MPWhat The Shift Is
AI search is a new category, and it is moving faster than anything search has done in twenty years. The platforms change how they choose sources without announcing it. Google ships updates that reach AI Overviews and blue links at the same time. New measurement tools appear monthly, and the research about them contradicts the research from the month before. A business owner cannot follow all of that, and should not have to.
So every Monday morning we read what changed across Google AI Overviews, ChatGPT, Perplexity, Copilot and the studies around them, and we write down the handful of things that matter to a business trying to get found, understood and recommended. Plain English first. The detailed brief underneath for the marketing people. If you have ever wished somebody would just tell you what changed and what to do about it, without a webinar, this is that.
Two labels do the honest work in every issue. Confirmed means a platform documented it or the data is public. Our read means it is our interpretation, which is worth exactly what you paid for it, and we say so.
In Plain English: What Changed and What To Do
Google says a human should fact-check every AI-written page
On October 1, Google updated its guidance on AI-generated content. It now says it is critical to manually fact-check and review all AI-generated content for accuracy and trustworthiness before publishing, and that the review covers page titles, meta descriptions, structured data and image alt text too. The same day, Google's helpful content page gained a definition of main content and four qualities its quality raters look for: effort, originality, talent or skill, and accuracy. Meanwhile the spam update Google started on September 24 had not been marked complete as of this morning.
This is Google putting in writing what it rewards, and Google's AI Overviews and AI Mode draw from the same pages and quality rules as regular search. Google is not banning AI-assisted writing. It is saying the machine guesses and a person has to check. In Google's own words, generative models do not retrieve facts, they predict a likely sequence of words. Anyone who has asked one for a phone number has met that sentence in person.
Pick your five most important pages and have a person who knows the business read each one for accuracy, including the page title, the description that shows in search, and any FAQ or schema markup. Fix anything wrong or generic, and add something only you could say: a real process, a real example, a real answer to the question customers actually ask. If traffic moves before the spam update finishes, note it and wait. A dip during an update is weather, not a verdict.
AI Overviews moved onto brand-name searches
DemandSphere, which tracks search results daily, published data on October 1 showing AI Overviews appeared on about 26 percent of the brand-name searches it tracks on September 1. By September 27 that was 90 percent, and it stayed above 80 percent through September 29. Most of the jump came in the last two weeks of the month. Google has not announced a change.
When someone searches your business by name, they are already close to buying. If an AI summary now sits on that page, it is part of your first impression, and you did not write it. DemandSphere counted whether an overview appeared, not whether it was accurate or mentioned the brand, so nobody yet knows how often these summaries get a business right. That part is worth checking yourself.
Search your business name, and your name plus words like reviews, pricing and near me, in a private browser window. Read any AI Overview the way a customer would. Is it accurate? Is it current? Which sites does it cite? If something is wrong, fix the source it is reading, usually your own site, your Google Business Profile or a directory listing, and use the feedback option on the overview.
ChatGPT lets shoppers try on clothes before they buy
On October 1, OpenAI added a Try on button to clothing and accessory listings in ChatGPT. A shopper uploads a photo and ChatGPT shows how the item might look on them. Shoppers can also save products as favorites in their ChatGPT library.
Each feature like this keeps a shopper inside ChatGPT for one more step of the purchase. The more of the decision happens in the chat, the more it matters what ChatGPT knows about your products before anyone reaches your website. For anyone selling physical products, the listing ChatGPT shows is becoming a storefront you do not get to decorate.
If you sell products, look them up in ChatGPT and check what comes back: names, prices, photos and availability. Keep the product information on your own site complete and consistent, with clear photos, accurate prices and current stock. If you sell services, there is nothing to change this week except noticing which way the wind is blowing.
Ranking yourself number one is losing its charm
On October 4, search consultant Lily Ray published research on 100 searches for the best B2B software. Compared with June, Google's AI Overviews cited 38 percent fewer best-of lists written by a company that ranks itself first. When the AI did cite one, the company that put itself first was left out of the actual recommendation 83 percent of the time, up from 69 percent in June.
A list of the best roofers in town, written by a roofer who put himself first, is not a review. It is an ad wearing a cardigan, and Google's AI appears to be noticing. The sting in the tail: in 70 percent of cases the self-ranking list helped a competitor get recommended instead. This is one consultant's sample in one industry, so treat it as a strong signal, not a law.
If your site has a best-of page with you at number one, rewrite it as an honest comparison with clear criteria, or retire it. Put that effort into getting named by people who are not you: customer reviews, local press, industry directories, partners and trade associations.
Your To-Do List This Week
- Put a human on the AI copy.Have someone who knows the business fact-check your five most important pages this week, titles, descriptions and schema included, and add one thing to each that only you could say.
- Search your own name.Read the AI Overview on your business name the way a stranger would, write down what it gets wrong and which sources it cites, and fix those sources first.
- Retire the self-ranking list.Replace any best-of page that puts you first with an honest comparison, and spend that effort earning reviews and mentions from other people.
The Full Brief
The detailed version, for marketing directors and anyone who has to defend a number in a meeting. Every item carries its source. Where we are interpreting, it says so.
Google tightens its AI content guidance and defines main content
On October 1, 2026, Google updated its guidance on using generative AI content. It now reads: "It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing." The review also applies to metadata such as title elements, meta descriptions, structured data and image alt text, which can appear in Search results. Google explains that generative models "don't retrieve facts, but predict a likely sequence of words based on their training data." Google's changelog says the update adds information from the Search Quality Rater Guidelines. Google Search Central: Using generative AI content
Google's helpful content documentation, updated the same day, defines main content as "any part of the webpage that directly helps the page achieve its purpose" and lists the attributes quality raters assess: effort, originality, talent or skill, and accuracy. For topics that could significantly affect people's lives, content "must be highly accurate and consistent with established expert consensus." Google Search Central: Creating helpful, reliable, people-first content
The September 2026 spam update began on September 24, and Google said the rollout "may take up to two weeks to complete." As of the morning of October 5, the incident page showed no completion notice. Google Search Status Dashboard
None of this bans AI-assisted content; using AI to mass-produce pages that add no value was already covered by the scaled content abuse policy. What changed is the emphasis. Google now calls human review critical and extends it to the titles, descriptions and markup that feed snippets and AI answers. Landing in the middle of a spam rollout, it reads as a signal of where enforcement is pointed, though Google has not connected the two. Ranking trackers reported a fresh wave of volatility around September 30; Google has not confirmed any phases.
Add a named human reviewer to every AI-assisted publishing workflow, covering titles, meta descriptions, FAQ copy and schema output. Audit high-value pages against effort, originality, talent and accuracy. Keep September 24 annotated in reporting and hold attribution until Google marks the spam update complete.
AI Overviews now appear on most tracked branded searches
DemandSphere reported on October 1 that AI Overview presence on its tracked branded keywords rose from 26.12 percent on September 1 to a peak of 90.48 percent on September 27, then read 80.23 and 82.06 percent on September 28 and 29. Measurements were taken once per day per keyword across all markets and devices, and an overview counted as present whether or not the brand itself was cited. DemandSphere notes Google has not announced a change. DemandSphere: AI Overviews on branded queries tripled in September
Presence is not accuracy, prominence or click loss, and this dataset measures none of those. Google may also dial it back. But branded searches are the highest intent queries a business gets, and an AI summary now frames many of them. Brand monitoring has to include what Google's summary says, which sources it leans on, and where it sits on the page.
Add branded queries (name, name plus reviews, pricing and location) to the weekly AI visibility panel. Record whether an AI Overview appears, whether it is accurate and which sources it cites. Watch branded click-through in Search Console, and correct inaccurate source pages first.
ChatGPT shopping adds virtual try-on
OpenAI's ChatGPT release notes for October 1, 2026 introduce new ways to shop: "A 'Try on' button will appear on product listings for clothes and accessories." Users upload a photo to see how an item looks before buying, and can save favorite products to their library. OpenAI: ChatGPT release notes
This is a conversion feature for apparel and accessories, not a change to how ChatGPT chooses what to show. It continues a clear pattern: each release moves another step of the purchase into ChatGPT, which raises the value of accurate, complete product data that the assistant can display without a click.
For retail clients, spot-check how key products appear in ChatGPT shopping results, including images, prices and availability, and keep on-site product data consistent. No change for service businesses.
Self-ranking listicles are losing ground in AI Overviews
On October 4, Lily Ray published an analysis of 100 "best B2B software" queries measured in June and September 2026. Google's AI Overviews cited 38 percent fewer self-promotional listicles in September than in June. Where one was cited, the publishing brand was left out of the recommendation 83 percent of the time, up from 69 percent, and in 70 percent of cases the list helped competitors get recommended while its publisher was omitted. Lily Ray on Substack
One practitioner, one vertical, one AI surface, and limited published methodology, so it is directional rather than definitive. It is consistent with Google's new emphasis on originality and trustworthiness. The tactic was always a little transparent; it now appears to be doing the competition a favor.
Inventory self-ranking comparison pages. Rewrite them as transparent comparisons with stated criteria and disclosure, or retire them. Move the effort to earning third-party mentions: reviews, press, directories and partner pages.
Questions People Ask
What did Google change about AI-generated content in October 2026?
On October 1, 2026, Google updated its guidance to say it is critical to manually fact-check and review all AI-generated content for accuracy and trustworthiness before publishing. The review also covers titles, meta descriptions, structured data and image alt text. The same day, Google's helpful content page added a definition of main content and the qualities raters look for: effort, originality, talent or skill, and accuracy.
Does Google penalize content written with AI?
Not for using AI. Google's guidance allows AI-assisted content that is accurate and useful. What it warns against is using AI to mass-produce pages that add no value, which falls under its scaled content abuse spam policy. Since October 1, Google also says it is critical to have a person fact-check and review AI-generated content before it goes live.
Why is there an AI Overview when people search my business name?
Google began showing AI Overviews on far more brand-name searches in September 2026. DemandSphere's tracking went from about 26 percent of branded searches on September 1 to more than 80 percent by the end of the month. Google has not announced the change, and the overview may not be accurate, so search your own name and read it the way a customer would.
How do I fix an AI Overview that gets my business wrong?
You cannot edit an AI Overview directly. Look at the sources it cites and correct those first, usually your own website, your Google Business Profile or a directory listing. Use the feedback option on the overview to report the error, then check again in a week or two.
Should I publish a best-of list that ranks my own company first?
It is a weak bet. New research from Lily Ray found Google's AI Overviews cited these self-ranking lists 38 percent less often in September than in June, and when they did, the company that ranked itself first was left out of the recommendation 83 percent of the time. An honest comparison with clear criteria, plus reviews and mentions from other people, does more work.
What is The Shift?
The Shift is Agency AEO's weekly brief on AI search news. Every Monday morning we review what changed across Google AI Overviews, ChatGPT, Perplexity, Copilot and the research around them, explain in plain English what it means for a business, and separate what is confirmed from what is our read. It exists because AI search is a new category that changes weekly.
Sources & Further Reading
- Google Search Central: Using generative AI content (updated October 1, 2026)
- Google Search Central: Creating helpful, reliable, people-first content
- Google Search Central: Documentation updates
- Google Search Status Dashboard: September 2026 spam update
- DemandSphere: AI Overviews on branded queries tripled in September
- OpenAI: ChatGPT release notes
- Lily Ray: Prediction: the next massive Google update is around the corner