, independently of the browser.
The supplier views its ongoing or completed batches, filtered by type and status, along with their progress. It knows where its import stands without rerunning the file to check. If it realises that a creation batch was launched by mistake, for example with the wrong file or the wrong setting, it can cancel it as long as the processing still allows.
Seeing what changes before applying it
Updating an existing catalogue carries another risk: unknowingly overwriting work already done on the platform, such as a reworked description or selected visuals. The flow therefore compares each imported row with the product as it currently exists, including its prices and images. The supplier sees exactly what the file would change.
It then decides row by row: it confirms the updates it accepts, or all of them at once if everything suits it. For a difference it does not accept, the current data is kept. As an illustration, a supplier of maintenance supplies that sends its file again to update its prices can accept the new prices while keeping the enriched descriptions on Scout. Only the information the file is meant to provide is applied, and a price update does not rewrite the entire product record.
When the review is complete, the supplier closes it and obtains the final counts for the processing. It knows that this file has been processed and can move on to the next one.
Picking up where you left off, correcting only what failed
With a large file, it is common for some rows to cause problems. Restarting the full import for a handful of errors would reprocess what had already succeeded. The flow offers two other options.
An interrupted import can be resumed: the work already loaded is kept, and processing continues instead of starting from scratch. For products that failed, the supplier can retrieve only those rows, correct them and resubmit them, without resending those already processed.
Staying in control of the catalogue after the import
Importing is not the only way to act on the catalogue. The supplier can create a product with its variants and images, edit an existing product record, its prices, dimensions or weight, and assign it a brand and a manufacturer chosen from shared lists. It does not need to reimport a file to correct a detail.
It can also export its full catalogue as CSV, with variants, prices, stock and images, to check it or reuse it. From a file, it can deactivate a selection of items that have become obsolete. And when it wants to start again from an entirely new base, it can delete its whole catalogue in one explicit operation, whose progress it can follow.
For a supplier on Scout, importing a catalogue no longer depends on a file that is perfect first time. It starts from its own tools’ export, sees what an update would change before accepting it, and corrects only the failed rows rather than rerunning everything. Property organisations thus find catalogues fed from existing files, and each supplier remains in control of what is published under its name.