One Shopper, Three Readers: How a Single Holiday Purchase Moves Between Search, Assistants, and Agents

Picture one shopper in the week before Thanksgiving. She has two jobs: a gift for her ten-year-old nephew, and the household stock-up she does every year when the Cyber Week deals land.

She starts on Google with "best STEM gifts for 10 year olds." An AI Overview answers before she clicks: three product types and a price range. She takes that into an assistant and adds what Google could not know: under $40, has to arrive by the 20th, and he already owns a microscope. The assistant suggests two kits. She goes to a retailer app, searches one of them by name, and asks the retailer's own assistant whether it needs batteries. Then she turns to the other job and asks her agent to reorder the dog food and the cleaning supplies before the deals end. She approves the basket from her phone.

That is one shopper, one week, and five handoffs. At Genrise we describe the digital shelf as having three readers: the Human Shopper, the AI-Assisted Human, and the Autonomous Agent. The shopper above was all three. The personas are not three different customers. They are three modes the same customer moves between, sometimes within a single purchase.

The clearest way to size each mode is by reach: how many shoppers use it at least once. Nearly every shopper still searches and browses as a Human Shopper. Between 40% and 50% now shop with an AI assistant at some point: NielsenIQ found 42% of U.S. consumers used an AI tool to shop in the past month, and Salesforce puts it at half of shoppers. About 5% have let a fully autonomous agent place an order for them, according to NielsenIQ. Reach does not tell us how much of each shopper's journey happens in each mode. It tells us that most shoppers now move between at least two.

This piece argues one thing: Holiday 2026 is the first peak season where all three modes are live at once, and each is changing on its own clock. The keyword click is being repriced. AI-assisted research is growing fastest and is the hardest to see. And agents can now finish a basket, starting with exactly the routine items that holiday stock-ups are made of.

The season in numbers: CPG is outgrowing the total

Adobe forecasts U.S. online holiday spending of $275.1 billion between November 1 and December 31, up 6.7% year over year, with Cyber Week accounting for $47.5 billion. The headline is steady growth. The category detail is where consumer brands should look.

Grocery is forecast to grow 10.3% and cosmetics 9.5%, both well ahead of the total. Electronics, the largest category, is forecast at 5.9% and apparel at 4.7%. Adobe also expects a sharp Cyber Week lift in everyday essentials compared with September averages: personal hygiene up 150%, baby products up 113%, pet products up 93%, and household cleaning up 49%.

Two other forecasts frame the rest of this piece. Adobe expects AI-referred traffic to U.S. retail sites to rise 130% year over year. And among marketing channels, Adobe forecasts revenue share growth of 17% for social media, 12% for affiliates, and 4% for paid search.

For consumer brands, the holiday is two seasons running in parallel. There is the gifting season, built on research and judgment. And there is the stock-up season, built on routine. As the rest of this piece shows, the same shopper treats them in different modes.

How far shoppers will let AI go

Surveys published ahead of the season agree on the shape of shopper comfort with AI, even when their numbers differ. Shoppers hand over research readily. They keep the purchase for themselves, except for small, routine items.

Two caveats apply to every row. These are stated intentions from surveys, not observed transactions. And the samples differ: Accenture surveyed more than 16,000 shoppers in ten countries, American Express 2,005 U.S. adults, and Shopify U.S. consumers within a nine-country study.

Read together, the ladder maps onto the three modes. Research and recommendation belong to the AI-Assisted Human. Delegated buying is real but narrow, and it starts where consumer packaged goods live: everyday, low-cost, routine purchases. Most Cyber Week stock-up baskets of shampoo, detergent, and dog food sit comfortably under a $102 approval threshold. Most gifts do not.

Mode 1: the keyword click is being repriced

The Human Shopper still carries most of the volume, and keywords still matter. What is changing is who pays for the click, and how often there is a click at all.

Organic is shrinking, and not only in the U.S. A Growth Memo analysis of Similarweb estimates for 10,000 domains per country found that organic search visits fell 13.2% in the U.S. and 24.4% in France over the 24 months to August 2026. Over the same period, paid search visits grew 21% in the U.S. and 17.3% in France. Paid growth did not fill the gap: it offset roughly 37% of the U.S. organic loss and about 7% of the French loss. The data is modeled traffic, not revenue, and the losses were uneven by site type, with publishers hit hardest. The direction for retail is still clear: the same intent is increasingly reached through a paid placement rather than an organic one.

AI Overviews answer before the click. Ahrefs found that top-ranking informational pages see a 58% lower click-through rate when an AI Overview appears. France offers a near-real-time test. Google launched AI Overviews there on July 22, 2026, and Ahrefs data on 149 ecommerce sites showed click-through rates down 5.4% within days. The pattern travels across markets.

The holiday detail: gift research is the most exposed query type. Seer Interactive's analysis of 5.47 million queries found AI Overviews on roughly 5% of transactional queries, but on about 86% of question queries and about 95% of comparison queries. Holiday gift research is almost entirely questions and comparisons: "best gifts for," "this versus that," "top stocking stuffers." That is precisely where the answer now arrives before the click. Seer also found that paid click-through rates rise when an AI Overview is present, which fits the organic-to-paid shift above.

What this means for keywords. The shopper in our opening used keywords twice: a broad comparison query at the start, and a specific product name at the end, after an assistant had narrowed her choice. The first kind is increasingly answered by Google. The second kind is still a click, and increasingly a paid one. The long tail between them is turning into prompts. On Amazon, the move to 75-character titles and a searchable Item Highlights field concentrates the words a Human Shopper scans into less space, which makes every one of them matter more.

Mode 2: the fastest-growing mode is the hardest to see

The AI-Assisted Human is where most of the growth is. It is also where measurement is weakest.

Three numbers, three definitions. Adobe forecasts AI-referred traffic to U.S. retail sites will rise 130% this season, measured by clicks from AI tools. Growth Memo's Similarweb analysis puts AI referrals at roughly 0.5% of total visits. Salesforce forecasts that 20% of holiday ecommerce traffic will originate from AI agents, but its definition includes consumer-facing bots, backend agents, and competitor scrapers. All three can be true at once. AI referrals are growing fast from a small base, and broad definitions of "AI traffic" capture machines as well as people.

Influence is not referral. Go back to the opening. The assistant shaped her choice, but she bought through a retailer app after searching by name. In referral data, that sale looks like a branded search. Retailer assistants such as Alexa for Shopping and Walmart's Sparky never appear as referrals at all, because the shopper never leaves the retailer. Every time the shopper switches mode, the record of the AI's role gets thinner.

AI is a research-day tool. Adobe's peak-day forecasts make the point. AI-referred traffic is expected to grow 159% on Thanksgiving, 95% on Black Friday, and 88% on Cyber Monday. Growth is highest on the day shoppers plan and lowest on the day they buy. Adobe's own consumer survey adds that shoppers who use AI report more confidence in their purchases and fewer returns.

Prompts carry constraints that keywords never did. "STEM gift" is a keyword. "Under $40, arrives by the 20th, he already has a microscope" is a set of constraints. An assistant can only recommend a product whose content answers those constraints: age range, what is in the box, size, dietary fit, delivery promise. Salesforce expects one in three ecommerce sites to have a branded shopper agent live by Cyber Week, which means more of these questions will be asked on retailer and brand sites as well as in general assistants. (Our AI shopping assistants field guide covers how the major assistants read product content.)

Mode 3: the first holiday agents can finish a basket

The Autonomous Agent is the smallest mode by reach: about 5% of U.S. consumers have let a fully autonomous agent place an order (NielsenIQ). What is new is that, for the first time at holiday peak, the infrastructure to complete a purchase on a shopper's behalf is live in several places.

Personal agents such as Meta's Muse and Instinct can now act across a shopper's connected accounts. Google began rolling out Universal Cart across Search and the Gemini app in the U.S. on October 15, with saved carts, price and stock alerts, and checkout through Google Pay or the merchant's site. TikTok announced Buy Direct on October 5, a native checkout built on the Universal Commerce Protocol in which the brand remains the merchant of record.

Each of these keeps a person in the loop. Muse requires approval before every purchase. Instinct routes purchases through Stripe's Link for approval. Universal Cart tracks prices and holds the cart, but the shopper still decides when to buy. Delegation this holiday means an agent doing the monitoring, comparing, and cart-building, then asking.

The same shopper delegates selectively. The survey ladder explains where this mode will appear first. Gifts stay in Mode 2: they depend on taste, judgment, and knowledge of the recipient. The stock-up moves toward Mode 3: known products, low prices, and a clear reorder pattern. American Express found 52% of consumers comfortable with an agent buying everyday, low-cost items, against 15% for high-value ones. For consumer brands, the first agent-completed holiday baskets are more likely to contain detergent than drones.

Access is still being negotiated. Retailers have taken different positions on outside agents, from blocking them, to sending their own agents into other assistants, to opening checkout to them. Each posture has its own commercial logic. For a brand, the practical effect is that a product can be chosen in one place and bought in another.

One purchase, three readers

Put the modes side by side and the shift becomes visible. It is not a migration from one channel to another. It is a shopper moving from click, to answer, to delegation, and back, depending on the task in front of her.

The bottom row is the one that matters most for the season. Every handoff is a point where a product can drop out of consideration or lose credit for being considered. The product content that carries a shopper across those handoffs is the same content, read three ways: scanned by a person, interpreted by an assistant, and parsed by an agent.

How the season is likely to unfold

The forecasts above suggest a sequence. It is a reading of the data, not a prediction of results.

What to watch

Whether AI-referred shoppers keep converting better. Adobe measured AI-referred conversion moving well above other sources during the 2025 season. Its 2026 actuals will show whether that held as volumes grew.

Whether paid keeps absorbing what organic loses. If the organic-to-paid shift continues through peak season, the cost of reaching the Human Shopper rises even when demand does not.

Whether any agent reports holiday purchase volume. No personal agent has published how much of its usage is shopping. A holiday disclosure would be the first real measure of Mode 3.

Whether branded shopper agents arrive on schedule. Salesforce's one-in-three forecast for Cyber Week is a test of how quickly retailers and brands put their own assistants in front of shoppers.

None of these change the underlying picture. The shopper has not split into three. She has learned to move between three readers, and she will do it more this season than in any season before. The brands whose content holds up at every handoff are the ones that stay in her basket.

References

Primary sources and research

• Adobe, Adobe: U.S. Holiday Shopping Season to Hit Record $275.1 Billion Online, Rising 6.7% YoY (September 28, 2026). https://www.businesswire.com/news/home/20260928285775/en/
• Adobe Digital Insights, 2026 Holiday Forecast. https://business.adobe.com/resources/holiday-shopping-report.html
• NielsenIQ, 42% of consumers now use AI tools to shop, NIQ data shows (May 5, 2026). https://nielseniq.com/global/en/news-center/2026/42-of-consumers-now-use-ai-tools-to-shop-niq-data-shows/
• Accenture, Consumer Holiday Research 2026 (surveyed August 13 to September 1, 2026). https://www.accenture.com/en/insights/consulting/consumer-holiday-research
• Salesforce, Holiday Retail Predictions 2026 (July 20, 2026). https://www.salesforce.com/blog/holiday-retail-predictions-2026/
• Shopify, Holiday Retail Research 2026 (consumer survey conducted July 2026). https://www.shopifyholidayretail.com/
• American Express, Agentic Commerce Business Playbook and Amex Trendex research (August 2026). https://www.americanexpress.com/en-us/newsroom/articles/amex-for-business/agentic-commerce-business-playbook.html
• Kevin Indig, Growth Memo, Organic traffic is down, but it's not all AI (October 26, 2026). https://www.growth-memo.com/p/organic-traffic-is-down-but-its-not
• Ahrefs, AI Overviews reduce clicks (May 2026). https://ahrefs.com/blog/ai-overviews-reduce-clicks/
• TikTok for Business, TikTok agentic solutions (October 5, 2026). https://ads.tiktok.com/business/en-US/blog/tiktok-agentic-solutions

Reporting

• mind Retail, E-commerce: click-through rate down by 5.4% since Google AIO launch in France (August 20, 2026).
• Search Engine Land, Google AI Overviews CTR shows early signs of recovery (Seer Interactive study, 2026).
• PPC Land, Google's Universal Cart reaches US Search and Gemini but not YouTube (October 2026).

Related Genrise insights

• Personal AI agents and the new origin of purchase intent
• Universal Cart and agentic commerce
• Alexa for Shopping: what CPG brands need to know
• AI shopping assistants field guide
• Amazon listing content optimization
• Walmart algorithm and Sparky
• The three-shopper shift in snacking, Q1 2026
• The three-shopper shift in pet nutrition, Q1 2026