top of page

Generative Engine Optimisation for Black Friday: How Retailers Keep AI Assistants Accurate About Their Deals

Nov 9
5 min read

This guide was produced by AI, TELL ME!, a full-cycle AEO & GEO agency with our own AI search monitoring platform.

Summary

A shopper who asks an AI assistant for a Black Friday code gets one with total confidence, and in an August 2026 test by a coupon-verification company, 59% of the codes that could be checked failed. Generative engine optimisation for Black Friday therefore starts with the facts of the deal, because an LLM (large language model) builds its answer from whatever pages describe a sale, and a retailer can edit the record behind an answer with no way to edit the answer itself. YouGov puts planned AI use at 34% of likely sale shoppers in both the US and Great Britain, so a wrong code now reaches a third of the people about to buy. The sections below cover the studies, six changes before 27 November and a twice-weekly test.

Cover of the generative engine optimisation for Black Friday guide: 34% of likely sale shoppers plan to use AI, 47% in Great Britain find it hard to tell if a discount is genuine, 24% of codes for the most-searched stores worked, and 59% of checkable AI promo codes failed.

Black Friday falls on 27 November, and a growing share of its shoppers will ask an AI assistant for the price, the code and the start date before opening a retailer's site. The retailer sees none of those replies and has 18 days to make sure the facts behind them are right. Generative engine optimisation for Black Friday is the work of keeping the codes, dates, prices and terms that LLMs state about a sale accurate.

How many shoppers will use AI for Black Friday 2026?

A third of likely sale shoppers will, according to YouGov, which puts planned AI use at 34% in both the US and Great Britain. YouGov's US survey of 1,495 adults compares the 34% with 16% a year earlier. In Great Britain the figure stood at 19% last year, and YouGov's Great Britain report adds that 47% of likely sale shoppers find it hard to tell whether a discount is genuine.

Do AI assistants give working Black Friday discount codes?

Often they don't: SimplyCodes, a coupon-verification company, put 468 questions about promo codes at 150 stores to ChatGPT, Gemini, Claude and Perplexity in August 2026, and 59% of the codes it could check against its records of 12.5 million promotions failed. The company lowered that figure from 67% after a manual audit showed that some unmatched codes were valid. Working rates were 47% for Claude, 43% for Gemini, 41% for ChatGPT and 37% for Perplexity.

Search demand shaped the result: codes for the 50 most-searched stores worked 24% of the time, against 44% for the middle 50 and 56% for the 50 least-searched. The study offers no explanation, and the likeliest candidate is volume, since the busiest stores attract the most voucher and affiliate pages and any of those can carry an expired code.

Three limits apply to the study:

  • SimplyCodes sells code verification, used its own database as the yardstick and published without independent review.

  • The test covered promo codes in August, which leaves Black Friday dates, prices and discount terms unmeasured.

  • The invented-code result applied to API access: Perplexity, Gemini and ChatGPT made up 632 codes for six stores that do not exist, Claude made up none and the consumer ChatGPT and Gemini apps declined.

The record behind the answer is the part a retailer can fix

A retailer can't edit an assistant's reply, so the work sits on the pages the reply is built from. In AI, TELL ME!'s studies of CRM, pharma and ski brands, brand-owned websites supplied between 5.6% and 32.0% of the citations or citation weight behind AI answers. None of those categories is retail, so the range shows a pattern and no retail result. Paid placement leaves the reply untouched too: OpenAI says ChatGPT ads run apart from the chat model, as the article on ChatGPT ads in the UK sets out.

A single check proves little: Ronald Sielinski's arXiv study of Perplexity Search, OpenAI SearchGPT and Google Gemini found citations varying substantially between repeated samples of the same queries, with many apparent differences inside the measurement noise.

Generative engine optimisation for Black Friday: six changes before 27 November

Six changes fit inside the 18 days left, and each targets a fact that an assistant can misstate.

  1. Put every offer on one dated page that states start and end date, time zone, discount depth, code or "no code needed" and exclusions in plain sentences, at a URL that stays unchanged until Cyber Monday on 30 November.

  2. Write out the terms that make codes fail: minimum spend, excluded brands, one use per customer and app-exclusive offers.

  3. Mark last year's sale pages as ended, with a link to this year's offer page, because a 2025 Black Friday page stays crawlable until someone edits it.

  4. Match every channel to the offer page: feeds, marketplace listings and email landing pages carry the same dates and prices, and affiliate and voucher partners get the live codes with a request to delete the rest.

  5. Show prices in the page text, as the guide to answer engine optimisation for e-commerce asks, and see the breakdown of generative engine optimisation for e-commerce for the wider retail picture.

  6. Brief customer service on whether the team honours a code that an assistant produced.

Prompts to test and how often to run them

A retailer should run each shopper prompt at least three times on every assistant and repeat the round twice a week until 1 December. The table lists three prompt types and the error to look for in each.

Prompt a shopper types

What can go wrong

What to check

"[Brand] Black Friday discount code"

The reply lists codes that expired or never existed

Compare every code with the live offer page

"When does the [brand] Black Friday sale start?"

Last year's dates appear as this year's

Compare the dates with the offer page

"Best Black Friday deals on [category] UK"

The brand is missing or shown at an old price

Record whether the brand is named, at what price and which sources are cited

Each wrong code, date or price goes into a log with the prompt, the assistant and the date, then gets traced to the page the reply cites: a partner page calls for a request and the retailer's own 2025 page calls for an edit. The prompt types come from the guide to checking whether ChatGPT recommends a brand, and the guide to AI hallucinations about brands sets out the claim-by-claim method.

What an agency adds in the 18 days before the sale

AI, TELL ME! is a full-cycle GEO agency with 17+ years in marketing, SEO and PR, working in the order of monitoring, diagnosis, strategy, implementation and measurement. For a sale, AEO Monitoring comes first as a free scan of what assistants say about the brand today, GEO Readiness then diagnoses the pages behind those replies and the fixes follow by impact. The agency works as a consultant beside an in-house team or takes the whole job on, with PR for GEO as one extra because many pages behind a reply belong to publishers and review sites.

FAQ

Do AI assistants give working Black Friday discount codes?

Often they don't: in SimplyCodes' August 2026 test, 59% of the codes from ChatGPT, Gemini, Claude and Perplexity that could be checked against the company's records failed. The test covered promo codes in general and left Black Friday unmeasured.

How many shoppers will use AI for Black Friday 2026?

YouGov found 34% of likely Black Friday and Cyber Monday shoppers planning to use an AI tool in both the US and Great Britain, up from 16% and 19% a year earlier.

Can a retailer pay to have its deal stated correctly in ChatGPT?

No, because OpenAI says advertisers can't shape, rank or alter ChatGPT's responses, so the accuracy of a reply depends on the pages the model reads.

How should a retailer test what AI says about its sale?

A retailer runs four to six shopper prompts per category at least three times on every assistant twice a week until 1 December and compares each code, date and price with the live offer page.

Before the sale starts, see what AI assistants say about your brand with a free AEO Monitoring scan from AI, TELL ME!.

bottom of page