What Majid Al Futtaim’s 9% AI Shopping Figure Measures

A product suggestion can influence a purchase without an AI agent ever pressing the buy button. That distinction matters after Majid Al Futtaim’s chief executive put a striking figure on the rise of AI-assisted retail.

At the Sharjah Investment Forum on 15 October 2026, Ahmed Galal Ismail said AI agents now play a role in around 9% of the group’s customer transactions, through recommendations, influencing choices or directing purchases. Emirates News Agency reported the statement. It is a company executive’s estimate, not a separately published audit. The group has not disclosed the calculation, the markets included or how it distinguishes an AI recommendation from an autonomous purchase.

For UAE shoppers, the practical question is whether a recommendation is genuinely useful, paid for, or based on personal data. For merchants, the question is whether AI changes completed purchases and customer outcomes rather than merely collecting credit for transactions that would have happened anyway.

What the 9% does and does not measure

The language reported from the forum is unusually broad. AI can recommend a product, influence a shopper’s choice or direct someone toward a purchase. Those are different levels of involvement. A recommendation displayed on a retail app is not the same as a software agent selecting an item, entering delivery details and authorising payment.

Nor does the statement establish that 9% of UAE purchases are AI-driven. Majid Al Futtaim operates across several countries and businesses, including grocery retail, malls, entertainment and lifestyle. The reported figure refers to the group’s customer transactions, without a published UAE-only breakdown.

There are other missing denominators. Is the 9% a share of transaction count or sales value? Does it cover digital and physical purchases? How long after a recommendation can a sale still count as AI-influenced? Does the company count any recommendation exposure or require a click, referral or measurable interaction? The public account does not answer those questions.

That does not make the number meaningless. It makes it a useful directional claim whose precision should not be overstated. Until the method is disclosed, Robius would not translate it into “one in eleven UAE purchases completed by an AI agent”.

There is a measurable business behind the claim

The stronger evidence comes from the company’s previously published operating disclosures. In April, Majid Al Futtaim and Visa said AI-led personalisation within the SHARE loyalty ecosystem had increased click-through rates for personalised offers by 23%. That is a specific engagement metric. It is not a 23% increase in purchases or proof of incremental revenue.

In its September first-half results, the group reported roughly 14 million SHARE members across the region, with more than 190,000 customers transacting daily. Retail digital revenue grew 11% to AED1.8 billion in the first half of 2026. Precision Media, its retail advertising business, reported an 89% year-on-year revenue increase to AED75 million. These are company-reported figures, and each measures a different activity.

Read together, they show why AI recommendations have become commercially important. A retailer with loyalty accounts, transaction records, apps, online orders and in-store advertising has multiple opportunities to shape product discovery. But growth in advertising revenue, digital sales and loyalty engagement cannot by itself establish that AI caused a particular purchase.

For investors, the useful question is whether these tools improve gross margin, repeat purchasing and customer retention after technology and advertising costs, not simply whether more interactions can be labelled AI-assisted.

Recommendation and advertising are not interchangeable

One layer is product relevance. A grocery app might suggest a replacement item or show a frequently purchased category. Another is sponsored visibility: a brand pays to place its product in front of shoppers. Both may use machine learning, but the shopper deserves to know which type of recommendation is being shown.

Majid Al Futtaim’s own 2025 announcements describe how its Precision Media operation links digital and physical retail advertising. The company said its partnership with Publicis Media included 212 digital screens across 25 Carrefour hypermarkets in the UAE, with sensor-supported audience engagement. A separate partnership with Advertima described connecting in-store audience signals with online shopping profiles for more targeted campaigns.

Those announcements establish the direction of the business, not the behaviour of every screen or the identity of any particular shopper. They also make the consumer question concrete: when a recommendation appears, is it chosen for usefulness, because a supplier bought placement, or through a combination of both?

A retailer can disclose commercial placements clearly while still offering useful recommendations. Problems arise when paid influence is indistinguishable from neutral advice, or when customers cannot understand which data categories are used to shape their shopping journey.

Online growth does not erase the physical store

At the forum, Ismail also said around 25% of the group’s retail business is conducted online. That is a separate claim from the 9% AI-influence figure, and the public account does not define whether the online share is measured by transactions, sales or another basis.

It would be a mistake to assume AI matters only in an app. A shopper can see an offer on a phone, compare it later, visit a supermarket and complete the purchase at a till. Conversely, someone may discover an item on a store shelf and reorder it online. Retail attribution becomes harder when the same customer crosses channels.

Majid Al Futtaim’s full-year 2025 results reported 20% e-commerce revenue growth and 38% growth in quick commerce. Its 2026 first-half statement described weaker overall retail revenue alongside rising digital revenue. The contrast matters: faster online growth can coexist with pressure elsewhere in the business. It is not evidence that AI has transformed the profitability of the entire retail portfolio.

For UAE retailers competing on delivery speed, stock availability and price, the operational basics still decide whether an AI recommendation becomes a satisfactory order. A model cannot rescue a checkout that quotes the wrong availability or a delivery promise that cannot be met.

An assistant is not necessarily an authorised buyer

The term agent can imply software with permission to act, not merely suggest. Robius previously explained how AI-assisted shopping infrastructure is being developed for merchants and payment networks. The crucial boundary is the transition from discovery to an actual financial commitment.

Another Robius analysis of the Amazon and Perplexity shopping-agent dispute showed that user permission, platform access and payment authority are separate questions. A shopper may want an agent to compare prices without granting it the right to place an order.

Nothing in the 15 October statement establishes that Majid Al Futtaim has deployed autonomous checkout across its stores or that 9% of transactions are executed without a human. The reported wording covers recommendation and influence as well as purchase direction. Those distinctions should survive the headline.

A shopper’s five-point check

First, identify the recommendation. Is it an organic result, a loyalty offer, a sponsored placement or an answer from a third-party assistant? Look for available labels and do not assume that the top item is the cheapest or best match.

Second, compare the final price and conditions. An AI suggestion is only a starting point. Check package size, substitute items, delivery fees, timing, return eligibility and any subscription or minimum-spend condition before approving payment.

Third, separate personalisation from permission. Loyalty programmes and shopping apps may have different settings for marketing, data sharing and account access. Robius’s AI app permission audit explains why granting access to device data is a separate decision from accepting a recommendation. Review the actual settings and terms rather than assuming every AI-labelled feature uses the same data.

Fourth, retain the human approval point for purchases that matter. Review the merchant, seller, quantity, total and delivery address. Do not give an unfamiliar browser agent unrestricted access to saved cards or shopping accounts simply because it promises convenience.

Fifth, keep the receipt and know the resolution route. If a recommendation leads to the wrong item or a disputed charge, the retailer’s returns and customer-service process matters more than whether the suggestion was generated by AI.

What UAE merchants should measure instead

Retailers should resist treating “AI influenced” as a single success metric. A useful measurement framework has at least four layers: recommendation exposure, customer engagement, completed transactions and incremental business outcome. Each layer answers a different question.

Exposure asks how many eligible shoppers saw a suggestion. Engagement asks whether they interacted with it. Conversion asks whether an order followed. Incrementality asks whether the recommendation caused more valuable purchasing than a comparable group would otherwise have made. Without a credible comparison group, a system can take credit for habitual purchases.

Merchants should also separate paid placements from product advice and report outcomes after returns, cancellations and discounts. If the recommendation boosts basket size but increases substitutions or complaints, the headline conversion rate hides the cost. If the tool steers customers toward sponsored products, measurement should distinguish advertiser performance from shopper benefit.

For small UAE businesses, the governance challenge is similar to the one in Robius’s practical policy for workplace AI use: assign ownership, define what data tools can access, and keep accountable humans around consequential actions. An AI-driven storefront needs those rules even when it never calls itself an agent.

What would make the next 9% claim stronger?

The next disclosure should define the reporting period, transaction denominator and geographic scope. It should explain what counts as an AI-influenced transaction, how paid recommendations are treated, whether the share is based on orders or revenue, and whether independent testing or holdout groups support a causal claim.

Consumer measures matter too: price comparison quality, recommendation transparency, opt-out controls, dispute outcomes and satisfaction after delivery. Investors should watch whether AI and Precision Media contribute sustainable earnings rather than only more impressions and attributed clicks.

The 9% figure is worth covering because it comes from a major UAE-headquartered retail operator with a real digital and loyalty business. It is not yet proof that autonomous agents are buying one in eleven baskets. For now, the clearest conclusion is that AI is entering the path to purchase, while the public evidence still leaves the size of its independent effect unresolved.

Sources

• Emirates News Agency, 15 October 2026: CEO statement at Sharjah Investment Forum
• Majid Al Futtaim and Visa, 30 April 2026: digital commerce and SHARE personalisation
• Majid Al Futtaim, 9 September 2026: first-half operating results
• Majid Al Futtaim, 16 March 2026: full-year 2025 digital retail results
• Majid Al Futtaim, 2 June 2025: Precision Media and Publicis retail screens
• Majid Al Futtaim, 13 May 2025: Advertima in-store audience technology

Checked 15 October 2026. The 9% figure is attributed to the chief executive’s public statement; its methodology, UAE-only share and incremental sales effect were not disclosed in the sources reviewed.

Robius.news | Dubai, UAE | 2026 | Built to be first. Built to be trusted.