Where AI helps
Start with the dull, repeated work behind the shop. That is where the checkable results are.
| Work | What has been shown or reported | Where it comes from |
|---|---|---|
| Search, chat and listings on a platform | Seven randomised experiments on one large cross-border platform raised sales by between nothing and 16.3%, mostly by turning more visits into sales. Returns and ratings did not get worse. AI-written Google ad titles showed no gain. | Experiments on one platform. Not peer-reviewed. Some authors worked for or advised the platform. |
| Search and translation | When eBay translated searches and listing titles for Spanish-speaking buyers, US exports to them on eBay rose by 10.9%. | eBay case. A peer-reviewed study. |
| Product data | Walmart said generative AI created or improved more than 850 million pieces of catalogue data. | Walmart case |
| Customer support | Human agents with an AI assistant resolved 15% more issues per hour. Klarna’s AI assistant took two-thirds of support chats, and the company later hired people again. | Field study of 5,172 agents, Klarna case |
| Marketing images and copy | Klarna said it cut image production from 6 weeks to 7 days. | Klarna marketing case |
| Counting stock | Starbucks retired an AI stock-counting tool after about nine months, following staff complaints about wrong counts. | A press report |
Read the big numbers with care. In a 2026 survey by the chip maker NVIDIA, 91% of retail and consumer-goods companies said they were using or assessing AI. “Assessing” is not using. McKinsey estimates that generative AI could be worth $240–390 billion to retailers. eBay says more than 10 million sellers have used its AI listing tools. That counts everyone who tried a tool once. A trade publication’s tests found the tool misidentified a product, and the error spread into the title, category and description.
Most published numbers come from very large companies. Small retailers mostly get AI through the tools they already pay for, or through the platform they sell on, which switches it on for them. In the US Census Bureau’s survey, about 14% of retail firms used AI. See micro and small companies.
Risks and rules
- What your chatbot says, you said. A Canadian tribunal ordered Air Canada to pay a customer who relied on a wrong answer from its website chatbot. The tribunal found the company responsible for all the information on its website, chatbot included.
- “AI” that needed people. The US Securities and Exchange Commission found in January 2025 that the “vast majority” of drive-through orders taken through Presto Automation’s own voice AI needed a person to step in. Presto had told investors that its AI removed the need for people to take orders (Presto case). If you buy “AI ordering”, ask what share of orders run with no person involved.
- AI answers can pass you by. When Google showed an AI summary, people clicked a normal search result on 8% of visits, against 15% without one, in the browsing records of 900 US adults collected by the Pew Research Center. In researchers’ tests with skincare products, three AI models recommended the better-known brand every time when the products had identical specifications. Those tests have not been peer-reviewed.
- Tell customers they are talking to AI. In the EU, the AI Act says AI systems that talk directly to people must be designed so that people are told they are dealing with AI, unless that is obvious. The duty falls on the provider: whoever develops the system and puts it into service under its own name. If you buy a chatbot, ask your vendor how it meets this. If you build your own, the duty may be yours. The rule has applied since 2 August 2026.
- Fake reviews. In the United States, a federal rule announced in August 2024 bans fake reviews and testimonials, including AI-generated ones. It also bans suppressing negative reviews through threats.
- Do not judge customer-facing AI on cost alone. Klarna published large savings from AI support. Its chief executive later said cost had weighed too heavily. Measure quality too. See measure time saved against a baseline.
Practices for this industry
All 30 practices in the library apply in this industry. None of them is specific to one sector. For the ones below, a study, a company case or a worked example from this sector shows why they matter here.
- Ask AI vendors for proof before you buy, and test on your own material Rite Aid never asked its two facial recognition vendors about accuracy testing. It was banned from the technology for five years.
- Keep a person checking anything that goes to a customer or carries legal weight A tribunal held Air Canada to a wrong chatbot answer. A regulator banned Rite Aid from facial recognition after staff acted on false matches.
- Train staff on the tools you actually chose Rite Aid’s staff acted on facial recognition alerts without being told the system produced false matches.
- Budget for the work AI creates A tribunal held Air Canada responsible for a wrong chatbot answer. Keeping answers in line with policy is work someone must do.
- Know the rules that already apply to your AI use A Canadian tribunal held an airline to its chatbot’s wrong answer. A US regulator banned a retail chain from facial recognition.
- Measure time saved per task against a baseline Klarna judged AI support mainly on cost, and its chief executive later said that was a mistake. Measure quality as well as time.
- Name one accountable AI owner Air Canada’s chatbot gave an answer that did not match company policy, and the airline had to pay. Someone must own what a chatbot says.
- Start with one costly process, not a company-wide rollout eBay and Walmart each applied AI to one narrow, high-volume task: translating listings and fixing catalogue data.
- Start with ready-made AI in software you already use, before building your own A small online shop bought a ready-made chatbot for WhatsApp, which its customers already used. Its founder estimates 30% less time on customer questions.
- Set the limits before the tool acts Rue La La let its pricing tool recommend rises only, capped them, and accepted a rise only if the price stayed below other websites.
Pick your company size
Size changes what is practical more than sector does. Each link opens the library filtered to that size.
Company cases
Each case is a named company with numbers and a public source. All cases.
- accessiBe: an AI plug-in sold as making any website accessible, and a $1 million order accessiBe sold website owners an AI plug-in and said it would make any site meet accessibility standards. The US consumer regulator said it did not, and that reviews praising it were paid for. The company paid $1 million.
- Air Canada: held responsible for what its website chatbot told a customer Air Canada’s website chatbot gave a customer wrong information about refunds. A Canadian tribunal ordered the airline to pay. It rejected the argument that the company was not responsible for its chatbot.
- eBay: machine translation raised cross-border sales by about 11% eBay began translating search queries and listing titles for Spanish-speaking buyers. A peer-reviewed study found that US exports to those buyers on eBay rose by 10.9%. It is one of the few AI results in retail measured by outside researchers.
- Klarna: an AI assistant took two-thirds of support chats, then the company hired people again Klarna reported large savings from an AI support assistant. About a year later its chief executive said the company had judged the work too much on cost, and it began hiring human agents again. The AI assistant stayed.
- Milkforbubs: a small online shop put a ready-made chatbot on customer questions A small online shop in Singapore bought a ready-made AI chatbot to answer customers. The team loaded it with common questions, trained staff, and keeps correcting its answers. The founder estimates the team spends 30% less time on customer questions.
- Presto Automation: the US regulator found most “AI” drive-through orders needed a person Presto sold restaurants a voice AI for drive-through orders and told investors it eliminated human order-taking. The US securities regulator found that people working off site, mainly in the Philippines and India, entered the vast majority of orders. Presto settled without admitting or denying the findings.
- Rite Aid: banned from facial recognition for five years after buying it without testing it Rite Aid used facial recognition to spot suspected shoplifters. The US consumer regulator said the chain never asked its vendors how accurate the system was, never tested it and did not train staff. Shoppers were wrongly accused. The chain was banned from the technology for five years.
- Rue La La: machine-learning price recommendations raised revenue by about 10% in a test Rue La La, an online fashion retailer, had a tool built that predicts demand for products it has never sold and recommends a price for each. In a five-month test, researchers estimated that revenue on the tested products rose by 9.7%. The retailer set firm limits on what the tool could do before it let it near a price.
- Temple & Webster: an online furniture retailer with about 200 staff put AI into customer care first Temple & Webster sells furniture online with about 200 employees. It reports that AI and automation handle 80% of customer contacts in whole or in part, and that customer care now costs over 60% less as a share of revenue. It then set up central approval, training and a contact person for AI in every department.
- Walmart: generative AI used to fix 850 million pieces of product data Walmart told investors it used generative AI to create or improve more than 850 million pieces of product data. The task is narrow, repetitive and easy to check, which is where AI tends to work.
Studies from this sector
These studies were run on work that is typical of this sector.
Sources
- Fortune (28 May 2026). Report that Starbucks retired an AI inventory-counting tool. fortune.com
- NVIDIA (2026). State of AI in Retail and CPG. A survey by a technology vendor. blogs.nvidia.com
- McKinsey. LLM to ROI: How to scale gen AI in retail. An estimate of potential value. mckinsey.com
- European Union. AI Act, Regulation (EU) 2024/1689, Article 50 (transparency obligations). Text on the European Commission’s AI Act Service Desk. ai-act-service-desk.ec.europa.eu
- US Federal Trade Commission (14 August 2024). Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials. Rule: 16 CFR Part 465. ftc.gov
- US Census Bureau (May 2026). Large Firms With at Least 20 Employees Biggest AI Users. Business Trends and Outlook Survey. census.gov
- Value Added Resource (August 2026). eBay Main Street Touts AI For Small Business, While Sellers Carry The Risk. A trade publication. valueaddedresource.net
- Pew Research Center (22 July 2025). Google users are less likely to click on links when an AI summary appears in the results. pewresearch.org
- Chu and Hou (2026). Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems. arXiv 2606.17443. Not peer-reviewed. arxiv.org
- US Securities and Exchange Commission (14 January 2025). Settled administrative proceeding against Presto Automation Inc., Release 33-11352. sec.gov