AI in Retail Media: How Artificial Intelligence Is Changing Advertising in Retail

Personalized advertising in retail is nothing new. What is new is the speed and level of detail with which artificial intelligence can process data and turn it into decisions. Instead of assigning customers only to relatively static audience segments, systems can increasingly take the specific shopping context into account.

Retailers have an important resource for this: first-party data, meaning data collected directly through their own customer interactions and channels—for example, search queries, shopping carts, transactions, loyalty programs, or digital touchpoints. AI systems can identify patterns in this data, build audience segments, and dynamically select advertising content.

The German digital industry association BVDW sees this as a key area of AI application across the retail media value chain. Algorithms can continuously analyze large volumes of data and optimize campaigns based on purchasing and behavioral patterns. Ad placement, frequency, and audience targeting can also be adjusted dynamically.

The difference from traditional audience segmentation lies in the stronger focus on the specific situation. Instead of addressing relatively static groups such as “people between the ages of 30 and 45 who are interested in sports,” systems can incorporate additional factors depending on the available data—such as current product searches, previous purchases, product availability, pricing, the channel being used, or purchase intent.

This is also changing what personalization means. Good retail media advertising does not have to feel as personal as possible. Above all, it has to be relevant to the specific shopping situation.

This will become equally important as product search increasingly takes place through AI assistants and conversational applications. Customers can express complex needs instead of entering individual search terms. Retail media systems will then need to understand not only keywords, but also intent, product relationships, and context.

Retail media is operationally complex. Audiences have to be defined, budgets allocated, bids adjusted, ad creatives produced, campaigns monitored, and results evaluated across multiple platforms. As the number of retail media networks and channels grows, so does the workload.

AI can automate several of these steps. The BVDW describes applications across the entire value chain: forecasting models support campaign planning, generative AI creates different advertising assets, and algorithms manage their delivery. A/B testing can also be automated—for example, by comparing different ads, images, or copy to determine which version performs better. Documented use cases range from AI-powered budget allocation and content creation to analytics and reporting.

This is changing the role of technology. The first generation of automation primarily answered one question: Which task can we complete faster?
The next generation increasingly answers another: What decision should the system make right now?

This can include:

• allocating budgets across campaigns and products,
• determining bid levels and bidding strategies,
• selecting and varying advertising creatives,
• choosing target audiences,
• managing ad frequency,
• forecasting campaign performance and demand.

Mirakl, a provider of e-commerce and retail media technology, expects AI to become more deeply integrated into retail media platforms in 2026. Examples include automated campaigns, AI-powered recommendations for audience selection and bid levels, as well as systems that not only analyze campaigns but can also trigger adjustments.

More automation, however, does not automatically lead to better results. The more decisions a system makes, the more important transparent rules, effective controls, and explainable outcomes become.

One of retail media’s biggest promises is better advertising measurement. Yet this is also an area where major challenges remain.

Compared with many other advertising channels, retail media has a fundamental advantage: retailers can connect ad exposure with search, shopping cart, or transaction data, allowing them to link advertising and purchases within their own ecosystems.

In practice, however, the market is still far from standardized. Retail media networks use different metrics, time frames, and methods to determine how a later purchase should be attributed to one or more advertising interactions. This makes it difficult to compare providers and campaigns.

In a 2025 IAB Europe survey, 53% of respondents cited a lack of standards and 51% cited the large number of different retail media networks as major barriers to further growth. Against this backdrop, transparency, demonstrable results, and reliable measurement are becoming especially important for advertisers choosing retail media partners.

Expectations around measurement are also increasing. It is no longer enough to know that a product was purchased after someone saw an ad. Marketers want to understand whether the advertising actually generated an additional purchase.

That is why incrementality—the additional impact created by an advertising measure—is becoming more important. The key question is whether a purchase happened because of the advertising or would have occurred anyway.

A simple example illustrates the difference: If 1,000 people buy a product after seeing an ad, that number alone does not tell us how many of them would have bought the product anyway. Incrementality attempts to determine what share of those purchases was actually generated by the advertising.

In Skai’s 2026 U.S. survey, 77% of the brands surveyed prioritized better measurement. At the same time, more than half cited limited data analytics resources as a major obstacle to measuring incrementality.

AI can help analyze large datasets, identify patterns, and support optimization based on that data. But as AI becomes more deeply involved in optimization and budget allocation, it becomes increasingly important that advertisers can understand how those decisions are made.

In the long run, a retail media network is not simply selling reach and advertising space. It is also selling proof of impact.

AI is now being used in many areas of retail media. The BVDW identifies applications across the entire value chain—from campaign planning and content creation to ad delivery, analytics, reporting, and optimization.

What initially was often tested in individual pilot projects is increasingly becoming part of day-to-day operations. In 2026, the BVDW is seeing more automated workflows and greater use of AI agents. These systems can connect multiple tasks and independently support or take over individual steps in a workflow.

For these applications to work reliably over time, companies need dependable data, suitable interfaces, and clearly defined responsibilities. Data privacy and governance are also essential—in other words, clear rules governing what AI is allowed to do, how results are reviewed, and who ultimately remains responsible.

These foundations become especially important when companies want to scale AI applications and integrate them into existing processes. At the same time, different retail media platforms do not yet work seamlessly together in every area. Common standards for media buying, measurement, and reporting continue to evolve but have not yet been adopted consistently across the market.

For companies, the challenge is increasingly to turn individual applications into reliable workflows. What matters is not how many AI tools a company uses, but how well those tools fit its existing systems, data, and processes.

Osmos, a retail media technology provider, identifies four developments as particularly important: AI-driven campaign processes, the growing importance of in-store retail media—advertising within physical stores—greater consolidation among retail media networks, and the convergence of retail media and commerce media. The latter extends advertising across additional digital stages of the purchasing journey.

According to the company, the focus of AI is increasingly shifting away from content creation alone and toward campaign management. This includes automated bid optimization, campaign management, and measurement.

At the same time, AI agents could transform product search. These are systems capable of carrying out tasks with a high degree of autonomy. In this context, Mirakl’s outlook on AI and agentic commerce points to shopping assistants that can understand complex search queries and provide relevant product recommendations.

As these systems become more deeply integrated into purchasing processes, retail media could change as well. Advertising may no longer be directed only at people, but increasingly at the digital systems that help prepare product searches and purchasing decisions.

This raises a new question for retail media: How should product data, offers, and advertising information be structured when it is no longer only people, but increasingly AI systems, that determine which products are considered in a specific shopping situation?

For retailers, 2026 is therefore primarily about creating the conditions for broader and more automated use of AI in retail media.

The key principle is Build the infrastructure first, then scale the intelligence.

Retailers looking to use AI effectively in retail media should be able to answer three questions:

• Are first-party and transaction data sufficiently structured and connected?
• Can campaign results be measured across channels using consistent criteria?
• Is it clearly defined which decisions can be automated and where human oversight is required?

Competitive advantage will not come from deploying as many AI tools as possible. What matters is whether retailers can bring data, campaign management, and measurement together within a robust infrastructure. Only then can AI move beyond isolated pilot projects and become a scalable part of the retail media business—delivering more relevant advertising, more efficient campaigns, and more reliable proof of their actual impact.

The use of AI in retail media is part of a broader transformation taking place across the retail industry. The EuroCIS Hot Topic AI & Data-Driven Retail explores how data and artificial intelligence are changing processes, decision-making, and business models