What PIM is and why it determines the quality of your product information

The same item can exist in four places in your organisation and still be missing when someone needs it. Dimensions sit in the business system, an image with marketing and an older description on the website. When a reseller asks for up-to-date information, someone starts searching, comparing and forwarding files.

That’s usually when the question of what PIM is becomes concrete. PIM stands for Product Information Management and is about gathering, structuring, quality-assuring and distributing product information from a controlled source. The question is who owns the information, how changes are approved and how the same product can be shown correctly in several channels.

What product information typically looks like today

A product manager changes a technical detail in the business system in the morning. In the afternoon, a salesperson is still using an older PDF with the previous value. Meanwhile, marketing has published a third version on the product website, because no one knew which file was the latest approved one.

None of these people has necessarily done anything wrong. Each department worked with the information available. The problem arises when item numbers, descriptions, dimensions, images, documents and translations are maintained separately. A change then has no obvious route to the web shop, the customer portal, the reseller feed and the sales material.

The consequences show up in everyday situations. A customer portal’s filters rely on a dimension that is missing for part of the range. A reseller receives an image that belongs to a different variant. A new market can’t publish the product because the Swedish text exists but the approved translation is missing.

The more channels and recipients use the same product, the harder it becomes to keep the versions in step manually.

As the range grows, every manual copy becomes a potential source of error. The work also takes time away from enrichment, since staff first have to work out which version applies. A review of how product data is handled in a product data management system often shows that the problem lies in the structure. More files and more routines rarely fix an information model where ownership is unclear.

That is the starting point to keep in mind. PIM becomes relevant when you want to move away from parallel versions and create a way of working where an approved change can be used everywhere.

What PIM actually is

PIM is a central way of working and system support for product information intended for use across several channels. Its technical core is a shared information model. Item numbers, attributes, units of measure, images, documents, languages, markets and channel rules are managed there as structured data points with a clear meaning.

The model starts with your product families. Each family needs an attribute matrix that sets out which fields are required for that type of product. One family may require weight, length, width, height, material and safety documents. Another may require power, voltage, IP rating and installation instructions.

Mandatory fields need to be combined with validation rules. Weight can be required in grams and dimensions in millimetres. The check can stop publishing when a technical specification, an image or an approved language version is missing. The rules make gaps visible before the customer or reseller encounters them.

Ownership and approval are the next part. Each field should have a responsible role, a status and a version history. That way you can see who changed a value, when the change was made and who approved it. Traceability is especially important for information that affects safety, compatibility or the choice of spare part.

Distribution is then governed by channel rules. An open product website can show a short description and selected documents. A customer portal may need more detailed specifications, while a reseller feed requires a specific set of attributes. The base information is maintained once and adapted according to each recipient’s rules.

The difference between PIM, ERP and e-commerce

Many discussions stall because three systems have overlapping responsibilities. The business system should own item master data, prices, stock and orders. Business Central, Monitor and Jeeves are examples of systems where transactions belong. They determine what can be ordered, at what price and with what availability.

The e-commerce platform is responsible for the encounter with the customer. It handles presentation, categorisation on the website, the cart, checkout and the order flow. The platform needs reliable product information, but it is rarely the right place to maintain every language version, document and channel adaptation.

The PIM is responsible for sellable, channel-governed product information. This is where descriptions, images, technical attributes, documents and translations are enriched. It is also where you decide which information is required for a given market or channel before a product may be published.

The boundary needs to be defined field by field. The item number can be created in the business system and imported into the PIM. Marketing copy can be created and approved in the PIM and then published on the website. Prices should normally be fetched from the business system at display or ordering, since customer agreements and price lists can change independently of product content.

Digital commerce covers more flows than an open web shop. According to Statistics Sweden’s definition of e-commerce, it includes orders via a website or app and orders through EDI messages designed to receive orders. Product data therefore needs to combine with customer-specific ranges, prices, stock status and order rules in several environments.

When responsibility is clear, changes become easier to assess. You can see which system needs changing and which channels are affected. A deeper look at ERP integration in practice helps you review that boundary before deciding on scope. Unclear boundaries often lead to double maintenance and manual checks that no one really owns.

How PIM connects to websites, customer portals and ERP

A working flow usually starts when the item is registered in the business system. The item number and other base data are passed to the PIM through a defined integration. There, the item is linked to the right product family and gets the attributes that family requires.

Enrichment then happens in the PIM. Product managers complete specifications, marketing reviews copy and those responsible for each market approve language versions. Images and documents are linked to the right item or variant. Validation shows what remains before the product can be published.

Once the item is approved, it is distributed according to the channel rules. Litium can receive information for the product website and e-commerce. A customer portal can show the same base data with documents and ranges adapted for logged-in customers. Resellers can receive a feed with the fields and formats their systems require.

The same product may therefore need different presentations without the base information being copied. The open website might show the area of use and main dimensions. The portal can add spare parts, manuals and the customer’s own product names. The reseller feed may require GTIN, packaging details and specific category values.

The flow also needs to work when things change. When an attribute is corrected in the PIM, the approved version should reach the affected channels through the integrations. If a value is owned by the business system, the change should be made there and then passed on. That order reduces the risk of channels gradually developing their own versions.

Your products are also increasingly compared in environments where language, units and attribute requirements differ.

At Glanser, such connections are built with functions and integrations from the Function Library, with a defined scope before development. A digital customer portal with structured product information then becomes a publishing surface that uses the same approved base data as the website.

PIM as a prerequisite for AI to give credible answers

PIM is not an AI platform and shouldn’t be sold as a shortcut to automated content. The quality of product data does, however, determine what an AI function can do with reasonable confidence. Incomplete or contradictory

specifications give the model a weak basis and increase the risk of invented details.

Structured attributes make it possible to separate facts from marketing. A dimension, a material detail or a safety classification can be locked to an approved field. Descriptive copy can get an AI-generated draft, but the draft should be reviewed before publishing. The rules need to state which fields may be suggested, changed or translated with AI support.

Traceability becomes critical when a technical detail is questioned. You need to be able to see the source value, the generated version and the person who approved publication. Otherwise, a wrong formulation can spread to several channels before anyone discovers where it came from.

Swedish companies’ use of AI gives a sober backdrop to that priority. According to Statistics Sweden, 35 percent of Swedish companies with at least ten employees used some form of AI technology in 2025. For many businesses, getting product data in order is therefore a more fundamental first step than text generation.

Three metrics make quality possible to track. The share of complete items shows how much of the range meets the requirements for each product family. The number of validation errors per 1,000 items shows where the model or data entry falls short. The time from registration in the business system to publication shows whether enrichment and approval flow or get stuck.

The metrics give the AI effort a checkpoint. Once the base data is complete and versioned, translation, classification and drafts can be tested in limited areas. When the base data falls short, data quality should be fixed before broader automation.

How a PIM implementation works in practice

An implementation should start with an inventory of product types, data sources and publishing channels. You map which attributes exist, where they are maintained and where values differ between systems. The inventory should also show which copy, images, documents and translations lack a clear owner.

You then model the product families. The attribute matrix sets out mandatory fields, permitted values, units and the responsible role. The model needs to stay simple enough to use every day. An overly detailed model can create heavy data entry and lead staff to work around the system.

Enrichment is often the most labour-intensive part. Copy has to be written or moved, images linked to the right variants and documents checked against the current version. The work becomes manageable when you prioritise product families and channels by business importance. The whole range rarely needs to reach the same level at once.

Integration and publishing come after the model and responsibilities. Data then flows from the business system to the PIM and on to websites, portals and reseller feeds. Testing should cover missing fields, invalid values, language versions and changes to already published items. A successful first import says little about how the flow works after six months.

Differences between industries affect priorities. According to Statistics Sweden’s data on companies’ web sales in 2024, 30 percent of Swedish companies sold via their own website, app or e-commerce platform, while the share was 21 percent in manufacturing and 54 percent in motor vehicle trade and repair. A manufacturer may therefore need to prioritise resellers and customer portals over its own open sales channel.

Ongoing management needs a named owner in the business. New attributes, changed channel rules and new markets require decisions on content and quality. IT can own the integrations, but product managers need to own what the fields mean.

PIM fits less well when the range is very small and only one channel is used. Clear routines in existing systems may then be enough. The assessment should be based on the number of items, variants, languages, recipients and changes in your business.

Three questions to ask the supplier before you decide

The first question can be put like this: what do ownership, approval steps and version history look like when your own staff work in the solution every week? Ask the supplier to show a change from registration to publication, including a stopped error. The answer reveals whether the way of working can be run by you or creates dependency for small adjustments.

The second question is about follow-up. Which data quality metrics does the supplier track after go-live, and who acts when they deteriorate? Concrete answers should cover complete items, validation errors and lead time to publication. The answer reveals whether PIM is treated as ongoing management or left behind after implementation.

The third question is about upgrades. Who is responsible for integrations and functions continuing to work when the platform or business system is upgraded, and is that work part of a contracted responsibility? Ask for a clear description of version tracking, testing, and bug fixing. The answer reveals whether you are buying continued product ownership or a handover where every upgrade requires a new assessment.

The answers should be traceable to named roles and concrete routines. General promises of flexibility say little about everyday life after launch.

Next steps if you want to see what it would look like for you

A sensible start is to test the model against your actual range. Choose a few product families and follow their path from the business system to the website, portal and resellers. You’ll then see which fields are missing, where responsibilities overlap and which integrations are actually needed.

The material should show both the possible scope and the work that remains on your side. The data model, enrichment and approval can’t be fully delegated to a supplier. The scope can, however, be made clear before you decide on implementation, functions and ongoing management.

Alongside PIM, you can see how we build data models, validation, and publishing to the web, portals, and retailers.