Black Friday's shortlist is written by AI now — and it changes every year
Profound tracked 279,741 AI responses across 25 retail categories: AT&T's phone visibility rose 645%, Wayfair's furniture visibility hit 85%.
- Profound analyzed 279,741 AI responses and over 1.2 million citation events across 25 retail categories. Self-reported m:…6175
- AT&T's visibility in US phone recommendations grew 645 percent year over year, and Wayfair's visibility in US furniture recommendations grew from 64.4 percent to 85.0 percent. Self-reported m:…6175
- 51 percent of B2B decision-makers start research with an AI chatbot more often than with Google, according to G2 survey data cited by Pixis. m:…5805
- EnterAI guarantees that at least 10 of 40 tracked keywords will be mentioned in AI responses, extending work six months free if it misses. Self-reported m:…0380
The shortlist moves
Profound's Black Friday study of AI search analyzed 279,741 AI responses and over 1.2 million citation events across 25 retail categories, and its headline finding is volatility. Brand visibility in AI search shifts substantially year to year: AT&T's visibility in US phone recommendations grew 645 percent year over year, and Wayfair's visibility in US furniture recommendations grew from 64.4 percent to 85.0 percent. Amazon leads the combined four-market leaderboard at 48.4 percent visibility, but Ulta and Chewy retained first place in their respective US categories while increasing visibility, which suggests the leaders are not standing still either. Commercial conversations in ChatGPT increased year over year, with spikes around Black Friday and Christmas, so the shortlist that matters is the one the AI writes in December, not the one it wrote in January. m:…6175
The Black Friday shortlist can change substantially from one year to the next, according to Profound, and new content can enter AI answers while the campaign is underway, as seen with a Times Union sale calendar. That is a different game from traditional search, where rankings move slowly and a page built in November pays off in March. In AI search, the answer is assembled at query time, and a fresh, well-cited page can show up the same week. m:…6175
Who wins, and why
The winners are not always the biggest brands. Local specialists lead in their respective categories, including Currys in UK appliances, Ocarat in French jewelry and Bergfreunde in German outdoors, by organizing content around specific regional purchase decisions and product conditions. Third-party publisher coverage remains a critical driver of AI visibility: Wayfair's placement in a Black Friday furniture guide contributed to its visibility growth. The AI does not rank your site; it ranks what trusted sources say about you, and then cites them. m:…6175
The B2B side tells the same story with different stakes. G2 survey data cited by Pixis indicates 51 percent of B2B decision-makers start research with an AI chatbot more often than with Google, and comparing vendor strengths and weaknesses is the leading use case for AI-assisted software research. Pixis warns that AI-generated shortlists can include inaccurate information about product tiers or integrations, and argues B2B SaaS companies should prioritize the recurring buying questions that block qualified evaluations, and have product owners confirm integration behaviour and security statements. One of its clients, Confido Health, saw a 52 percent increase in organic clicks from May to July, a threefold increase in Google impressions, and AI citations rising from two in June to fifteen in August. m:…5805
The industry selling the fix
An industry has formed to sell the fix, and it is starting to offer guarantees. EnterAI, a Taiwan-based agency that launched its dedicated GEO offering in 2025, uses a free website health check as the main entry point for new projects, evaluating seven areas from crawlability to freshness. It guarantees that at least 10 of 40 tracked keywords will be mentioned in AI responses, and if it misses its target it extends the work by six months at no additional charge, or offers a 30 percent refund if the target remains unmet after a year. Its founder argues that traditional search exposure does not indicate AI recommendation frequency, which is the pitch in one sentence: being found and being recommended are now different jobs. m:…0380
Buyers are also getting a buyer's guide. AirOps frames the AEO platform decision around six questions, from revenue connection to enterprise governance, and recommends a 30 to 60 day pilot with written decision gates before signing an annual commitment. It notes that 90 percent of evaluation answers now include pre-purchase checks, and that Google AI Mode citation rates for new pages can reach 56 percent by day 7. Meanwhile the infrastructure is localizing: Exa has launched search services in the European Union, processing queries and storing data in AWS Stockholm, with EU API keys region-locked and requests that cannot be served in the EU refused rather than failing over to the US. The EU endpoint reduces latency by over 100 milliseconds and shows less than a three percentage point difference in quality versus the US. AI search is becoming a regulated, regional product, and the brands that win it will be the ones cited by others, not the ones shouting loudest. m:…8526 m:…0510
What is still unresolved
- Whether Profound's visibility figures reflect actual purchases, or only appearances in AI answers.
- Whether GEO guarantees like EnterAI's survive once AI models change how they cite sources.
- How much of B2B shortlist accuracy depends on vendors keeping their own pages current.
Provenance by Synorb. Every claim links to its source. This story was written by an agent from the Manifests listed below. Read the methodology and limitations, or send a correction.
Free to reuse, including commercially. Credit “Hanging Context, with provenance by Synorb” and link to this story. Keep the cited Manifest ids. Reuse terms v1.