How AI reads this page
ChatGPT, Claude, Gemini and Perplexity don't look at colours, images or animations. They receive text, headings, links and structured data. Below is exactly what they read when they open “AI answers are personalized: why you see your brand and your customers don't”.
The AI receives 1,972 words in 8 sections, 18 links and 3 blocks of structured data.
Text
The page content in Markdown, without menus or graphics. It is the format closest to how a language model reads text.
# AI answers are personalized: why you see your brand and your customers don't > AI Mode uses preferences, Gmail and Photos; ChatGPT uses memory. With Google's temporary threads (November 17, 2026) you can test brand visibility neutrally. URL: https://geosnap.ai/en/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand Language: English Version IT: https://geosnap.ai/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand Author: Rinald Sefa, CMO Geosnap · Category: Guides Publisher: Geosnap (Maind Group S.r.l.), https://geosnap.ai ## In short Yes, Google AI Mode and ChatGPT answers are personalized: they use past preferences, history and, in the US, Gmail and Google Photos too. That is why checking your brand from your own account inflates the result. Since November 17, 2026, temporary threads in AI Mode switch personalization off in every supported language: use them, together with ChatGPT's Temporary Chat, to measure what people who don't know you actually see. - Google states that a normal AI Mode thread uses your "past preferences" and shapes future recommendations; with Personal Intelligence (Labs, US, Pro and Ultra subscribers) it also reads Gmail and Google Photos. - ChatGPT remembers through two mechanisms, saved memories and chat history; free users get a lightweight version, Plus and Pro a long-term memory. In Europe chat-history referencing arrived later: check your own account. - AI Mode temporary threads (global since November 17, 2026) and ChatGPT Temporary Chat use no preferences or memory and are not saved to history: they are your neutral test session. - Protocol: neutral mode, prompts written the way a customer writes them, at least three repeats on different days per engine, comparison with your personalized session, automation beyond ten prompts. - You cannot optimize personalization: measure the neutral answer and work on sources, pages and name consistency. If your report comes from personalized sessions, the numbers are inflated. Written by the Geosnap AI agent, reviewed and approved by Rinald Sefa. Translated from the Italian original. [Read the original](https://geosnap.ai/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand) You open Google AI Mode, ask "what is the best software for running a real estate agency", and your brand is there. A prospect asks the same thing and it isn't. Neither of you is necessarily wrong: AI answers are personalized, and yours is built partly on what you searched, clicked and bought over the past months. Since November 17, 2026 Google says so openly with temporary threads in AI Mode, a mode that switches personalization off. Here is what AI Mode and ChatGPT use to tailor answers, why checking your brand "by hand" gives you a distorted picture, and how to run a neutral test. ## Are AI Mode and ChatGPT answers personalized? Yes, and two accounts can get different answers to the same question. Google states it in the [launch post for temporary threads](https://blog.google/products-and-platforms/products/search/temporary-threads-ai-mode/): a normal AI Mode thread uses your "past preferences" to tailor answers, lands in your history and shapes future recommendations. In January 2026 Google added a further layer, [Personal Intelligence in AI Mode](https://blog.google/products-and-platforms/products/search/personal-intelligence-ai-mode-search/): users who opt in connect Gmail and Google Photos, and AI Mode can suggest a product based on the brand you just bought or plan a trip by reading the hotel confirmation in your inbox. For now it is a Labs experiment, by invitation, in English, in the US, for AI Pro and AI Ultra subscribers with a personal account; it stays opt-in and can be switched off under "Search personalization". ChatGPT has done something similar for longer. According to the [official memory page](https://openai.com/index/memory-and-new-controls-for-chatgpt/), last updated in June 2025, memory has two parts: "saved memories" (things you ask it to remember or that it notes on its own) and "chat history", from which the model draws cues for its answers. Free users get a lightweight version with short-term continuity; Plus and Pro get a longer-term understanding. When the chat-history feature launched in April 2025, [TechCrunch](https://techcrunch.com/2025/04/10/openai-updates-chatgpt-to-reference-your-other-chats/) reported that the EU, the UK, Switzerland and Norway were excluded pending reviews under local rules: if your customers are in Europe, check what your own account has enabled before assuming their experience matches yours. Neither document says "memory changes the brands we recommend". But both say the answer is built on your preferences and your past, and a buying recommendation is an answer like any other. Google's own example, sneakers "based on the brand you recently bought", is exactly that. ## What do AI Mode temporary threads and ChatGPT Temporary Chat actually do? They switch personalization off for that conversation. A temporary thread in AI Mode starts from an icon in the top right corner, with no settings to change; Google describes it as "completely detached from your account": answers do not use past preferences, do not affect future personalization and are not saved to history. Close the thread and it is gone. It is [rolling out globally](https://techcrunch.com/2026/11/18/googles-ai-feature-in-search-gets-its-own-incognito-mode/) on desktop and mobile in every language AI Mode supports. Before launch some users had spotted it in testing, as [Search Engine Roundtable](https://www.seroundtable.com/google-ai-mode-temporary-thread-42539.html) reports, and Google's Robby Stein confirmed the rollout on November 17. In ChatGPT the equivalent is Temporary Chat: according to OpenAI it does not appear in history, does not use memory and is not used to train models. In the [dedicated FAQ](https://help.openai.com/en/articles/8914046-temporary-chat-faq) OpenAI adds that a copy may be kept for up to 30 days for safety reasons and that custom instructions, if enabled, still apply: for a clean test, turn them off. | | Temporary thread (AI Mode) | Temporary Chat (ChatGPT) | | --- | --- | --- | | Uses preferences or memory | No | No (but custom instructions still apply) | | Saved to history | No | No | | Shapes future answers | No | No | | Availability | Global, all AI Mode languages, since November 17, 2026 | Since 2024, all plans | | Where to start it | Top-right icon in AI Mode | Model menu at the top | ## Why is checking your brand "by hand" unreliable? Because you are the least neutral user imaginable. You have visited your own site hundreds of times, searched your own name, asked the AI to rewrite your product pages, and your Gmail is full of your suppliers' invoices. A system that adapts to "past preferences" has everything it needs to put you at the centre of the answer. The reverse holds for your competitor: if they use AI to study you, they will see you more often than the market does. Add the normal volatility of these engines: as we showed in our article on [AI answers that change every week](https://geosnap.ai/en/blog/risposte-ai-cambiano-ogni-settimana-brand-volatilita-dati-2026), the same question asked on different days returns different brands even without personalization. A single check from your own account mixes the two effects and tells you nothing reliable, neither that you are visible nor that you are not. A typical case: the owner of a travel agency in Turin asks AI Mode for "tailor-made travel agency Turin", sees her agency first and files the topic away. Her prospects, who have never opened her site, get a different list. The problem is not the answer; it is the method of checking. ## How do you test brand visibility without personalization? With a simple, repeated protocol on a neutral session. This is what we recommend: 1. **Use the neutral mode.** A temporary thread in AI Mode, a Temporary Chat in ChatGPT with custom instructions off. Gemini has had temporary chats for over a year. If you can, also use an account that has never interacted with your brand: neutral mode removes personalization, not the context of the session you are typing in. 2. **Write prompts the way a customer does.** Not "Geosnap reviews", but "how do I choose a tool to monitor my brand in AI answers". Start from the questions your sales team actually hears; if you are not sure where to begin, read [how to choose the prompts to monitor](https://geosnap.ai/en/blog/quali-prompt-monitorare-risposte-ai-come-scegliere). 3. **Repeat.** Same prompt, at least three times on different days, on every engine you care about. Note the brands cited, their position, the sources linked and the tone. One mention out of three is not "we are visible": it is a 33% frequency. 4. **Compare with your personalized session.** The gap between what you see and what the temporary thread shows is the measure of how much you were overestimating (or underestimating) your presence. 5. **Automate once you pass ten prompts.** Doing this by hand on three engines, three times a week, does not scale. Geosnap queries ChatGPT, Gemini and Google AI Mode by [simulating a real user in the browser](https://geosnap.ai/en/blog/abbiamo-abbandonato-le-api-e-simuliamo-l-utente-reale), not your session, and keeps the answer history: comparing it with your manual check is the first clue to how much personalization weighs in your case. ## What does this change for your GEO strategy? You cannot optimize personalization: you measure the neutral baseline and work on that. You have no control over what sits in your prospects' Gmail or history, and no content technique changes it. You do control three things. First, the neutral answer, the one a stranger receives: it depends on the sources the engine reads, the pages that answer the question and the mentions on third-party sites, in other words classic GEO work. Second, how consistently your name appears across your site, listings, reviews and the documents that end up in customers' inboxes: invoices, order confirmations and newsletters are the raw material Personal Intelligence uses to build "the brand you recently bought". Third, the measurement method: if your visibility report comes from personalized sessions, the numbers are inflated and every decision built on them is wrong. There is a second-order effect worth keeping in mind. If engines learn from past preferences, people who have already bought from you are more likely to see you again in their answers: the loyalty of existing customers becomes a signal that also feeds AI recommendations. One more reason to look after the post-purchase experience, not only acquisition. ## Sources - [Temporary threads are now available in AI Mode in Search](https://blog.google/products-and-platforms/products/search/temporary-threads-ai-mode/) (Google, 2026-11-17) - [Personal Intelligence in AI Mode in Search: Help that's uniquely yours](https://blog.google/products-and-platforms/products/search/personal-intelligence-ai-mode-search/) (Google, 2026-01-22) - [Google's AI feature in Search gets its own incognito mode](https://techcrunch.com/2026/11/18/googles-ai-feature-in-search-gets-its-own-incognito-mode/) (TechCrunch, 2026-11-18) - [Google AI Mode Start Temporary Thread](https://www.seroundtable.com/google-ai-mode-temporary-thread-42539.html) (Search Engine Roundtable, 2026-11-16) - [Memory and new controls for ChatGPT](https://openai.com/index/memory-and-new-controls-for-chatgpt/) (OpenAI, 2025-06-03) - [Temporary Chat FAQ](https://help.openai.com/en/articles/8914046-temporary-chat-faq) (OpenAI Help Center, 2026-11-20) - [OpenAI updates ChatGPT to reference your other chats](https://techcrunch.com/2025/04/10/openai-updates-chatgpt-to-reference-your-other-chats/) (TechCrunch, 2025-04-10) ## Frequently asked questions ### Are Google AI Mode answers personalized outside the US as well? Yes, for the part based on past preferences and history: Google says so in the temporary threads post, and the feature is rolling out in every language AI Mode supports. Personal Intelligence, which reads Gmail and Google Photos, is for now a Labs experiment in the US, in English, for AI Pro and AI Ultra subscribers only. ### Does a temporary thread show the "real" answer my customers get? It shows the answer without your personalization, which is the common baseline. Your customers have their own personalization, built on their preferences: no test will give you the exact answer one specific person sees. The temporary thread is the neutral reference for measuring your visibility in a repeatable way. ### Is ChatGPT's Temporary Chat enough for a clean test? Almost. OpenAI explains that Temporary Chat does not use memory and is not saved to history, but custom instructions still apply if you have them on: turn them off before testing. Then repeat the same prompt on different days, because answers change even without personalization. ### Can I turn personalization off permanently instead of using temporary threads? In ChatGPT, yes: memory is switched off under Settings, Personalization, Memory, and saved memories and chat-history referencing can be disabled separately. In Google you manage Personal Intelligence connections under "Search personalization". For testing, the temporary thread is still simpler because it leaves your everyday account untouched. ### Why does my competitor always show up when I check, and never for customers? If you search for them often, visit their site and ask the AI to compare them with you, your session has learned that this brand matters to you. The same mechanism that inflates your own presence when you search for yourself inflates theirs when you study them. Repeat the test from a temporary thread and compare.
Structured data
Schema.org data (JSON-LD) embedded in the page code. It tells the AI explicitly what the page is about and who published it.
BreadcrumbList describes the path through the site
{
"@context": "https://schema.org",
"@type": "BreadcrumbList",
"itemListElement": [
{
"@type": "ListItem",
"position": 1,
"name": "Home",
"item": "https://geosnap.ai/en/"
},
{
"@type": "ListItem",
"position": 2,
"name": "Blog",
"item": "https://geosnap.ai/en/blog/"
},
{
"@type": "ListItem",
"position": 3,
"name": "Guides",
"item": "https://geosnap.ai/en/blog/#guide"
},
{
"@type": "ListItem",
"position": 4,
"name": "AI answers are personalized: why you see your brand and your customers don't",
"item": "https://geosnap.ai/en/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand"
}
]
}
BlogPosting describes the article: author, dates and sources
{
"@context": "https://schema.org",
"@type": "BlogPosting",
"@id": "https://geosnap.ai/en/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand#article",
"headline": "AI answers are personalized: why you see your brand and your customers don't",
"description": "AI Mode uses preferences, Gmail and Photos; ChatGPT uses memory. With Google's temporary threads (November 17, 2026) you can test brand visibility neutrally.",
"inLanguage": "en",
"url": "https://geosnap.ai/en/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand",
"mainEntityOfPage": "https://geosnap.ai/en/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand",
"image": "https://geosnap.ai/img/blog/p2026-11-20-risposte-ai-personalizzate-thread-temporanei-en.webp",
"articleSection": "Guides",
"wordCount": 1364,
"author": {
"@type": "Person",
"name": "Rinald Sefa",
"jobTitle": "CMO",
"url": "https://geosnap.ai/en/about",
"worksFor": {
"@id": "https://geosnap.ai/#organization"
}
},
"publisher": {
"@id": "https://geosnap.ai/#organization"
},
"isAccessibleForFree": true,
"abstract": "Yes, Google AI Mode and ChatGPT answers are personalized: they use past preferences, history and, in the US, Gmail and Google Photos too. That is why checking your brand from your own account inflates the result. Since November 17, 2026, temporary threads in AI Mode switch personalization off in every supported language: use them, together with ChatGPT's Temporary Chat, to measure what people who don't know you actually see.",
"datePublished": "2026-11-20",
"dateModified": "2026-11-20",
"citation": [
"https://blog.google/products-and-platforms/products/search/temporary-threads-ai-mode/",
"https://blog.google/products-and-platforms/products/search/personal-intelligence-ai-mode-search/",
"https://techcrunch.com/2026/11/18/googles-ai-feature-in-search-gets-its-own-incognito-mode/",
"https://www.seroundtable.com/google-ai-mode-temporary-thread-42539.html",
"https://openai.com/index/memory-and-new-controls-for-chatgpt/",
"https://help.openai.com/en/articles/8914046-temporary-chat-faq",
"https://techcrunch.com/2025/04/10/openai-updates-chatgpt-to-reference-your-other-chats/"
],
"translationOfWork": {
"@id": "https://geosnap.ai/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand#article"
}
}
FAQPage describes the frequently asked questions
{
"@context": "https://schema.org",
"@type": "FAQPage",
"@id": "https://geosnap.ai/en/blog/risposte-ai-personalizzate-ai-mode-thread-temporanei-come-testare-visibilita-brand#faq",
"mainEntity": [
{
"@type": "Question",
"name": "Are Google AI Mode answers personalized outside the US as well?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes, for the part based on past preferences and history: Google says so in the temporary threads post, and the feature is rolling out in every language AI Mode supports. Personal Intelligence, which reads Gmail and Google Photos, is for now a Labs experiment in the US, in English, for AI Pro and AI Ultra subscribers only."
}
},
{
"@type": "Question",
"name": "Does a temporary thread show the \"real\" answer my customers get?",
"acceptedAnswer": {
"@type": "Answer",
"text": "It shows the answer without your personalization, which is the common baseline. Your customers have their own personalization, built on their preferences: no test will give you the exact answer one specific person sees. The temporary thread is the neutral reference for measuring your visibility in a repeatable way."
}
},
{
"@type": "Question",
"name": "Is ChatGPT's Temporary Chat enough for a clean test?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Almost. OpenAI explains that Temporary Chat does not use memory and is not saved to history, but custom instructions still apply if you have them on: turn them off before testing. Then repeat the same prompt on different days, because answers change even without personalization."
}
},
{
"@type": "Question",
"name": "Can I turn personalization off permanently instead of using temporary threads?",
"acceptedAnswer": {
"@type": "Answer",
"text": "In ChatGPT, yes: memory is switched off under Settings, Personalization, Memory, and saved memories and chat-history referencing can be disabled separately. In Google you manage Personal Intelligence connections under \"Search personalization\". For testing, the temporary thread is still simpler because it leaves your everyday account untouched."
}
},
{
"@type": "Question",
"name": "Why does my competitor always show up when I check, and never for customers?",
"acceptedAnswer": {
"@type": "Answer",
"text": "If you search for them often, visit their site and ask the AI to compare them with you, your session has learned that this brand matters to you. The same mechanism that inflates your own presence when you search for yourself inflates theirs when you study them. Repeat the test from a temporary thread and compare."
}
}
]
}
Metadata
The information in the page header, which crawlers and engines read before the content.