A supplier's catalogue is rarely ready for publication without editing. Names, units of measurement, packaging options and photographs may be inconsistent. Importing that file carries the problems into your shop: filters behave inconsistently, prices are unclear and stock levels refer to the wrong variant.
Before importing, prepare one consistent dataset that both the website and the shop team can use. Start with a small product group, agree on the rules and then apply them to the whole range.
Assign data sources and responsibility
Decide where names, prices, stock levels and images come from. Sources may differ: the manufacturer provides specifications, an inventory system holds stock levels, and your team writes the descriptions.
Assign someone to keep each type of information current. Agree which system is authoritative and what happens when records disagree.
If prices are changed in both the shop and the inventory system, the next synchronisation may restore an old value. Update rules therefore need to be in place before the first bulk import.
Separate the model, variant and identifier
A model brings related variants together. A variant is a specific item sold, such as a T-shirt in a particular colour and size. It may have its own SKU, price, photograph and stock level.
Do not use the name as the only way to identify a product: wording can change. You need a stable identifier so the system can update an existing record instead of creating a duplicate.
The SKU, the platform's internal ID and the manufacturer's code may serve different purposes. Record which field is used for imports and which for exchanging data with your inventory system. Do not change these values during routine catalogue editing.
Prepare a working data template
A spreadsheet is a convenient place to maintain the product file. First agree on the fields and how they should be completed, then adapt them to your chosen platform's format.
Include the following in your working template:
- Identifiers: a stable item code, SKU and a link to the parent model if the item is a variant.
- Content: name, short description, full description and included items.
- Structure: category and attributes used by filters.
- Variants: size, colour, capacity or another distinguishing feature.
- Sales terms: price, currency, unit of sale and rules for displaying taxes.
- Availability: stock quantity or an agreed availability status and waiting time.
- Shipping: weight, dimensions and units if needed for calculations.
- Images: an image file or URL, image order and association with the variant.
- Language versions: translated names and descriptions linked to the same identifier.
This is a checklist for preparing information, rather than a universal upload format. Column names and the way variants are represented depend on the platform. For example, Tilda describes separate rows for variants and their link to the parent product in its catalogue import instructions.
Agree on consistent data entry rules
Use controlled lists for attributes, including consistent material names, colours and units. Values such as “black”, “Black” and “black colour” should not accidentally produce three filters in one language version.
Store numerical specifications in a usable form, with the value separate from its unit if the system supports this. Pay particular attention to decimal separators, leading zeros in SKUs and codes automatically being converted into dates.
Distinguish a blank field from zero. A missing price must not automatically make a product free, and missing stock data must not confirm availability. Check how the importer handles each state.
Keep identifiers and commercial data consistent across language versions when the sales terms genuinely match. Translating a name should not create a new, independent inventory item.
Prepare images and check their associations
Name files so they can be linked to the product and variant. Make sure the platform can access image URLs and that they point to the intended images, rather than a preview page or a private folder.
Check the main image, the order of additional photographs and the correct colour. If you use AI visuals, prepare source material and accuracy checks separately, as explained in the article on preparing a product for AI visualisation.
A practical process before bulk importing
Hypothetical example: the catalogue contains a mug in two colours and a set of two mugs. This could be one model with colour variants and a separate set with its own inventory record. The decision depends on how the shop sells and stores these items.
Use this small sample to check your preparation:
- Agree on identifiers, units of sale and variant associations.
- Complete all required fields and check images.
- Import records into a test or unpublished catalogue.
- Compare product pages, filters, prices and stock levels with the source data.
- Change one value and repeat the import, checking that it updates without duplicates.
- Record errors and correct the template rules.
- Back up the data before importing the rest of the range.
When developing an online store, the catalogue's contents and source quality affect the scope of work. Separate data preparation, import configuration and result checking in the project estimate. Other budget components are covered in the article on online store launch costs.
A prepared catalogue should support both the initial upload and reliable updates after launch.