AI in Spanish retail: what five leading retailers already do, and where reviews fit
By Daniel Pisano, Partner & CGO at Vurdere
Five of Spain's largest retailers already run AI in live operations: customer service, logistics, catalogue management and conversational search. The study that brings these cases together names agentic commerce as the next step. Getting there takes reliable product data, and the richest source of that data is what customers write: reviews and questions.
What did the La Distribución Inteligente study show?
On Thursday, 29 October 2026, La Distribución Anged, which represents Spain's large retail groups, and the digital industry association DigitalES presented La Distribución Inteligente in Madrid. The study collects the experience of Alcampo, Eroski, MediaMarkt, Leroy Merlin and El Corte Inglés and sets out a roadmap for tying AI spending to business results.
The roadmap is refreshingly sober. Start from a concrete need, measure, and only scale a project once there is evidence of value. The returns it lists are practical too: product availability, customer satisfaction, and staff time freed up for serving people.
I read the full report looking for one thing: where customer-generated content shows up. The answer says a lot about what comes next for European retail.
What is each retailer doing with AI?
Retailer | Where AI is applied | What has been disclosed |
Alcampo | Online grocery, order picking and logistics, through its partnership with Ocado | Gains in productivity, availability and scalability, with no published figures |
Eroski | Customer service: voice, email and messaging on a single platform | More than 23,000 cases a month, about 42% less handling time per interaction, transfers between agents down from close to 40% to close to 4% |
Leroy Merlin | Continuous improvement of its digital catalogue (e-merchandising) | A change in operating model, with results described qualitatively |
MediaMarkt | Conversational search, an after-sales assistant and a store staff assistant | Search that answers with specs, comparisons, stock, reviews and recommendations |
El Corte Inglés | Quick answers during the purchase and cross-sell recommendations | Brings its sales staff's expertise into digital channels |
Three details stood out. MediaMarkt's after-sales assistant targets purchases above 500 euros, exactly where shoppers hesitate most. Leroy Merlin's programme covers a catalogue of tens of thousands of products and uses an engine that flags where a product page is incomplete. And Eroski's service heat map combines contact-centre interactions with customers' online reviews, cutting the time to spot problems by around 30%.
Why does conversational search depend on reviews?
Picture a shopper asking MediaMarkt's site: "Which robot vacuum handles rugs and dog hair?" The spec sheet lists suction power. The people who can say whether it copes with dog hair are dog owners who bought it and wrote about it.
That is why MediaMarkt's search returns reviews alongside specs and stock. Conversational AI can only answer well what someone has already answered. With three reviews, it has little to say. With 300 recent reviews from verified buyers, it can summarise what matters for each kind of use.
The same logic applies to El Corte Inglés. "Quick answers during the purchase" are, in practice, the questions shoppers have been leaving on product pages for years: will it fit, is it compatible with my model, can it go in the washing machine. If those questions are already answered and organised, the assistant has somewhere to get its answer.
Where does the customer's voice appear in the report?
Rarely, and at the edges. Leroy Merlin lists voice of the customer as one of seven data layers feeding its catalogue programme, next to PIM, taxonomy and web analytics. Eroski uses online reviews to find service problems. At MediaMarkt, reviews are an input to search.
The report also includes what I consider its most important warning: only a small share of the information available is actually used by AI. The bottleneck is not the model. It is the data. And much of the most useful data for a purchase decision is not in the ERP or the PIM. It is in what customers wrote after using the product.
How does Vurdere complement these strategies?
None of these five retailers is a Vurdere customer, and I am not suggesting they are missing anything. My point is that the projects in the study share one thing with our work: they all get better when customer content is plentiful, recent and well organised. If I were building one of these projects, this is how I would connect the pieces:
Conversational search needs review volume. In A/B tests with Vurdere customers, review collection grew up to 10x. In Brazil, our study of more than 120,000 consumers found that review requests sent by WhatsApp convert at up to 15%, against 5% for email.
Quick answers: Vurdere's smart Q&A ranks questions by relevance, giving an assistant a library of real answers.
Cross-selling: cross reviews and the recommendation showcase use what other customers bought and rated to suggest the next product.
A catalogue that improves each cycle: customer content fills gaps the spec sheet leaves open, and Social SEO helps those pages get found.
Marketplaces: for retailers with third-party sellers, seller reviews and the Seller Portal bring the seller's reputation into the decision.
In A/B tests with current customers we measured up to 3x conversion rate, +15% incremental sales, +20% organic traffic and +30% average order value. Installation takes about 3 hours, with one week of validation, on Shopify, VTEX, Magento and other platforms.
What changes when the buyer is an AI agent?
The study closes by naming agentic commerce as the next challenge: assistants that compare options or complete purchases on the shopper's behalf. One line in the report sums up the shift: "Estamos pasando del browsing humano a la compra dirigida por agentes" (we are moving from human browsing to agent-led buying).
An agent does not see shop windows or banners. It reads data. In the agentic commerce protocols I covered in the first article of this series, the aggregate rating and the review count are already product fields. A store with plentiful, recent, machine-readable reviews gives the agent a reason to pick it. A store without them looks like every other option.
The report also covers the other half of the equation: protecting the customer relationship, managing data with clear responsibilities, and keeping humans in the loop. I agree. An authentic review is exactly that: data that comes from the customer, with consent and a name attached.
What can other European retailers copy now?
Pick a measurable problem first, as Eroski and Leroy Merlin did, then pick the tool.
Map where the answers shoppers look for actually live. If they only exist in the heads of store staff, AI cannot reach them.
Grow review volume and recency on high-ticket products, where doubt is most expensive.
Treat product page Q&A as a knowledge base, not a loose comment thread.
Make sure the aggregate rating exists as structured data, for the agents that are coming.
My view
What impressed me most about the study was not the technology. It was the discipline. None of the five tells an "AI does everything" story. Each one went after a specific part of the operation and measured it.
Still, I think the next big gain will not come from logistics or customer service. It will come from the product page. MediaMarkt already sees this, putting reviews inside its search answers. Every retailer should be asking a simple question: when my AI, or the shopper's agent, has to explain why this product suits this person, where will the answer come from? If the answer is "the spec sheet", you are only halfway there.
Frequently asked questions
Which retailers are featured in La Distribución Inteligente?
Alcampo, Eroski, MediaMarkt, Leroy Merlin and El Corte Inglés. The study comes from La Distribución Anged and DigitalES and was presented in Madrid on 29 October 2026.
What are Spanish retailers using AI for today?
Productivity, operations and customer service: logistics and order picking, unified customer service, catalogue improvement, conversational search and cross-sell recommendations.
Why do reviews matter for conversational search?
They answer what spec sheets cannot: real-world use, fit and durability. MediaMarkt's search, for example, returns reviews together with specs and stock.
What does the study say about agentic commerce?
It names it as the next challenge: assistants that compare options or complete purchases for shoppers, with attention to the customer relationship, data management and human oversight.
Comments