i
News
News · 2026-11-17

AI agents could steer holiday shoppers away from store websites

Dataist Dataist

During the holiday shopping season, an AI agent may do the shopping for you

Cover: AI agents could steer holiday shoppers away from store websites

Shoppers are increasingly asking AI agents to find products, compare reviews and spot discounts without visiting store websites. But an agent can recommend a product only if it can read the page. In an AudioEye study, researchers ran 1,560 AI-agent sessions across six websites. On the site with the most accessibility problems, agents completed 96% of tasks after fixes, compared with just 31% before. Inaccessible pages can cost retailers sales both through AI search and from shoppers with disabilities who face the same barriers.

AI agents could take holiday shoppers away from store websites

What keeps agents from reading pages

Marketing teams are trying to influence which products AI search recommends to users. According to a Pew Research Center survey, nearly half of US adults have used AI chatbots. Adobe Analytics reports that during the 2025 holiday season, traffic to US retail websites from generative AI sources rose 693% year over year.

Before an agent can recommend a product, it has to read the page. Many agents, including those in the AudioEye study, use an accessibility tree—a hidden map of a page’s contents. Screen readers, which read pages aloud for blind and low-vision users, have relied on this for decades.

When the accessibility tree lacks information, an agent has to guess or stop searching. Problems often arise when teams are in a hurry or do not know what to check. For example, a size chart may be uploaded as an image because that is faster, but without a text description, an agent cannot read the sizes. Product colors may be shown as small swatches: without text labels, an agent cannot tell dark blue from black.

These oversights may not get in the way of a sighted shopper. But an agent looking for a discounted dark blue wool coat in size 16 may fail to find a suitable option. In the study, every model tested failed a task that required reading numbers shown only in an image with no text description—including the most advanced models in the test.

If an agent cannot find the information it needs, it may recommend a competitor’s product instead, because it could read that page. The shopper was ready to buy, but the store never got the chance to make the sale: the agent could not make sense of the page.

Barriers for shoppers with disabilities

AI agents are not the only ones who rely on the accessibility tree. According to the CDC, about one in four US adults has a disability. The Return on Disability Group’s 2024 report, “Global Economics of Disability Report,” puts the disposable income of Americans with disabilities at about $1.3 trillion.

The same oversights that keep agents from finding a product also create problems for shoppers. A screen-reader user cannot choose a coat color if the swatches are unlabeled. Someone navigating a site with a keyboard may not be able to reach checkout if the payment field responds only to a mouse click.

A shopper who runs into a problem like this may leave without telling the store why. To the marketing team, it looks like a product-page exit or an abandoned cart, with no explanation. When sales fall, teams typically review the messaging, offer, advertising materials and competitors. Website accessibility often does not make the list, even when it is what is preventing a purchase.

The effects can last beyond the holidays. A shopper who could not place an order in December may not return in January; they may also tell friends and family about a store where buying was easier. Conversely, a site that works well for these shoppers may keep them coming back after the holiday season ends.

What stores can fix now

As teams prepare for the holidays, they may feel it is too late to change the website. But in this period, inaccessibility costs more: traffic rises, the share of visits from AI tools grows, and an unlabeled color swatch or an image-based size chart can turn away both an agent and a shopper ready to buy. Some fixes do not require a new code release and are within the marketing team’s reach this week.

Start with best-selling products. On many platforms, product images and descriptions are stored in a content management system, so marketers can update them without a new code release. Add text descriptions to product images, and put size charts and discounted prices on the page as text.
Check key pages. Ask an AI agent to briefly summarize a product page, then check it with AudioEye’s free accessibility scanner. Neither method replaces a full audit, but together they can help identify problems that are quick to fix.
Track AI-service traffic as a separate channel. Monitor visits from ChatGPT, Perplexity and other such tools, and see how often they lead to purchases. These data will provide a starting point for planning next year.
Plan larger changes. Buttons, form fields and page layouts are usually set in website templates, so fixing one template can improve every page built from it. This matters more as agents move from finding products to placing orders: Meta’s Muse, launched in September, can already fill out forms and make purchases for people.

Every holiday season, stores compete for shoppers’ attention; this year, more of those interactions are taking place through AI agents. Fixes that let an agent read a product page also help people with disabilities find what they need and complete a purchase. That could expand a store’s customer base during the holidays and beyond.

Dataist’s take: Website accessibility affects two sales channels at once—the work of AI agents and purchases by people who use assistive technology. But the AudioEye test covered six websites, and quick checks do not replace a full audit. Individual fixes may help, but on their own they do not confirm that a site is accessible in every scenario.

Daily AI news

Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.

Only what matters — every day

Follow on X