How to track ChatGPT and AI traffic in Google Analytics 4
GA4's new AI Assistant channel misses some assistants and app clicks land in Direct. The custom channel group, the regex, and what GA4 still can't see.
Playbooks, research and teardowns on getting AI engines to recommend your brand - written by the team building the category.
GEO - generative engine optimization, aka AI GEO - is the practice of getting your brand named and cited inside AI answers. What it is, how it differs from SEO, and a step-by-step framework to get ChatGPT, Claude, Gemini and Perplexity recommending your brand.
GA4's new AI Assistant channel misses some assistants and app clicks land in Direct. The custom channel group, the regex, and what GA4 still can't see.
How to find GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot and PerplexityBot in your logs, verify their IPs, and read training versus live-fetch patterns.
Keyword tools can't show what buyers ask ChatGPT. Mine sales calls, tickets, Search Console and Reddit, sort by intent, and pick the prompts worth tracking.
How SaaS products get named when buyers ask ChatGPT, Claude or Perplexity for the best software: category clarity, comparison pages, G2, crawlable pricing.
How Microsoft Copilot grounds answers in Bing, how it differs from ChatGPT search, and how Bing Webmaster Tools, IndexNow and AI Performance get you cited.
A mention is your name in an AI answer; a citation is your page linked as a source. How engines show each, share of voice vs citation share, what moves both.
What ChatGPT, Gemini and Perplexity recommended - and cited - across law firms, clinics, ecommerce, hotels and six more industries in the US, UK, Canada, Spain and Turkey. 79% of cited domains appeared exactly once, ChatGPT and Gemini named the same #1 brand in 8.5% of questions, and Wikipedia made up 9.5% of ChatGPT's citations but none of Perplexity's.
"Best meditation apps", "app to plan a trip" - questions that used to be store searches are now asked in ChatGPT, and the answer is a shortlist of three to five names. Here's how assistants actually pick those names - listing signals, reputation bars, citable sources - and the six-step playbook for getting your app into the shortlist, with a free checker that grades where you stand today.
Your app is live, rated well by the people who found it - and invisible when someone asks an AI assistant for the best app in your category. The cause is almost always mechanical: a handful of listing, reputation and web-presence signals that assistants read and your listing fails. Here are all 20, grouped into the five categories that decide AI recommendability, and the order to fix them in.
App Store Optimization was built for one reader: the store ranking algorithm. Now your listing has a second reader - the AI assistants people ask for app recommendations - and it grades on different criteria. What ASO is, what GEO changes about it, where the two disciplines pull in opposite directions, and the listing playbook that satisfies both readers at once.
An AI assistant is telling your buyers something false about your company - a wrong owner, a dead product, a price from 2023 - and there's no edit button. Here's the playbook that actually works: trace the sources the engines read, fix them in the right order, file with each engine's official channel, and verify the correction sticks. Plus the honest part nobody says: how long each layer takes.
"Tools for AI Mode" hides two different questions: how do I use Google's AI Mode well, and how do I know whether it mentions my brand? A quick answer to the first, and an honest map of the second - what Search Console shows (and hides), what SEO suites actually capture, what AI visibility trackers measure instead, and which of it is free. (We make a tool in the last category, and we say so.)
People keep googling strings like turn0product7 and turn0search0 - and finding storefront junk instead of an answer. Here's the real one: they're ChatGPT's internal citation tokens, the invisible markers that place product cards and source badges inside an answer. Where they come from (we've captured them in the raw response stream), why they leak into copied text, and what their existence tells brands about how AI shopping answers are assembled.
AEO - answer engine optimization - is one of the fastest-rising terms in search marketing, and most explanations make it sound more novel than it is. Here's the plain version: what AEO actually means, where the term came from, how it relates to SEO and GEO, and the handful of practices that genuinely improve your odds of being the answer instead of a result.
"AI SEO tools" is one label hiding three different jobs: AI that helps you write and optimize content, AI features inside classic SEO suites, and tools that track your visibility inside AI answers. An honest map of all three - what each category actually does, who it's for, and how to avoid buying the wrong one. (We make a tool in the third category, and we say so.)
Thousands of owners search for the ChatGPT equivalent of a Google Business Profile. It doesn't exist - there's no submission form, no listing to claim, no fee to pay. But ChatGPT does decide which businesses to recommend, and you can influence that. Here's where its answers actually come from, the five things that function as your 'listing,' and how to check whether it worked.
LLM SEO (or LLMO, or GEO - the names haven't settled) is the work of getting your brand found, trusted and cited by large language models instead of just ranked by Google. This is the plain-English version for people who already know SEO: how LLMs actually choose what to say, which of your existing skills transfer directly, the handful of genuinely new moves, and the tactics you can safely ignore.
Getting named by ChatGPT is only half the battle - 'a budget option with limited features' is a mention that loses the deal. Tone is now a first-class metric: whether the engines frame you as the leader, the alternative, or the one to be careful with. Here's where AI sentiment actually comes from, how to measure it without fooling yourself, and the work that turns a lukewarm description into an endorsement.
AI Mode is Google's conversational search - a full AI answer instead of a results page, now sitting one tab away from every Google search. This is how it actually assembles an answer (and why it's not the same placement as an AI Overview), what that means for whether your brand gets named, and an honest look at the tools that can track your visibility in it today.
SEO gets you ranked; AI search optimization - some call it AI search SEO - gets you recommended. When someone asks ChatGPT, Gemini or Perplexity for the best option in your category, either you're in the shortlist or a competitor is. This is the practical playbook: how AI search actually assembles an answer, the handful of moves that measurably improve your odds of being named, and how to tell whether any of it is working.
The AI Overview is the paragraph Google now writes above the blue links - and for a growing share of searches it's the only thing people read. Ranking #1 no longer guarantees you're in it. Here's what an AI Overview actually is, how it picks the sources it summarizes, why being on page one isn't enough, and the concrete work that gets your brand named in the answer instead of buried under it.
Ask the same question through the ChatGPT API and through the app your customers actually open, and you can get two different answers - one from the model's frozen memory, one from a live web search with real citations. Most GEO tools only ever measure the first. Here's exactly why the two disagree, why the gap decides where AI sends customers, and what it really costs to capture the answer people see.
Ask ChatGPT for a product and sometimes you get plain text, sometimes a rich carousel with prices and ratings. We captured and decoded a real streamed shopping answer to show the machinery underneath: how it classifies the question, where the products actually come from, what the model 'sees' about each one, and what it takes to get your product on that shelf.
Every serious AI visibility tracker, honestly compared - Profound, Peec, Otterly, Scrunch, AthenaHQ, Ahrefs, Semrush, SE Visible, HubSpot, BrightEdge and more. Entry prices checked in October 2026, the engines each covers, who it's for, pros and cons, and where it stops short. (We make one of them, and we say so.)
Before AI can recommend your brand, it has to reach, read and cite your site - and that's measurable. Meet our free, open-source Chrome extension that scores any page's AI readiness in one click, per engine.
AI shopping is here - ChatGPT recommends products, compares them and even checks out. How ecommerce brands get surfaced: product data, reviews, structured feeds, and why it's relevance, not ad spend.
When someone asks ChatGPT or Google's AI for 'the best [X] near me', who gets named? How local businesses win AI recommendations - reviews, business profiles, consistent NAP, and the sites AI trusts for local.
AI visibility is the new line item clients are asking about. How agencies productise GEO - scoping it, reporting it, pricing it, and managing many brands across engines without drowning in manual checks.
Wikipedia is one of the most-cited sources in AI answers - and a backbone of what models 'know'. How to earn a legitimate Wikipedia presence (and why trying to game it backfires).
AI engines lean on review and comparison sites to decide which brands to name - yet most companies treat reviews as a support metric. How to turn G2, Capterra and reviews into AI recommendations.
Video keeps showing up as a source in AI answers. Why engines cite YouTube so often, what they actually read from it, and how to make your video content quotable to a model that can't watch it.
Payment giants are building rails for AI agents that discover, decide and check out for people. What agentic commerce means for brands - and why being the answer the agent trusts is the new shelf position.
ChatGPT now recommends products, compares them and checks out in-chat - ranked on relevance, not ad spend. A practical playbook for getting your products onto that shelf while it's still early.
An honest look at our own scan data: across 2,209 answers, only 39% cited a source - and which engine you ask changes everything. Plus the recognisable domains that keep showing up, and how fragmented the rest is.
Raw mentions is a vanity metric. The numbers that tell you whether AI engines recommend your brand - visibility rate, share of voice, answer position, citation share, sentiment and engine coverage - and how to read each one.
A practical setup for monitoring whether ChatGPT, Gemini, Perplexity, Claude and Grok recommend you - which questions to track, which engines to cover, what cadence, and when eyeballing it stops working.
The numbers behind the shift from Google to AI answers - engine reach, click-through, how buyers research, and how ChatGPT picks and cites the brands it names, including the August 2026 changes. Every figure links to its source, and vendor research is labeled.
ChatGPT, Gemini, AI Overviews, Perplexity, Claude, Copilot, Grok and the rest - who's behind each, whether it answers from the live web or from memory, and how every one of them decides which brands to name.
Short answer: not the recommendation. ChatGPT now tests labeled ads and Google sells ads in AI Mode, but neither buys you a place in the answer. What's actually for sale in October 2026 - and what the brands that get recommended did instead.
Weeks for some answers, months for others - here's why. Which AI answers you can move this month, which take a quarter, and how to tell which kind you're facing.
Yes - but its job changed. Why your Google rank no longer guarantees an AI mention, what SEO still feeds into every engine, and where to put this year's effort.
The answer that names you today can drop you next week. The three things that make AI answers drift - and a sane monitoring cadence that catches it without obsessing.
No tricks, no "prompt hacks" - the five things that actually move a brand into ChatGPT's answers, in plain language, in the order we'd do them.
A do-it-yourself audit: which questions to ask, how to ask them cleanly, what counts as "visible", and the mistakes that make self-testing misleading.
Where llms.txt stands in October 2026: Google says its search ignores the file, and Ahrefs found 97% of ~38,000 llms.txt files got no requests in a month. What the file does, what it can't do, and the 20-minute version if you ship one anyway.
The acronym soup, sorted: what GEO (generative engine optimization), AEO (answer engine optimization) and their synonyms each cover, where they overlap, and why the distinctions matter less than the two real disciplines underneath them.
Each engine builds its answer differently - training data, retrieval, citations. The mechanics explain why your visibility differs per engine, and what to do about each one.
Why "X vs Y" content is the single highest-leverage asset for AI visibility - and how to structure one that models actually cite instead of skim past.
Community threads punch far above their weight in both training data and retrieval. Why that happened, what the August 2026 ChatGPT citation data does (and doesn't) show, and how to participate without astroturfing.
A worked example, week by week: the exact sequence we'd run to take an unknown SaaS from zero AI mentions to consistent ChatGPT and Perplexity recommendations - baseline, citable core, third-party presence, re-measure.
Publishers block them; brands mostly shouldn't. What each crawler does, what blocking actually prevents, and a sane robots.txt policy for a SaaS.
The most transparent engine is the best one to learn from: how retrieval, ranking and citation work in Perplexity, and what reliably gets a page into the answer.
AI engines quote pricing pages constantly, but only when they can parse them. The patterns that make yours quotable, and the ones that make models guess.
You're on page one of Google, customers love you - and ChatGPT draws a blank. The three reasons this happens, and which one is yours.
Somebody is winning the answers in your category right now. A fifteen-minute exercise to find out who - and what to do if it isn't you.
Position, frequency, sentiment, sources: the components that make a visibility score honest, and the shortcuts that make one meaningless.
Structured data is back - for a new reason. It's how models find, parse and trust your facts. The five types worth shipping first, with markup examples.
One practical essay on AI visibility every other week. No fluff, unsubscribe anytime.