AI search
Black Friday shoppers are asking AI. It recommends the typical retailer in 0.4% of answers
Shoppers are asking AI assistants what to buy this Black Friday. Here's what our retail data shows about being named, being recommended and being bought, and how to measure each one.
This is the first Black Friday where AI assistants are a normal place to start shopping. YouGov found that 34% of US adults planning to shop Black Friday or Cyber Monday are likely to use AI tools this year, up from 16% in 2025. In our data, AI names the typical retail brand in 29% of answers to questions about its category, and recommends it as the one to pick in 0.4%.
Those are two different numbers, and neither of them is a sale. That's the point of this article: AI visibility for a retailer is three numbers, not one.
This article is based on a white paper by my colleague Søren Koch, who works with retail and e-commerce brands on exactly this. We covered what you can still change before the peak in Black Friday 2026 in AI answers. This one is about being ready for AI shoppers, and about measuring it, so you can tell the CFO what worked when it's over.
The short answer
- AI shopping has doubled. 34% of likely US Black Friday shoppers plan to use AI, against 16% last year (YouGov).
- Shoppers use more than one assistant. ChatGPT and Gemini are tied at 29% each among those shoppers.
- Named is not chosen. Across 27 retail brands in our data, AI named the median brand in 29% of category answers and recommended it in 0.4%.
- Chosen is not bought. 89.9% of shoppers who used AI for product research checked its recommendation somewhere else before buying (Product.ai).
- Brand still does the heavy lifting. Being ready for AI shoppers turns brand strength into accurate answers. It doesn't replace it.
- So prove it in three layers, the AI Trifecta: what AI says (3RD), what shoppers do next (your analytics), and what they buy (your order data).
What shoppers will do this year
- Planned AI use has doubled. 34% of likely US Black Friday and Cyber Monday shoppers say they're likely to use AI tools, against 16% in 2025. ChatGPT and Gemini were picked by 29% each (YouGov, 1,495 US adults).
- Standalone assistants lead. In McKinsey's survey of about 21,000 consumers in 11 markets, 56% had used an AI chat or search tool for shopping in the past three months, against 27% for a retailer's own assistant. Nearly a third of AI shoppers named AI chat or search as the biggest influence on their latest purchase (McKinsey).
- The influence is already real. Salesforce reported that AI and agents influenced 20% of global online orders during Cyber Week 2025 (Salesforce).
Keep the sizes apart, though. AI influencing an order, AI sending the visit, and the order being completed inside an AI assistant are three very different numbers. Influence is the big one. AI-referred visits are a much smaller share of traffic, and checkouts inside AI are smaller still. Show only the biggest number and you lose the CFO's trust the first time someone checks.
All of this describes a channel that's no longer an experiment for a large minority of shoppers, that's spread across several assistants, and that already shapes purchases. It doesn't describe a world where one assistant, or one visibility number, is enough.
Number 1: Are you named?
Visibility is the share of AI answers to your category questions that name your brand. For the 27 retail and e-commerce brands in our data (4 October – 2 November 2026), the median was 28.9%.
And every assistant shows a different picture. Across 207 brands, the typical brand's weakest assistant named it about half as often as its strongest, a gap of 14 points (AI visibility differs by assistant). A retailer can look healthy in ChatGPT and be close to absent in Google AI Overviews, or the other way around.
Two things follow from that. First, treat each assistant as its own shelf: an average across them hides where you're missing. Second, don't read survey popularity as reach. How many shoppers say they'll use an assistant is not the same as how often that assistant names brands in your category, so an assistant with a small share of shoppers can still matter for you.
Number 2: Are you chosen?
The recommended rate is the share of answers that tell the shopper to pick you over the alternatives. Being in a list doesn't count. Being listed first doesn't count either.
- Retail median: 0.4%. 18 of the 27 retail brands were recommended in under 1% of answers.
- Across all 168 brands in our data, a competitor was recommended more often than the brand for almost half.
- Assistants differ in how often they pick at all. Claude and Perplexity picked a winner in about 7% of answers, ChatGPT in 5%, and Google AI Overviews and Copilot in under 2%.
A retailer at 50% visibility and 1% recommended is on every shortlist and rarely wins it. And the same brand can be recommended often in one assistant and almost never in another, so the recommended rate needs reading assistant by assistant too.
Number 3: Are you bought?
A recommendation is where the shopper's checking starts. In Product.ai's survey, 89.9% of shoppers who used AI for product research verified the recommendation through another source before buying, up from 85.7% in June (Product.ai).
So a recommendation turns into a sale only if what the shopper finds next agrees with it: the price on your page, the stock, the delivery promise, the reviews. And the page has to be readable by the assistant in the first place. Adobe's checks of retail sites found product pages less readable to AI than homepages, 66% against 75% (Adobe AI Content Visibility Checker, reported by Chief Marketer, April 2026).
What drives the first two numbers
Brand matters most. A study of 30 brands published by WARC (Charlie Oscar), as presented by Dan Wilson, attributed 63% of LLM visibility to long-term brand equity and 11% to citations and structured content (Dan Wilson). Brands that have spent a decade building recognition start ahead. A challenger that only works on its AI visibility, with no brand building behind it, is fighting the heaviest factor with the lightest levers.
That doesn't make the rest pointless. Citations and machine-readable product information are what let AI describe you accurately, and recommend you at all. In our data, the brands AI recommends are the brands it names on most answers to a question, and recommendations show up most on "best", "vs" and price questions. Those are the questions shoppers ask in Black Friday week.
So the way to think about it: brand building does the long work, being ready for AI shoppers turns that brand into accurate, recommendable answers in each assistant, and measurement shows whether either one moved sales.
What to do before Black Friday
Use this as a working list, not a slogan.
1. Measure each assistant on its own
Track visibility and the recommended rate separately for ChatGPT, Gemini, Google AI Overviews and AI Mode, Perplexity, and any retailer assistant that matters in your market. And track the questions shoppers actually ask: "best wireless headphones under €150 with next-day delivery" tells you more than "[your brand] Black Friday". Questions that name you will always look good.
In 3RD: Benchmark has a switch between visibility and Recommended rate on the Models view.
2. Make your catalog readable by machines
Product feeds, structured data and product pages should show price, availability, variants, delivery promises and return terms, the same way everywhere. Fix the pages an AI can't read well: thin product pages, content that only exists as images, blocked resources and specs that contradict each other.
3. Survive the check
Make sure what AI says about you matches your live site on price and stock, because shoppers check, and a mismatch costs you the sale. Then strengthen what they check next: reviews, press, comparison content, store locator and customer service facts.
4. Plan for several assistants
Shoppers are split between ChatGPT and Gemini, and many use more than one. Working on one assistant is half a program. And don't skip an assistant because few shoppers say they use it: check how often it names brands in your category first.
5. Keep brand building in the budget
Treat AI visibility work as something that amplifies your brand, not a substitute for building it. That's where most of the weight sits.
6. Connect marketing to the shop floor
AI shoppers often skip your category pages and land straight on a product page. That makes product page speed, merchandising and delivery promises part of the conversion, so bring e-commerce and operations into the plan.
7. Decide what "worked" means before the peak
Pick the topics and products where you want AI to recommend you more, and the ones you'll deliberately leave alone, so you have something to compare with afterwards.
How to prove it worked: the AI Trifecta
Visibility without commercial proof won't survive the post-Black Friday review with the CFO. Søren's answer is a three-layer frame we call the AI Trifecta.
Layer 1: Named and chosen (3RD)
By assistant and by topic, measure:
- Visibility: the share of relevant answers that name you.
- Recommended rate: the share of answers that recommend you over the alternatives.
- Competitive gaps: where a competitor is recommended more often than you.
- The sources and questions that explain the movement.
Layer 1 tells you whether the assistants changed what they say. On its own, it doesn't prove sales.
Layer 2: What shoppers did (your analytics)
Follow the paths between the answer and the order, for example in GA4:
- AI referrals: visits where the referrer or UTM shows ChatGPT, Perplexity, Gemini, Copilot and similar.
- Branded search: more people searching for you by name after periods of AI exposure.
- Direct traffic: where journeys without a click, or with a copy-paste, often end up.
- Asking: a "how did you hear about us?" question at checkout, with AI assistants as an option.
Layer 2 will be incomplete, because many AI journeys don't leave a clean referrer. That's why asking and branded search sit next to referrals, not instead of them. Read it as a direction, checked from several angles.
Layer 3: What they bought (your own data)
Close on the numbers the business already trusts:
- Orders and revenue
- Margin or contribution, where you have it
- New vs returning customers, if AI is claimed as a way to win new ones
Prove it with a comparison, and report ranges
Rising AI visibility next to rising revenue proves nothing on its own. Prefer:
- Worked-on vs untouched topics inside the same brand, compared over the same period.
- Holdouts: questions, categories or products you deliberately don't work on.
- Market splits: working on one market and not another, where traffic and delivery allow it.
Report the result as a range with its uncertainty, not "AI drove 12% of sales." Say what 3RD shows (changes in what AI says) and what only your order system shows (revenue). If there's no time to set up a comparison before Black Friday, say so, and present layers 1 and 2 as leading indicators, not as proven return.
Common mistakes
- Reporting one AI visibility number. It hides the assistant where you're missing and the gap between being named and being chosen.
- Counting a mention as a recommendation. In our retail data, the typical brand was named in 29% of answers and recommended in 0.4%.
- Treating a recommendation as a sale. Nine in ten AI shoppers check somewhere else first.
- Skipping the comparison group. Without categories you left alone, you can't tell your work from the season.
- Doing AI visibility work instead of brand building. Brand is the biggest factor. The work amplifies it.
- Quoting only the biggest number. AI influence, AI referrals and checkout inside AI are very different sizes. Show them apart.
About the data
- 3RD data: AI answers to the prompts 3RD customers track, 4 October – 2 November 2026, questions that name the brand left out. Retail figures cover 27 retail and e-commerce brands with at least 500 checked answers; the cross-industry figures cover 168 brands and 1.26 million answers. Recommended rate as 3RD defines it. The sample leans Nordic. Every number is aggregated and anonymized.
- Other sources: YouGov (US, 1,495 adults, July–August 2026), McKinsey (about 21,000 consumers in 11 markets, June 2026), Product.ai (US online shoppers, October 2026 edition; the 89.9% is among those who used AI for product research), Salesforce (Cyber Week 2025), Adobe (AI Content Visibility checks, 2026), WARC / Charlie Oscar (30 brands, as presented by Dan Wilson). These are surveys and vendor studies, and we report them as published.
FAQ
How many shoppers will use AI for Black Friday 2026? YouGov found 34% of likely US Black Friday and Cyber Monday shoppers plan to use AI tools, up from 16% in 2025.
Which AI assistants do shoppers use? Among likely US Black Friday shoppers, ChatGPT and Gemini were tied at 29% each (YouGov). Many people use more than one.
What's the difference between AI visibility and being recommended? Visibility means AI names your brand. Being recommended means the answer tells the shopper to pick you. For the median retail brand in our data, that was 29% against 0.4%.
How do I measure whether AI drove sales? With the AI Trifecta, in three layers: what AI says (visibility and recommended rate), what shoppers do next (your analytics) and what they buy (your order data). Compare the categories you worked on with the ones you didn't, and report a range.
What does agentic-ready mean for a retailer? That AI assistants can read your catalog, describe you accurately and recommend you, and that what a shopper finds when they check (price, stock, reviews) agrees with the answer.
Next step
Before Friday, check your top ten Black Friday category questions in each assistant your shoppers use. Note where you're named, where you're picked, and which competitor is picked when you're not.
If you want those numbers per assistant, side by side with your competitors, start for free and open Benchmark.

