PIM in a best-of-breed architecture

In traditional setups, enriching product information using Excel spreadsheets or channel-specific applications was still feasible for relatively small assortments. However, this is no longer a workable solution in today’s landscape. Additionally, the demand for product information has increased significantly—not just for webshops but also for mobile apps, physical stores, point-of-sale (POS) systems, catalogs, flyers, affiliate networks, and more. To support this growth, a new system is needed to manage all product information: the PIM system.

A new system also means new processes and a new application architecture. In this article, we will explore the characteristics and challenges of the traditional architecture and how a PIM system transforms these challenges into an improved product proposition, higher product data quality, and more efficient enrichment processes, with a special emphasis on the role of Artificial Intelligence (AI).

Traditional architecture

The traditional product information architecture is often characterized by the use of Excel spreadsheets, channel-specific applications, and the ERP system as the foundation for enrichment and product information management. Each channel has its own product information requirements, leading to multiple document variations that need to be maintained manually. There is minimal control over content, often resulting in incomplete product information. To compensate, businesses frequently add extra fields in the ERP system—sometimes inappropriately—to manage product data.

In a more mature architecture, basic product information is enriched within the ERP system, while marketing product information is managed in content management systems (CMS) associated with various e-commerce solutions. However, since this product information is also required for other channels, this architecture still relies heavily on numerous disconnected documents containing (fragments of) product data.

Because product information is maintained in different locations, discrepancies arise regarding the "truth." For example, the color of a shirt may be listed as red in a printed catalog but appear as pink in the webshop. This inconsistency negatively impacts data quality and erodes consumer trust. Inconsistent product data leads to increased return rates, reduced visibility in search engines, and lower conversion rates.

Characteristics of Traditional Architecture

• Key characteristics of the traditional product information architecture include:
• (Local) documents containing fragmented product information, enriched separately for each channel.
• Product enrichment occurring in multiple systems/documents (CMS, ERP, Excel).
• No single source of truth (golden record principle) for product information.
• Incomplete product data.
• No automated integrations between systems, requiring manual exports and imports.
• Different data formats for affiliate networks, webshops, and POS systems, requiring manual updates.
• Custom ERP modifications to accommodate additional product information enrichment.
• Marketing information managed within CMS platforms.

The Role of AI in a PIM System

To overcome the challenges of the traditional architecture and establish a mature PIM system, AI can play a crucial role. AI provides advanced capabilities for managing and enriching product data at scale, leading to improved data quality, efficiency, and accuracy. Some key AI-driven enhancements in a PIM system include:

• Automated classification and categorization: AI tools can automatically classify and categorize products based on text and image analysis. This enables quick and accurate placement of products into the correct categories, improving search functionality and navigation across online stores and other channels.
• Automated translations: AI-powered translation software can automatically translate product descriptions and specifications into multiple languages. This facilitates international expansion and allows businesses to engage with customers in different regions.
• Image recognition: AI can analyze and recognize product images, automatically tagging them with relevant attributes and metadata. This reduces the need for manual data entry and minimizes errors.
• Personalization: AI can analyze customer behavior and preferences to generate personalized product recommendations. This helps retailers present more relevant and engaging products, increasing purchase likelihood.

Mature PIM Application Architecture

A mature product information architecture is characterized by a best-of-breed approach, where each system in the application landscape is used for its core strengths. By adopting this strategy, space is created for a PIM application, which takes a central role in the ecosystem. The integration of AI technologies further enhances the potential of the PIM system.

A new product is still created in the ERP system, but it is not enriched there. Instead, this basic product information is automatically exported to the PIM system, where it undergoes further enrichment—supported by AI functionalities such as automated translations and image recognition. Various processes and completeness checks ensure that product data is only exported to sales channels (such as webshops and affiliate networks) when it meets predefined standards.

This approach makes product information enrichment far more manageable. The PIM system serves as the single, complete, and most up-to-date source of truth for all products within the organization. Any necessary updates are made directly in the PIM system, which then automatically distributes the new information to downstream sales channels.

Applicaties in a best-of-breed architecture

While the interaction between different applications leads to the most efficient PIM system setup, integration is not always required to achieve process improvements. Even as a standalone application, a PIM system can deliver significant process optimizations.

Applications that characterize a best-of-breed architecture include:

• ERP (Microsoft AX, Oracle, SAP)
  A software suite that consolidates essential business information across departments. Articles are "born" in this system, receiving the basic information needed for purchasing, warehouse management, etc.
• PIM (inriver, Akeneo, Contentserv)
  A platform dedicated to enriching product data with marketing information to better position items in the market.
• Translation software (SDL, Google Translate)
  A tool for translating product information. Alternatively, translations can be managed directly within the PIM system.
• Feed management tool (Channable, Tradebyte)
  A tool that converts a single data source from the PIM system into multiple feeds for marketplaces and affiliate networks.
• Webshop/CMS (Magento, TableTop, Shopware, SalesForce)
  A platform where enriched products are sold to customers.
• Digital Asset Management (XCDN, Bynder, Ampliance)
  A system for storing, managing, and distributing digital assets (e.g., images). Alternatively, images can be stored directly within the PIM system.
• Print (InDesign, EasyCatalog)
  Software for designing printed and digital publications, directly integrated with the PIM system for real-time product information.

Key characteristics of a mature PIM architecture

• Best-of-breed solution (potentially with AI integration).
• High data quality through the golden record principle (a single source of truth for product data within the organization).
• Automated integrations ensuring timely and accurate data flows.
• Faster time-to-market.
• Efficient processes.
• Reduced manual tasks.
• Lower (license) costs compared to custom solutions or additional personnel.
• Completeness checks to ensure data integrity.

Implementing the PIM Application Architecture

Although a mature architecture follows a best-of-breed strategy, implementing a PIM system is an excellent first step in the product management roadmap. It significantly improves the efficiency of enrichment processes and enhances data quality. PIM serves as the bridge between the ERP system and webshops/affiliate networks.

A logical next step in streamlining product information distribution is adopting a feed management tool. This allows predefined mapping documents to automatically format and distribute product data from the PIM system to various marketplaces and affiliate networks.

For further optimization, consider outsourcing translations to an external agency and storing images and digital assets in a Digital Asset Management (DAM) system. This ensures consistently high translation quality and optimizes image loading speeds on webshops.

With AI integration, a PIM system becomes even more powerful, enhancing efficiency and accuracy in product information management. AI-driven capabilities not only improve data quality but also save time and enhance customer experience. This makes an AI-enhanced PIM system a valuable asset for modern e-commerce businesses competing in the digital marketplace.