Artificial intelligence is the talk of the year, but it is no longer a novelty in the retail industry: for years, intelligent algorithms have been optimising forecasts and replenishment processes, supporting customer service staff and managing personalised dialogues with customers. However, with the advent of generative AI, the topic has gained new momentum.
It is not so much the technology itself that is new, but rather the strategic dimension: how can retail companies embed artificial intelligence effectively, sustainably and holistically into their structures?
From project to practice
The new EHI study "AI Integration in Retail" reveals that almost all major retailers are now engaging with AI at a strategic level. What began as a pilot project a few years ago is now part of concrete transformation projects. Nevertheless, implementation remains selective in many cases: AI is used selectively, but is rarely embedded across the board. However, against the backdrop of the EU AI Act, which formulates new requirements for transparency and traceability, it is becoming clear that successful AI integration is less of an IT project and more of a transformation process that affects all levels.
A survey of AI managers found that 67% consider their data integration as poor or relatively poor, while only 13 percent consider it to be good. Although large amounts of data are available, they are distributed across different systems and are often of insufficient quality. A lack of interfaces and the historical development of IT landscapes further complicate smooth integration. These structural deficits are exacerbated by unclear responsibilities in data management, as in many places it is not clearly defined who is responsible for maintenance, quality assurance and governance. Forecasts and applications remain unreliable if the underlying data is inconsistent or incomplete. Meanwhile, early initiatives such as the establishment of data lakes or central platforms show that targeted investments can be effective – but so far, this has been the exception rather than the rule.
Price management systems
Without clear responsibilities, uniform standards and a well-thought-out data strategy, the potential of AI can only be achieved to a limited extent. The three biggest hurdles are clearly identified: the general availability of data, its technical integration into existing systems and ensuring its quality and usability. AI can only unfold its full potential in the retail industry when these three dimensions are systematically addressed.
Against this backdrop, price management systems are an insightful example of how technology and data strategy need to work together. At EuroShop, providers such as GK Software are demonstrating how precise AI-supported price optimisation works with a reliable data architecture. "The key to data architecture is finding the right balance. It's not necessarily a case of 'the more, the merrier'; it's much more important to create a clean, robust and structured database that enables pricing algorithms to work optimally," explains Sandy Preuß from GK Software. According to the expert, flexible interfaces to PIM, ERP or order management systems, as well as to external sources such as competitive price data, are particularly important in this context. Such integrations prevent manual data silos, increase the consistency of inputs and make automatic decision-making processes reliable.
At the same time, hybrid integration models and the option of keeping sensitive data in European or in secure environments specifically address the concerns raised in the study about loss of control and reliance on hyperscalers. For retail decision-makers, this means that investments in data pipelines, governance and interfaces are not a secondary IT task, but rather a strategic prerequisite for AI applications – such as in pricing – to function reliably and in compliance with legal requirements.
AI has moved into all dimensions of EuroShop
Artificial intelligence now influences almost all thematic dimensions of EuroShop. In particular, the advent of generative AI has seen its use develop far beyond traditional areas such as replenishment, pricing, forecasting and inventory management. We see the merging of technology, data intelligence and operational practice everywhere, from store planning, energy and building control to shelf management, all of which are making retail more intelligent and efficient step by step.
Artificial intelligence is used in the design and simulation of sales rooms, for example. Modern 3D planning tools enable the virtual replication and interactive modification of real shops: shelving systems, displays and lighting elements can be inserted into the floor plan using drag-and-drop and are immediately available as photorealistic 3D visualisations. This allows customer walkways, sightlines and lighting moods to be optimised in advance. For example, Umdasch Shopfitting provides shop.up, a 3D planning programme that retailers can use to display their store concept in real time in the real environment with the help of augmented reality. They can also use it to try out different lighting scenarios on test shelves, for example. Thanks to AI-supported design tools such as these, design cycles can be significantly shortened and shop layouts can be adapted flexibly.
Key findings of the “AI Integration in Retail” study
- Responsibilities are established: 100 percent of the companies surveyed have defined clear responsibilities for AI within the company, and 71 percent have specialist teams.
- Early adopters remain the exception: 16 percent of respondents see themselves as active pioneers.
- Data integration remains a weak point: 67 percent rate their current data integration as poor or relatively poor.
- Sovereignty is coming into focus: 21 percent of respondents prioritise processing sensitive data exclusively on European servers.
- The results also demonstrate that technological progress alone is not enough. Strong leadership, clear governance, targeted training and the active involvement of employees are the key drivers.
