Process innovation through artificial intelligence

In this blog post, we look at how AI process innovations can improve the data quality and content of online shops, what technical requirements are necessary for this and how you can utilise these new opportunities in your own business.

AI-supported maintenance of products and content

A successful online shop is not only characterised by an appealing layout, a well thought-out design and a wide range of functions, but is also measured above all by the quality of its data. Among other things, this forms the basis for search engine and shopping portal findability, which is crucial for the success of an e-commerce business.

Ensuring consistently high data quality used to be an extremely time-consuming process that required a lot of manual work. This included transferring content from manufacturers‘ or wholesalers’ catalogues, continuously maintaining and updating content and linking content to the purchase of a desired item. This process is ongoing and does not end after the initial content creation. Permanent maintenance is essential for long-term high quality.

Small and medium-sized online shops in particular often shy away from this effort, as the available resources are often limited. This is where the use of artificial intelligence (AI) offers an enormous opportunity. AI technologies already make it possible to build and maintain a high-quality pool of articles and content - and at significantly minimised costs. In the following chapters, we will show you how AI is revolutionising this process and what benefits it can bring to your e-commerce business.

Product data: The foundation of a successful webshop

Product data is the centrepiece of every webshop. It comprises the entire catalogue of a company and begins with the division of the article master into a clearly structured category tree. This enables users to navigate quickly and easily through the product range. The various attributes also play a central role, as they describe product characteristics and make them transparent for the customer. These attributes must be filled in a meaningful and standardised way to ensure a consistent and user-friendly presentation.

In addition to this basic data, high-quality descriptive texts are essential. They emphasise the advantages and special features of a product and often provide important information on the intended use. There is also accompanying information such as frequently asked questions (FAQs) or comparative presentations that help the customer to make informed decisions. These elements are not only crucial for good visibility of the shop, but also make a significant contribution to increasing customer satisfaction.

Figure 1 - Forrester Consulting survey shows the problems B2B companies have with data processing

Despite these obvious advantages, many retailers struggle to effectively implement a value-adding product and content strategy. A recent survey conducted by the consulting institute Forrester Consulting in January 2024 shows that almost no company in the B2B sector is able to make its complete product catalogue available online. The main reason for this is often an incomplete or incorrect database, which prevents the effective utilisation of product data.

This finding emphasises how crucial careful and comprehensive data maintenance is for the success of an e-commerce company. Without a solid foundation of product and content data, it will be difficult for retailers to maximise their online presence and meet the expectations of their customers.

Automated product data maintenance with AI

The traditional method of maintaining product data in online shops is based on manually transferring data from various sources into the shop system. A product information management (PIM) system is often used as a central collection point for the data. However, many smaller retailers do without this and maintain the data directly in their e-commerce platform. This process - regardless of the system used - often requires a considerable amount of human labour and is therefore associated with high costs.

This is where artificial intelligence (AI) comes into play, which is predestined for such tasks in particular. AI technologies can significantly reduce the need for human intervention in data maintenance and at the same time increase efficiency and accuracy. At valantic, we use a variety of AI models for automatic processing, including

• Generative text-to-text systems: These models can generate and process texts, e.g. to automatically create product descriptions or improve existing texts.
• Classifying visual and document-based question answering methods: These processes enable visual and textual information to be extracted from documents and converted into structured data.
• Image-to-text systems: Often in combination with traditional OCR (Optical Character Recognition) techniques, these systems extract text information from images, such as product photos, and prepare it for further use.

By skilfully nesting and linking these systems, we can guarantee comprehensive and automated data processing. For example, existing PDF documents can be systematically read, interpreted and applied directly to product master data.

Figure 2 - After automatic data enrichment, valuable data makes it easier for the customer to choose the right product (source: valantic CEC)

Let's assume that only a product photo and a data sheet in PDF format are available for an item: AI systems can extract attributes such as colour, quantity or technical properties such as dimensions, weight and ingredients from these sources fully automatically and almost error-free. After processing, these attributes are immediately available for other e-commerce features, such as filtering search results.

The process runs fully automatically after the initial set-up. Quality checks can be carried out as required to correct any deviations or fine-tune the results. In this way, the use of AI enables efficient, cost-effective and high-quality maintenance of product data, which can prove to be a decisive competitive advantage, especially for small and medium-sized retailers.

High-quality content through AI-supported automation

A successful online shop relies not only on excellent product data, but also on first-class text content. This content plays a crucial role in increasing visibility in search engines, positioning the company as an expert in its field and drawing customers' attention to products and their use. As important as good content is, its creation and long-term maintenance are often associated with considerable effort and high costs.

However, the introduction of advanced AI models for text generation opens up new opportunities to automate this process. It makes sense to utilise these AI results for web shops too. However, this still involves a certain amount of manual effort, as texts alone do not necessarily lead to higher sales. Only the targeted linking of text passages with the appropriate products or categories motivates readers to take the next step and buy the relevant items.

Our valantic solution relies on a combination of three interlinked AI systems to overcome these challenges. These systems generate suitable text modules on demand, automatically link them to the relevant products and integrate them into the page in a visually appealing way. This is done fully automatically and delivers high-quality content without any significant human intervention.

Figure 3 - Automatically created landing page with links to relevant articles in a shop in the DIY sector (source: valantic CEC)

Of course, the automatically generated texts can still be customised manually. This gives companies the freedom to add their own thoughts or change unwanted text passages before the content goes live. The system integrates seamlessly into the administration interface of your shopping portal, making the process of content creation and maintenance not only more efficient but also much simpler.

This technology enables small and medium-sized retailers in particular to provide first-class content with minimal effort, which both improves their own search engine ranking and increases conversion rates in the shop. Automation paves the way for a continuous and cost-efficient content strategy without the need for extensive human resources.

Infrastructure requirements for AI-supported systems

The implementation of the AI-supported processes described above usually requires an expansion of the existing infrastructure. A central component of this expansion is the integration of a vector database, which serves as the knowledge base for various AI processes. Similar to other systems, these databases are often open source and available free of charge, but require specialised expertise for continuous operation and long-term maintenance.

In principle, it is also possible to run AI models locally, i.e. directly on your own servers. This approach offers advantages, particularly with regard to the protection of personal data and intellectual property. At the same time, however, it presents challenges, particularly in terms of performance and processing speed.

Customisation and strategic planning

The necessary adjustments to your own server landscape must always be considered individually. They depend on the existing systems, the desired use cases and ultimately also on the available resources in terms of effort and costs. Careful planning and advice from experts is essential to ensure that the infrastructure meets the requirements and that the benefits of AI technologies can be fully utilised.

Preparing for the use of AI in the e-commerce infrastructure therefore requires not only technical knowledge, but also a clear strategy to achieve the desired results without unnecessarily burdening the existing systems. This is a crucial step in maximising the performance and efficiency of AI applications while ensuring the protection of sensitive data.

Conclusion: AI as a driver for e-commerce innovations

Automated product maintenance using AI and the creation of high-quality content without human intervention are just two of the many use cases that are already becoming a reality in e-commerce. There are also numerous other use cases that have either already proven themselves or are at a promising stage of development.

However, the successful introduction of AI-based features requires the collaboration of specialists from the fields of artificial intelligence, development and hosting. If you would like to examine the opportunities and special features of artificial intelligence in your own project, the experts at MaxCluster and valantic will be happy to help.

About the author

Elias Henrich has been programming since the age of 11 and currently works as a Senior Software Architect at valantic (netz98). In addition to his extensive experience as a Magento 2 developer, his broad expertise helps him design customized system architectures for projects. Since 2019, the trained engineer in medical and information technology has been living on the German North Sea coast, where he also contributes as a university lecturer.

About valantic

valantic is the N°1 for digital transformation and is one of the fastest growing digital consulting, solutions and software companies on the market. More than 500 blue chip customers already rely on valantic - including 33 of 40 DAX companies and a large number of international market leaders. With more than 4,000 digitalisation experts, valantic is represented in 18 countries worldwide and is expected to generate sales of EUR 600 million in 2024(e). valantic advises companies on all the challenges of digital transformation, helps them to better manage their corporate performance and leverage the potential of data and artificial intelligence. In addition, valantic supports its customers in shaping the customer experience, using core digitalisation technologies profitably and optimising company processes across the board.
As one of the leading e-commerce agencies in Germany, netz98 has been part of valantic since 2019. Within valantic, netz98 is the specialist for Magento and Adobe Commerce and has been implementing pioneering B2B and B2C commerce platforms and digitalisation projects for well-known companies for over 20 years.

Published on 29.08.2024 | Process innovation through artificial intelligence | EH