PIM for wholesalers and who owns the product data
Peter Nilsson
You have the same item in your ERP system, on your website and in a spreadsheet from the supplier. The warehouse counts single units, purchasing orders cartons and the customer sees a price with no clear pack size. When someone corrects the data on the website, the old version is back after the next import.
PIM for wholesalers starts with sorting out who is allowed to change what. The system can gather product information and pass it on. Your organisation has to decide which value is the valid one, how the product is identified and who approves changes.
What PIM gathers and who owns the data
PIM is a system for gathering, enriching and distributing product information. Product managers can add material, areas of use, images and documents to the basic data there. The ERP system normally continues to own orders, pricing logic and stock levels.
Ownership has two parts. A source system holds the governing value. A named person is responsible for that value being correct and for errors being fixed. Writing "PIM" in the owner column is therefore never enough on its own.
The internal item number and the GTIN are the identity that lets systems recognise the same product across the ERP system, website and EDI. The ERP system owns the basic data and PIM carries the reference. Units and packaging levels, meaning single unit, carton and pallet and the conversion between them, are owned by the ERP system and used in order flows, the website and the customer portal. The same goes for list prices, customer agreements and customer-specific ranges, as well as stock levels and delivery times. Dimensions, weight and technical attributes are owned by PIM and validated against the ERP system before they go out to every channel. Sales copy, product images, data sheets and certificates are owned by PIM and published on the website, in the customer portal and in the reseller channel.
The internal item number ties your own flows together. The GTIN is the standardised product identity linked to the barcode, as described by GS1 Sweden. Store both identities with clear rules for how they relate.
Units and packaging need an owner in, for example, purchasing or logistics. Finance and sales need to maintain pricing rules and customer ranges. The warehouse manager needs to make sure stock levels and delivery estimates mean the same thing to the sales rep and the customer.
Product managers can own technical attributes while marketing is responsible for copy and images. Documents also need an appointed reviewer who checks that they are still valid. The boundaries between these responsibilities belong in your work on shared master data and ownership. Anyone who can edit a field also needs to know who is allowed to approve it.
Gather your spreadsheets, supplier emails and system views. Mark which data already has a defined home. Pay particular attention to fields where different departments each keep their own version.
How PIM connects the ERP system to the channels
The flow starts when an item is created in the ERP system with an item number, unit and price data. PIM receives the basic data and adds copy, images, attributes and documents. Each channel then gets the information it needs.
The illustration shows the order between the systems. Prices and stock levels can also be fetched straight from the ERP system through a separate integration.

According to SCB's statistics on ICT usage in enterprises, 36.7 per cent of Swedish companies with at least ten employees had e-commerce sales in 2024, with 31.2 per cent selling via the web, 11.2 per cent via EDI and 5.7 per cent via both channels. The figures apply to companies as a whole.
For your channel architecture, the overlap is what matters. Web and EDI need to work side by side. An EDI order is created in the customer's ERP system and has to match your item numbers, units and agreed formats.
Your own webshop needs copy, images, attributes and search fields from PIM, while prices, stock levels and the order flow come from the ERP system. The customer portal needs product information and technical documents per customer group, with customer-specific prices and ranges from the ERP system. The reseller channel needs sales copy and channel-specific attributes, plus list prices and stock status. EDI to business customers needs structured basic data with stable keys, and item numbers, units and prices from the ERP system.
PIM can supply structured product data to an EDI flow. The order exchange itself has to be handled by the integration responsible for the transactions. That is also where acknowledgements and rejected messages need a clear recipient.
Decide how quickly each piece of data needs to update. A product description can be published after review, while a stock message has to follow the update frequency you have agreed. The same transfer interval rarely suits every field.
Your work on ERP integration therefore needs to describe both direction and error handling. Draw which system each channel reads from. Then mark where staff re-enter the same data and who notices a failed transfer.
Why stable keys decide whether an order can be placed
A product can look correct on the website while the order is rejected. The customer picks a carton, but the order line is sent with the item identity for a single unit. The ERP system then receives a combination it has no rules for.
Stable keys let the systems tell product, variant and packaging apart. The GTIN shouldn't be your only internal business key. You also need to handle supplier relationships and several packaging levels without overwriting the wrong record.
You need four keys. The internal product identity points to the product regardless of channel. The variant identity separates, for example, size, material or finish. The packaging identity separates single unit, carton and pallet. The supplier reference links the supplier's item to your own.
Missing dimensions can mean an order has to be handled manually to calculate shipping. The wrong unit can lead to picking errors. An unclear pack quantity can trigger quote requests or a rejected EDI order.
Every attribute needs an owner, a source system, a validity period and a validation rule. For some fields, validity means the value applies until it is replaced. Other values need an explicit end date.
Validation finds the errors someone has written rules for. Use schema validation before publishing. It checks, for example, that mandatory fields are present and that values follow the permitted format. Version control should make it possible to see earlier values and roll back an incorrect change.
Separate channel views can control what resellers, end customers and internal sales staff are allowed to see. The rules need a named owner who tests changes against the receiving systems. Otherwise a technically approved import can still cause errors further down the line.
Settle four things before the first import: which keys are permanent, how supplier items and packaging are linked, which values may be replaced automatically and who fixes records that are blocked.
A well-designed PIM data model needs to describe these relationships. Before the import, you should be able to trace an item from supplier reference to orderable variant. Also test what happens when a packaging level is discontinued.
Which metrics show whether enrichment is working
Start enrichment with the fields that make the item possible to order and handle. Set requirements for unit, net and gross weight, dimensions, product class and delivery status. Require a valid image or technical drawing where the product and channel need one.
After that you can add marketing copy and in-depth content. Measure completeness separately for identification, logistics, sales and compliance. A well-written description should never hide the fact that a required weight is missing.
Identification means the product has valid identities. Logistics means units, weight and dimensions are usable. Sales means the customer gets the right content and range, and compliance means the required documents have been reviewed.
According to SCB's overview of ICT usage in enterprises, 30 per cent of Swedish companies received orders via their own website or app. Among companies selling online, e-commerce accounted for 13.5 per cent of turnover for small companies, 19.5 per cent for medium-sized companies and 32.2 per cent for large ones. For your own follow-up, you need to measure what happens between item registration and a placed order.
Language, customer segment and range status should control publishing. A discontinued item may need to stay as a spare parts reference but be blocked from purchase. A customer-specific range has to follow the customer's permissions all the way to the order.
Track four metrics. The share of items that pass validation shows which ones can be published without manual correction, and it should be tracked per channel. The time from a new item in the ERP system to publication shows how quickly products become sellable. The share of manual order corrections shows how often incorrect data reaches the order, and a rising share needs investigating. Channel discrepancies per item show differences between the channel's data and the agreed source, so recurring errors and their causes can be followed up.
A new pack size can affect price, shipping calculation and order quantity at the same time. That is why you should test the whole purchase whenever packaging rules change. Page views say nothing about whether the customer could order the right quantity, while order corrections can show where the chain breaks.
Where AI can enrich product data and where people approve
Decide which fields AI may suggest values for before you switch on enrichment. Also decide which suggestions have to be reviewed by a person. Permissions need to follow how much the field matters for purchasing, use and safety.
According to SCB's statistical news on ICT usage in 2025, AI use among companies with at least ten employees rose from 25.2 per cent in 2024 to 35.0 per cent in 2025, while 72 per cent of AI users reported one or more specific purposes. For product information, that purpose needs to be tied to specific fields and approvals.

AI can suggest normalised attribute names, channel copy and translations based on approved source material. Price, stock, chemical classification, technical tolerances and item relationships need rule-based validation and approval by a named person. For price and stock, the review should cover the rules and the source, while the current values come from the ERP system.
Say a supplier document contains a technical tolerance with an unclear unit. AI should send the value for review. Otherwise a plausibly worded sentence can give the customer the wrong information about the product's properties.
Log the field and its previous value, so the change can be understood. Log the source material used, so the basis for the suggestion can be checked. Store both the suggested and the approved value to separate suggestions from decisions, and record the approver and time so you can trace who made the decision.
More automation can also spread the same error across the whole range. So limit which records can be changed together and review a defined sample before publishing widely. Make sure approved values can be restored. An AI suggestion should keep its status as a suggestion until the right check has been done.
Split your data model by which fields can be processed automatically and which require review. Add fields that AI may never change. That gives product managers and system administrators clear boundaries to work within.
When PIM isn't the first step in digitalisation
A limited range with consolidated basic data and one sales channel can work well without a separate PIM. A cleaned-up item register with clear responsibilities may then be a better first step. Start by checking which recurring problems you actually need to solve.
If the errors are mainly duplicates and mixed units, you need to fix the source. Otherwise a new system means handling the same errors in one more place. The cost of maintaining the register remains after implementation.
If the basic data is already consolidated and used in one channel, the first step is to clean up the register and appoint owners. If supplier files lack a common structure, you need to decide on an import template and intake check. If several people change the same field, you need to set mandates and an approval order. If the same content is processed for several channels, it makes sense to trial a limited PIM flow.
PIM takes time from the business. Someone has to remove duplicates, sort out packaging and decide how conflicting supplier data should be handled. Purchasing needs to set requirements for the source material and product managers need to review it.
When purchasing, marketing and IT share the responsibility, one person needs to own the whole. That person has to be able to prioritise corrections and decide when an item can be published. The role needs a mandate even when departments want different things. You also need to contribute the product knowledge that decides whether the data is correct.
At Glanser we assemble proven building blocks and keep new development to a minimum. You can start with a limited solution and build on the same foundation. A ready-made setup is a poorer fit, though, when the workflows you depend on require major deviations.
A cleaned-up spreadsheet can be enough during preparation if changes are few and easy to control. In that case, decide who may edit, how versions are saved and how approved data is passed on. Move the work to PIM when more channels or reviewers make that process hard to sustain.
Three questions to ask your next vendor
How do you handle stable keys when the same item exists at several packaging levels or with several suppliers? Ask the vendor to show a product with single unit, carton and separate supplier references. The answer should separate the standardised product identity from your internal keys and show what happens during an update. Also ask for a demonstration where the vendor changes a supplier reference while you watch. You will see whether already published variants are left intact and whether the vendor understands the product's relationships through the whole order flow. A concrete answer can be followed from the changed field to the receiving channel.
Which validation rules block publishing, and who approves fields that AI has suggested? Ask for an example where an incorrect value is stopped and sent to the right person. You should see the rule itself, the approval and the change log. That tells you whether the check exists in the workflow or still has to be built, and who on your side will maintain the rules as the range changes.
How do you follow up channel discrepancies, and what happens to PIM when the ERP system is upgraded to a new version? Listen for a description of comparisons between systems, error handling and responsibility for testing after upgrades, because the answer shows which responsibilities are contracted after go-live. A weak answer stops at saying the systems are kept in sync, without naming who tests the connection or who pays when an upgrade breaks it.
Three common answers need follow-up questions. If the vendor says data syncs automatically, ask how errors are detected and who acts on them. If the vendor says AI output is reviewed before publishing, ask which fields are reviewed and by whom. If the vendor says upgrades are handled on an ongoing basis, ask what the agreement covers and who does the testing.
Start with ownership before you compare PIM systems
Put your spreadsheets and supplier material next to the item register. Mark which data has an owner and a validation rule. Write a name where you would otherwise have written a department.
Bring the mapping to the vendor meeting. A missing weight, a conflicting material value and an unclear carton relationship lead to better discussions than a generic feature list. The examples also show what work your organisation needs to do before implementation.
Every mark in the mapping becomes a requirement. If an owner is missing, the role has to be decided on your side. If the same field is changed in several systems, a governing source has to be set. If an approval is missing, a reviewer has to be appointed. If the channel shows outdated content, transfer and error handling need to be described.
The PIM page describes how product information is gathered in a structured system. There it is managed and enriched once and distributed to websites, customer portals and reseller channels, connected to the ERP system. Comparing it with your mapping helps you spot requirements that are still missing.
The gaps can concern both functions and responsibilities on your side. An empty owner field needs a responsible person before system permissions are decided. Use the description of PIM as a basis when you assess how your own flow should work.
We help you structure product information and connect PIM to the ERP system and your sales channels. In a 72-hour analysis you set aside two hours, and within 72 hours you receive an inventory of functions and integrations, a function specification, a prototype and a complete cost overview 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 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
Ordering portals: how they work and what to demand
An ordering portal lets customers see their contract prices and stock levels and place orders themselves. How the order flow reaches the ERP system, when EDI is the better fit and what to demand from your vendor.
Kenneth Lyngshede
Master data management that governs your shared data
One source and one owner per attribute keep your systems giving the same answer.
Peter Nilsson