A product catalogue that holds together in every channel
Peter Nilsson
The customer opens your website and sees a price for the item. The salesperson sends a quote with a different price. The spreadsheet that went to the CFO shows a third. None of the figures is made up, and that is exactly the problem.
A product catalogue is an agreement about which system owns each piece of data, who may change it and how it reaches each channel. Without that agreement, you get several versions of the same truth, and every version costs someone time.
For a company with a few thousand items, it shows within a few weeks. The range changes, prices move and the channels multiply. The hard decision is which system owns which field.
What a product catalogue actually contains
A product record is the smallest unit of the catalogue. It needs to carry everything a buyer, a salesperson and a business system need to find, compare and order the same item. In practice, that means a number of fields that live in different places.
The item number and SKU, a stable number that does not change over time, come from the business system. The EAN, the barcode for each packaging level, comes from the business system or PIM depending on where the item is created. Customer prices and contract prices per customer, quantity and validity period belong in the business system, as does the stock level per warehouse.
PIM owns the rest: product names and descriptions for each language version, categories with product group and subgroup, variants such as size, colour or material, technical attributes such as dimensions in millimetres or pressure rating in bar, relationships to accessories and spare parts, images and documents such as datasheets, manuals and certificates, and Swedish and English copy for the same item.
The split looks simple on paper. It still decides which system may overwrite what. An item number that changes when marketing renames a product creates a new item in the business system, and you lose history, agreements and order statistics.
Images and body copy are enough for a buyer who just needs to recognise a product. A purchaser choosing between three options needs something else: dimensions, pressure rating, material, compatibility and available quantity. Without those fields, the website becomes something to look at, and the comparison happens in a spreadsheet on the side.
Take a customer ordering a spare part. What decides whether the order lands correctly is rarely the image. It is that the spare part has the right technical attributes and the right relationship to the main product, and that the stock level comes from the right warehouse. If one of the fields above is missing, the error has consequences far down the flow: the pick list, the invoice and the next repeat order.
How product data, categories and channels fit together
Split responsibility by what the data is used for. The business system owns transactional values such as item numbers, prices, stock levels and orders. PIM owns enrichment, classification, translations and publishing rules. The rule is easy to remember and hard to keep, because both systems can edit the same product.
The reason to keep them apart is concrete. If someone types a new name straight into the business system, editorial changes risk overwriting transactional data. If someone copies prices into PIM, you get two versions of the same price, and neither is clearly authoritative. Many organisations solve this with a manual check afterwards, which becomes the most expensive part of the whole solution.
The flow has four steps. The business system supplies item numbers and core data. PIM receives them and enriches them with copy, attributes, images and documents. From there, the content is published to the website, customer portal and reseller channel. Orders and stock levels go back to the business system through the integration.
The connection has to deliver machine-readable data that other systems can read. An ERP integration that only fills a web page is not enough when the customer wants to pull items, prices and stock into their own business system. A catalogue for business customers needs stable item and variant IDs, and pricing logic tied to customer, quantity and validity period.
This is how the process looks from core data to published channel.

Categories work the same way. The same category model in PIM is used for the website's navigation, the customer portal's filters and the reseller portal's sorting, but each channel gets its own display rules. The product group "pumps" sits in the same place in the model whatever the channel. The website shows it as a menu, while the reseller portal shows it together with the stock level per warehouse. A changed category in PIM takes effect in every channel at once, which is why categories are owned in one place.
Categories, variants and attributes that make the catalogue usable
A pump in three sizes and two materials is one product with six variants. The spare part that only fits two of the sizes is a separate item with a relationship to the main product. Product, variant, accessory and spare part are different objects with different rules, and a manufacturer that mixes them up ends up with a catalogue nobody can filter.
Every attribute needs four things: a data type, a rule for whether it is mandatory, a list of valid values and an owner. Dimensions are stored as numbers with units, never as free text. Then you can filter on dimensions, validate against a range and export the same value to a reseller without manual translation. Free text in the dimension field is the most common reason a filter stops working after six months.
For a purchaser, dimensions, compatibility and area of use decide whether they find the right product. The keyword in the heading rarely helps someone who needs to replace a seal in an existing machine. That buyer filters on thread, diameter and pressure rating, and only then on price.
An implementation usually starts with one product family or market. First, the attribute model for filters and variants is defined, then the data sources are connected through PIM and the business system where needed. After that, the catalogue grows step by step, and functions are added at the same pace. We have previously described how a product information data model is built up with field types and relationships.
Variants, attributes and relationships fit together in the same model.

A task you can do in an hour: write down the three most important attributes for your target group and give each one a named owner. Without an owner, the attribute is a wish, and it will be empty in the catalogue a year from now.
The same product has to match in every channel
Swedish companies sell through more than one digital route. According to Statistics Sweden's figures on ICT use in enterprises, 36.7 percent of companies with at least ten employees sold online in 2024. Of these, 31.2 percent sold via the web and 11.2 percent via EDI. 5.7 percent used both channels.
The difference is practical. The website has to present and enrich information for people. A company that sells in both channels needs the same identity and the same price in both. EDI needs structured fields and stable data flows between systems, where nobody reads the text and only the fields are interpreted.
Statistics Sweden counts e-commerce as orders via a website, an app or EDI-type messages, and the payment does not have to be made electronically. The definition says something about the requirements: the same item data has to be readable by a person in an interface and by another system in an order file.
Manufacturers have a particularly strong need for structured product information. Industrial products have more variants, technical attributes, compatibility rules, documents and spare-part relationships. The information is also used by salespeople, customers, resellers and service organisations, so the catalogue's model matters more than the publishing surface.
An example of what goes wrong: a reseller orders via EDI and gets a packaging unit that does not match what the website shows for the same item. The order goes through, the stock level drops and the error is only discovered at goods receipt. The same item had two identities in two channels, and neither was wrong in itself. How item and order data meet between the channels and the business system is a separate question, which we cover in the post on e-commerce and the business system.
Start with one product family. Take the family with the most variants and the most documentation, and check that item number, unit and price are identical on the website, in the portal and in the EDI flow before you touch the rest of the range. Write down the channels where the same product exists today and compare the three fields in each. The discrepancies you find are your list of integration requirements.
What the catalogue requires after launch and when it does not fit
A catalogue project is not finished when the website goes live. Prices, stock, product status and rules change every week, and a web catalogue can look correct while a field in the business system is wrong. The error first shows up in order handling.
Four things need to be written down before you can talk about ongoing management. First, who owns each attribute group. Second, how quickly a change must reach the channels, expressed in hours or days per field group. Third, how a conflict between the business system and PIM is resolved. Fourth, how data quality is measured per channel, with completeness, timeliness and the share of products that can be bought without manual handling.
Different systems can be authoritative for different fields. It is entirely possible for the business system to own the price while PIM owns which customers see which price. What does not work is letting both write the same value without a rule. A named owner per attribute group and a routine for changes solve more than an extra integration.
This approach fits less well when the range is small and rarely changes. Manual publishing can then be cheaper to run than a full model. It also does not fit when the business system lacks stable item numbers. Clean up the core data first, or you move the same disorder into the new catalogue and pay for it twice.
Include ongoing management, upgrades of the platform and business system, and the work of keeping fields correct in the total cost of ownership. Two things can be checked straight away in your current situation. Every attribute group should have a named owner, and every change should go through a routine for approval and publishing. If either is missing, that is the first thing to fix.
Three questions to ask the supplier
Ask which fields are owned by your business system and which by the product information, and who approves changes to prices and item data. The answer shows whether the supplier has thought through the split of responsibility. Without a clear owner per field, the catalogue becomes a second copy of the business system within a year, with its own prices and its own item numbers.
Then ask what happens when an attribute changes in the business system after publishing, and how quickly the change shows on the website and in the EDI flow. The answer reveals whether the integration is built for ongoing operation or for a one-off migration. An answer without a time frame means nobody has measured it, and you will discover the lag through a customer.
Finally, ask which data quality the supplier measures in operation and how often you get a report. Completeness, timeliness and the share of products that can be bought without manual handling are reasonable measures to track per channel. The answer shows whether ongoing management is planned or will have to be invented after go-live.
A supplier that cannot answer these three questions has no plan for the catalogue once the project is over. That is worth knowing before the contract is signed, whichever platform you choose.
Next steps for your product catalogue
Choose one product family or market to start with. List which of the eleven fields above you have today, which are missing and which system owns each field. The result is the basis for a first attribute model, and the work usually takes an afternoon with the right people in the room: someone from sales, someone from IT and someone who owns the range.
On the page about product information and PIM, we describe how you gather, enrich and publish product information from one source to the website, portals and resellers, with item numbers and prices kept in the business system where they belong.
In a 72-hour analysis, you set aside two hours, and within 72 hours you get an inventory of functions and integrations, a functional specification, a prototype and a complete cost picture with a timeline, free of charge.

About Peter Nilsson
Peter Nilsson writes about integrations and product information, focusing on how business systems, PIM and the web can share the same data without duplicated work.

ERP and EDI that keep orders and invoices together
ERP EDI integration explained: which messages are included, how to validate data, and which metrics indicate whether the flow is working.
Peter Nilsson 
ERP systems as the hub for your website and portal
Business systems consolidate finance, orders, inventory, and product data into a single source. This is then connected to the web, portals, and EDI.
Peter Nilsson
PIM for wholesalers and who owns the product data
PIM for wholesalers starts with deciding who can change what. How to split ownership between ERP and PIM, choose stable keys and set limits on where AI can enrich product data.
Peter Nilsson