Monday, September 21, 2026
8 stories · 2 topics · 6 outlets
Answer Engines
6With ChatGPT and Claude, the brand is no longer an experience, it is a channel
The article says ChatGPT and Claude turn a brand into a channel rather than an experience: the brand keeps its content but loses control of how it is staged, because the assistant frames the comparison. It argues that what separates brands inside those answers is real product experience rather than marketing presentation. Marketers should expect their products to be judged on substance found in third-party and review content, since they no longer set the scene themselves.
Cited more than 15,000 times by Copilot, and still no paying customers
A case study circulating among independent developers describes PeriodHub, a women's health app priced at $19, that was cited more than 15,000 times by Bing and Copilot and still signed up no paying customers. The write-up traces the gap to intent: the citations came from informational questions that Copilot answered inside the chat, so most readers never clicked through, and the people asking about symptoms were not shopping for an app. Marketers should treat AI citation counts as a reach metric rather than a buying signal, and check whether the prompts they appear in sit near a purchase decision.
Research Shows Google AI Mode Sends Less Clicks & Is A Poor User Experience via @sejournal, @martinibuster
Search Engine Journal reports on research it says found Google's AI Mode sends fewer clicks to websites and gives users a poor experience. The outlet frames the findings as a challenge to Google's own statements about AI search. Marketers should check their own referral numbers from Google before assuming AI Mode will replace the clicks they used to get from classic results.
The AI Search Optimization Checklist – With Worksheet [Updated Sep 2026]
Aleyda Solis updated her AI search optimization checklist and worksheet in September 2026. The checklist sets out a process: decide which AI-assisted journeys matter commercially, find where a brand is shown, recommended, linked or described inaccurately, trace which owned, earned and shared sources drive those outcomes, and measure progress by platform and confidence level. Marketers can use it as the audit step before committing budget to an AI visibility tool or agency.
How To Build Local Pages AI Systems Can Find And Trust via @sejournal, @lorenbaker
Search Engine Journal published rules for building local service pages that AI systems can find, rank and cite, drawn from its SEJ Live session. The guidance covers how AI systems read passages and alt text and how they repeat numbers a page publishes. Marketers running local pages should structure text for passage-level reading and keep the facts they publish consistent, because those are the parts answer engines quote.
Agent-to-Agent Marketing
2Instacart launches custom AI shopping assistants for retailers and its own
Instacart launched a white-label Cart Assistant that retailers can put on their own sites and apps, and Clementine, its own AI grocery shopping assistant for its US and Canada marketplaces. The assistants turn conversational requests such as a week of lunches for kids into a cart tied to the chosen store's inventory; Instacart's chief commercial officer told Modern Retail the company does not share the same level of embedded-assistant data with OpenAI. Marketers selling grocery or retail products should expect more AI assistants to build carts from store inventory, and should make product data readable to those assistants if they want to be included.
Agentic commerce: the real battle happens before payment
Journal du Net argues that in agentic commerce the decisive work happens before payment, in how a product is described, priced and stocked for a shopping agent, rather than in the autonomous payment protocols that get the attention. The piece notes Walmart has withdrawn from its ChatGPT integration, which it reads as a sign that the expected order of adoption has not held. Marketers should put their effort into agent-readable product data and their standing inside assistants, because both come before any checkout protocol decision.