Search for a Black Friday sofa and the recommendation may come from a magazine’s furniture guide. Search for a coupon for the very same store and the answer may come from a deals site. Both can mention the retailer without quoting the retailer’s own promotion. In two November 2026 studies, Wayfair provides a particularly good example of how those paths diverge.
With Black Friday on November 27, the immediate job for an ecommerce team is to follow one of its offers all the way through. Record the exact question, read the page the answer cites, and try the offer in an eligible basket. That sequence tells you whether you need to fix your own sale page, contact a publisher, or investigate checkout. It also prevents an easy mistake: recording a mention as a working discount.
- 01The question changes the sourceA broad buying question can draw on a publisher’s buying guide. A search for a retailer’s code can surface an affiliate or coupon site.
- 02The coupon count covers GoogleThe Search Post saw AI Overviews in 59 of 60 searches. Of those 59, 36 displayed a particular code. Checkout success was outside the study.
- 03A retailer can be its own cited sourceProfound found AT&T’s sale hub in a growing share of monitored answers. Eligibility, trade-in terms and timing give a shopper reasons to consult the original offer.
- 04Test the customer’s version of the dealMatch country, eligibility and basket contents. A discount for an app user or new subscriber can fail for an otherwise identical shopper.
01 — The evidenceTwo studies asked different questions
On November 13, Profound published its Black Friday in AI Search 2026 research. For its US year-over-year comparison, it followed the same 1,250 English-language prompts on ChatGPT and Google AI Overviews. It compared November 4–10 last year with November 3–9 this year. Both windows ended the same number of days before Black Friday. The researchers weighted the two engines equally. That design is useful for studying changes in the answers to a fixed set of questions, even as shoppers’ actual mix of questions changes.
On November 21, The Search Post asked something narrower. Its researchers ran a desktop Google coupon-code query for each of 40 US and 20 UK stores. They recorded whether an AI Overview appeared and whether it printed a particular code. They also read the listed sources. This gives a retailer a close look at what someone asking for a code might see on one morning. It gives no direct estimate of how often ChatGPT dispenses codes or how much revenue either platform generates.
| Study | What it observed | What the result can answer |
|---|---|---|
| Profound, Nov 13 | Matched US prompts on ChatGPT and Google AI Overviews across two seasons | Which brands and pages gained visibility in the tracked answers? |
| The Search Post, Nov 21 | 60 Google store-code searches on desktop in the US and UK | Which answers displayed codes and which sources did Google credit? |
02 — The merchantWhen the retailer supplies the answer
Profound reported that AT&T’s Black Friday hub was cited in 11.7% of the monitored answers in its US phone comparison, up from 0.7% a year earlier. Its discussion points to a page that puts monthly prices alongside named devices, trade-in conditions and purchase options. That figure belongs to the tracked answers in the study. It cannot tell a retailer which change caused the increase or how many people went on to buy a phone.
Look at the offer through a shopper’s eyes. A headline price may depend on the device handed in, the plan selected or the way a credit is paid out. A shopper asking about the best phone deal needs those conditions before two advertised prices can be compared. A page that places the terms next to the price gives that person something concrete to verify. The same principle applies to simpler offers. A return visitor should be able to tell which deal is active today, even if last week’s promotion used the same URL.
Open your offer on a phone and hand it to someone who has not seen the campaign. Ask what they would pay, which product qualifies and when the offer ends. The first question they cannot answer identifies a useful edit.
03 — The publisherFollow Wayfair through two kinds of search
In Profound’s US furniture analysis, a Forbes guide was the most-cited page in the comparison week. Wayfair appeared in that guide. That is one way a retailer can enter a shopper’s shortlist before the retailer’s own page comes into view. In the separate coupon study, The Search Post reported that Google credited DealNews for Wayfair codes. A furniture buyer asking which store to consider and a buyer asking how to pay less at that store encountered different sources.
That distinction changes who can correct an error. If a magazine describes a promotion as applying to every sofa, a narrower statement buried in the retailer’s terms may do little for the person reading the guide. Contact the editor with the exact sentence, a current public offer page and the missing condition. If the merchant’s own terms also disagree with checkout, resolve that discrepancy first. An editor needs a stable reference, and the customer needs a store that will honor the corrected offer.
Which sofa should I shortlist?
Check that the featured model and sale price remain available. A price for one fabric or size can be misleading beside a photo of another.
Can I use a code at this store?
Determine whether the code is public, tied to one customer, limited to an app, or already built into the displayed price.
04 — Coupon resultsWhat 36 answers tell a retailer
Of the 60 searches in The Search Post’s sample, 59 returned a Google AI Overview. Of those, 36 gave at least one specific code. The split was 24 of 39 US overviews and 12 of 20 UK overviews. Keep the denominator with the number. Thirty-six describes answers that displayed a code, including answers that displayed more than one. It is neither a count of distinct discounts nor a count of successful purchases.
Some answers cited a store’s own offers page, while others drew on publisher pages, coupon aggregators or social posts. The researchers did not try the codes at checkout. They also observed that rerunning a search could change the answer within minutes. Capture the answer your customer saw, including the date, market and source, before trying to reproduce it. Finding a different answer a few minutes later can be informative without resolving the original complaint.
| Market | Searches | AI Overviews | Named a code |
|---|---|---|---|
| US | 40 | 39 | 24 |
| UK | 20 | 20 | 12 |
| Combined | 60 | 59 | 36 |
05 — Offer testingTry the terms in an eligible basket
Consider an illustrative offer promising $20 off a $100 order before shipping. A customer’s basket contains $90 of eligible items plus $12 shipping. The checkout total reaches $102, yet the qualifying merchandise stays below the threshold. The public offer page might explain this clearly while the AI answer simply says “$20 off $100.” Both the displayed answer and the merchant’s checkout behavior matter to the person who has been promised a discount.
Try the same promotion on a basket that meets every stated condition, then change just one condition. Use a regular-price item if sale goods are excluded, a qualifying country and the right customer type. Stop at the order review screen once the discount appears. Save the before-discount subtotal, the applied reduction and the text shown when a basket is ineligible. If a cart requires someone’s personal one-use code, the right public description may be how to qualify for the program. Republishing that person’s code would create a new problem.
- Write down what the AI answer promised in the shopper’s words.
- Compare it with the source and the official terms for that market.
- Change one basket condition at a time and record the observed outcome.
06 — Search operationsGive the right team the correction
When the store’s own page is outdated, the merchandising owner can amend the end date, eligible products or minimum spend. If a publisher reproduces an old campaign, send the publisher the precise correction and the current URL. If a clean eligible basket fails everywhere, give the checkout team the item, market, timestamp and expected discount. Each outcome has a different owner. Escalating all three to the SEO team leaves the person closest to the problem out of the loop.
If you also submit promotions to Google Merchant Center, check that the approved markets, product mapping and promotion terms agree with the store. Google’s promotions specification has separate fields for details such as minimum purchase amount and effective dates. Keeping those fields accurate helps that channel’s customers. A change there alone gives you no assurance about which page an AI Overview or ChatGPT will cite. For a fuller look at product identifiers and feeds, see our merchant guide to product data in AI shopping. Google’s promotions data specification.
07 — MeasurementSeparate seeing a deal from redeeming it
Track the answer, the referral and the sale as separate observations. An answer capture tells you which brand or URL appeared for a particular question. A referral shows that someone arrived through a link identifiable to your analytics. A redeemed code shows that a checkout accepted the code. A shopper can see a recommendation on a phone and buy from a laptop, or use a broadly shared code they originally received by email. In an AI search review, record the cited page and the successful redemption separately so a change in either measure can be investigated on its own.
Use the days before November 27 to check the offers customers ask about most often and the deals with the most costly exclusions. Check again when early access closes or participating products change. Keep one record with the question, cited source, market and checkout result. That gives support a current answer and lets the next campaign owner see which external pages needed attention. Our peak-season ecommerce playbook covers the operating calendar, and our guide to discount margins helps decide whether an offer still pays for itself.
Trace one promotion from the question to the basket
Pick a high-volume offer and check it once as a shopper planning a purchase and again as a shopper ready to pay. The sources can differ. Fix the first place where the cited terms diverge from the deal a qualifying customer can actually receive.