eCommerce, Loyalty, Personalisation, Identity & FraudRetail, FinTech & Agentic Commerce

← All writing
Research11 Nov 2026Harini Ramalingam

The Brand Demand Blindspot

Shoppers now find products through AI assistants and creators. Shelves are still planned on last quarter's sales. The gap between the two is where brands are quietly losing sales.

This research asks one question: is there a real difference between the products shoppers discover, mostly through AI assistants and social media, and the products actually available on the shelf? The short answer is yes, and the gap is widening. The shift in discovery is well evidenced. How large the gap is in GCC stores still needs to be measured directly.

Today a product can be picked by an AI answer or a creator video and see demand jump within days. The shelf it needs to be on was set months earlier, by a category manager working from past sales. The brand sits between the two and sees neither clearly. This is the brand demand blindspot. The sections below cover how discovery has changed, why existing tools miss the gap, why the timing is right, the 2027 outlook, and what brands and retailers can do to close the gap.

The gap in four numbers

Gulf shoppers moved to AI and social fast, and the shelf has not kept up
70%of UAE consumers have used an AI assistant to shop (EMEA average: 34%)Adyen 2025
73%of UAE and KSA consumers bought through social media in the past yearDeloitte 2025
~4products in a typical AI shopping answer, from 134,271 active FMCG SKUs in the UAEcloro 2026 · NielsenIQ 2025
68%of shoppers check 2 to 4+ stores before finding a trending product in stock (US)RELEX 2025

The two blue figures show how fast Gulf shoppers have moved to AI and social discovery. The two amber figures show the mismatch: an AI answer narrows a huge shelf to a handful of products, and when one of them trends, most shoppers struggle to find it.

The gap: discovery and the shelf run on different signals

Discovery is now decided by algorithms and creators, one SKU at a time, in days. The shelf is decided by category managers and planograms, using past sales and trade terms, over months. The brand sits between the two and sees neither clearly.

To see why this gap is widening now, start with where shoppers begin their search.

Discovery has moved, and the Gulf is ahead

Around the world, search is slowly losing its place as the first stop. Social has caught up with search among younger shoppers, and AI assistants are the newest starting point. Globally, AI is widely tried but rarely first: only about 7% of US shoppers start a product search on ChatGPT.

How shoppers start looking: global vs GCC
Global and USGCC
SearchStill leads50%+of online adults worldwide research purchases on search engines GWI 2025Level with social53–58%of UAE and Saudi internet users research brands on search engines We Are Social
SocialCatching up#2research channel overall; first only for younger shoppers and in South-East Asia GWI 2025Leads60%of Saudi internet users research brands on social; 73% of UAE and KSA consumers bought through it We Are Social 2026 · Deloitte
AI assistantsTried, rarely first34%of EMEA consumers have used AI to shop; only about 7% of US shoppers start a product search on ChatGPT Adyen · Smarty 20262× faster70%of UAE consumers have used AI to shop; Saudi AI use rose from 49% to 66% in a year Adyen · Deloitte KSA 2026

The GCC is further along the same path:

  • Social already leads. 60% of Saudi internet users research brands on social media, ahead of search.
  • AI is growing twice as fast. 70% of UAE consumers have used AI to shop, and Saudi AI use rose from 49% to 66% in a single year.

This changes who decides what a shopper sees first. When search led, a brand could buy a listing and follow the click. Neither a creator feed nor an AI assistant knows what a nearby store has on its shelf, and neither tells the brand which SKUs it recommended. Paid search may also deliver less. Google has said AI Mode earns less from ads than classic search, so a growing share of questions may end without a sponsored click a brand can track to a retailer. The GCC is not an exception here. It is simply ahead of a global shift, so the lessons learned here will travel.

And when discovery starts in a feed or an AI answer, it runs straight into a shelf that was planned on a completely different system.

Two systems that never talk to each other

What a shopper wants and what a shopper can buy are decided by two different systems. They differ on every dimension that matters.

Where demand startsWhat the shopper can buy
Decided byAI assistants, creators, social feedsRetail category managers and distributors
Based onProduct data, reviews, online contentPast sales, planograms, trade terms
SpeedDays; trends peak and fade within weeksMonthly or quarterly range reviews
UnitA single SKUA category and a shelf position
Brand sees it throughMarketing tools, if at allRetailer and distributor reports, weeks later
Where the sale is lost, from discovery to an empty shelf
DiscoverAn AI answer or creator video picks a SKUAssistants weigh dozens of retailers and up to 290 brands for one question.
SpikeDemand jumps for one itemFeta sales rose 200%+ in days in one TikTok trend. NeeDoh sold a year's stock in nine weeks.
ShelfThe range was set months earlier45% of retailers and brands take over a week to react to a viral spike.
SearchThe shopper can't find it68% check 2 to 4+ stores for a trending item. Only 12% find it on first visit.
OutcomeThe sale goes elsewhereOr to a rival brand, or doesn't happen at all.

Each box is one step from discovery to purchase. The red box is where the brand loses the sale, while shelf space stays with SKUs that discovery has already moved past.

What shoppers do when the product isn't on the shelf

Share of shoppers. Global out-of-stock study (2002) and RELEX US survey (2025)

Buy at another store31%
Switch to another brand26%
Give up on a trending item19%
Don't buy at all9%

These figures are older or from outside the GCC, so read them as a guide to scale.

When a discovered product is missing, the brand keeps the sale only if the shopper finds it in another store. It loses the sale to a rival brand or to no purchase about a third of the time (26% plus 9%). That makes the gap a brand problem first. No public study yet measures these rates for GCC shoppers who arrive from an AI answer or a creator video.

Why nobody has closed it

Today's tools each see one side of the gap. Marketing tools can show whether a product appears in AI answers or social feeds, but not whether it is on the shelf. Retail and sales data show what is on the shelf, but not why demand moved. Retailers' own planning tools react faster to spikes, but only inside their own stores. Nothing joins the two sides product by product, store by store.

None of the reasons incumbents haven't closed the gap is technical. All four are about incentives and structure:

  • Different buyers. AI and social tools are sold to the CMO. Shelf and sell-out data is sold to sales and supply chain. No one is paid to connect them.
  • The retailer owns the shelf view. Retailers earn money from their shelf data through insight deals and retail media, and share it one banner at a time.
  • Market structure. The GCC runs on distributors, baqalas and quick commerce, in Arabic and English. Portal-based US tools don't fit.
  • Wrong unit and wrong clock. Marketing measures brands weekly. Supply chain measures categories monthly. Discovery moves one SKU at a time, in days.

That makes this as much a leadership problem as a data problem. Closing the gap needs a commercial, marketing and supply chain team to agree on one shared view of demand, and on who acts when it moves.

Why now, and the 2027 outlook

AI shopping is moving from experiment to infrastructure. Platforms are now building product feeds, carts and stock tracking into their assistants. Each launch puts more of the shopper's first decision inside an AI answer, and the gap between what it recommends and what a store carries grows with it.

WhenWhat changed
Oct 2026Google launches Universal Cart in AI Mode and Gemini (US), tracking price and online stock across retailers
Oct 2026Google names AI Mode's lower ad revenue as a drag on Search growth
May 2026OpenAI launches product-feed shopping ads in ChatGPT, up to 1M SKUs per advertiser
2024Quick retail reaches about 25% of GCC quick-commerce GMV, up from under 5% in 2019, raising shopper expectations of instant availability

The forward signals all point the same way. Adobe expects AI referrals to US retail sites to grow 130% over the 2026 holidays, after 693% in 2025. Looking further out, McKinsey projects that retail sales made through AI agents could reach $3 to 5 trillion globally by 2030. Yet only 36% of consumer products companies say their product data is fully machine-readable.

AI shopping is still growing fast, just less explosively than before, which is what a maturing market looks like. Two developments could narrow the gap: AI assistants that check live store stock, and retailers using demand-sensing tools to react faster to spikes. But both help platforms and retailers see their own stock. Neither gives a brand a view across every shelf its products sit on, so the blindspot is likely to stay open through 2027.

Closing the gap

The blindspot won't close on its own, and no single tool will close it. It needs brands and retailers to read demand the same way, at the same speed.

  • Brands need to know which of their products AI assistants and creators are recommending, and check those against what is actually on shelf, store by store. Clean, machine-readable product data helps make sure what an AI says about a product matches what is stocked.
  • Retailers can treat AI and social demand as an input to ranging decisions rather than marketing noise, and shorten the time between a spike and a shelf response.
  • Both need one shared view of demand, and clear ownership of who acts when it moves.

How brands and retailers can build that shared view, and what it takes to act on it in the GCC, is the subject of the next article in this series.

Sources

US and global benchmarks are used where no GCC figure exists. Out-of-stock behaviour figures are older or non-GCC and are a guide to scale only.

← All writing