E-Commerce PIM and AI: Personalization without Limits

A recent study by McKinsey has underlined this: In e-commerce, customers today expect a personalized customer experience that is consistent across a wide range of channels. In the meantime, retail "multichannel-first" and personalization in e-commerce is mandatory:

• 72% want to be able to shop via every possible channel - and that's almost ten different ones today.
• 72 percent expect a consistent customer experienceadapted to the channel.
• 74% want to be able to see product availability online immediately.
• 67% value digital self-service: shopping at any time and from anywhere.

Responding to these customer expectations pays off very clearly in retail. Compared to non-personalized shopping experiences, personalized shopping experiences achieve a clearly measurable increase in sales of 20 percent.

However, the growing customer demands for consistency and personalization across all channels can only be met if there is a correspondingly comprehensive flow of information and data in the company across many departments and business areas of the company. This is because highly detailed product descriptions are needed to convince customers, reviews and ratings from other buyers, and other materials such as video tutorials can also promote the purchase decision. It is clear: without a good system for Product Information Management, this vast amount of information cannot be managed for comprehensive and consistent personalization in e-commerce across all channels. Omnichannel without a good e-commerce PIM: unthinkable.

PIM systems have always been the central point in e-commerce for bundling and managing all product information. This is because all relevant systems are connected to the PIM, from the ERP to the store system. The PIM stores all product features, logistical data, price information, advertising data, and media assets - and displays them in the required form in all channels. Given the growing demand for consistent personalization in e-commerce - and localization - it is becoming increasingly important to automate processes relating to the e-commerce PIM system. It would be illusory to manually try to meet all the requirements of modern product information management manually. The good news is that PIM is the ideal platform for AI in e-commerce into e-commerce. PIM and artificial intelligence are a perfect combination. Automation via AI is also an excellent opportunity to question, optimize, and redefine existing PIM processes across all departments. This immediately clarifies who supplies which data, where AI functions should provide support, and who releases what.

Large language models like ChatGPT and translation tools based on neural networks such as DeepL are now achieving extraordinary things. They are rightly on everyone's lips. In the e-commerce context, there is a whole range of exciting use cases for this class of generative AI. We show three examples of use cases here.

Use case (bike) for creative text generation in an AI tool: 

Simply recognize an image and immediately know which details and properties are characteristic of the product shown? A modern image recognition service such as Azure Custom Vision from Microsoft can do this. By connecting Custom Vision to a PIM platform, product data can be generated even if only a product image is available. The AI can also classify the product based on the image. It even provides cross-selling ideas for the product based on the company's own product range recorded in the e-commerce PIM. Of course, this also requires seamless integration of the AI service into our PIM. Below are three application examples for AI-supported image recognition and image analysis.

Use case (camera) for creative text generation in an AI tool: 

Moving images are attractive, but localization can be time-consuming. Specialized AI can understand and translate spoken text in videos and even imitate the original speaker. This allows marketing videos to be effortlessly localized into different languages without re-recording or subtitles. Thanks to AI, the geographical personalization and localization of any video asset becomes completely scalable.

A good e-commerce PIM already provides the perfect basis for AI integration. This is because well-maintained and structured product data from your PIM system is the ideal basis for the work of modern AI services, from ChatGPT to Azure Custom Vision. A good PIM platform usually also has a flexible data model so that new fields can be created quickly so that the results provided by an AI can be stored in the PIM in a channel-, country- and language-specific manner. The following also applies: no integration of external AI services into the e-commerce PIM without corresponding interfaces. A PIM platform with robust and flexible interface management is therefore a must. Last but not least: workflows integrated into the PIM make your internal processes easier: from the efficient and largely automated creation of suitable prompts for the AI to smooth correction and approval processes for the results that the AI feeds back into the PIM. 

If you would like to find out what is possible and useful for you in terms of e-commerce PIM and AI and how you can use it for revenue-boosting personalization in e-commerce: We will be happy to advise you.