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.
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.
If you only ship one piece of content for AI visibility this quarter, make it a comparison page. Not a feature grid built for a sales deck - a genuine, opinionated "you vs. the alternative" page written for a buyer who is deciding.
The leverage shows up twice - once when the engine retrieves the page, again when it picks a sentence to quote. The structure below is built for both.
Think about what an AI engine is actually asked: "Notion or Obsidian for a small team?", "best alternative to X", "is Y worth it over Z?". These are comparison questions. When an engine searches for material to answer them, a page whose title and structure mirror the question is the easiest possible retrieval match - and once retrieved, a page with clear verdicts is the easiest possible source to quote.
The model can't be persuaded - it can only quote you. So write the sentence you'd want repeated back to a buyer.
Models lift summaries. Open with two or three sentences that actually answer the question - "Choose A if … choose B if …". Pages that bury the conclusion under 1,200 words of preamble give the model nothing to grab.
Conceding real weaknesses feels like sales malpractice. It's also what gets you quoted: engines cross-check sources against each other, and a one-sided page disagrees with the rest of the evidence and gets discounted. A page that says "X is genuinely better for offline use; we win on collaboration" matches reality and earns the quote.
Add FAQ and Product schema so the facts are machine-readable - we cover the details in 5 schema types that help AI engines cite you.
Don't guess. Ask the engines your buyers' questions and note which competitor keeps appearing next to you (or instead of you). That pairing is your first page. The competitor whose name the models already associate with your category is the comparison the models already want to make - give them the source for it. If you sell software, comparison pages are one part of a bigger plan: GEO for B2B SaaS covers category definition, review sites and crawlable pricing too.
A comparison page is the rare asset that serves the skeptical buyer and the citation-hungry model with the same words. Run a free audit to find which comparison you're losing right now - the competitor column usually names your first page for you.
Because the questions buyers ask AI engines are themselves comparisons - "Notion or Obsidian for a small team?", "best alternative to X". A page whose title and structure mirror that question is the easiest possible retrieval match, and once retrieved, a page with clear verdicts is the easiest possible source to quote. The leverage shows up twice: when the engine finds the page, and again when it picks a sentence to repeat.
Lead with the verdict in two or three sentences that actually answer the question, since models lift summaries and ignore pages that bury the conclusion. Be honest about where you lose, because engines cross-check sources and discount one-sided pages that disagree with the rest of the evidence. Then use comparable, concrete facts - a real HTML comparison table, specific numbers over adjectives, a short FAQ at the bottom - and mark it up with FAQ and Product schema.
Write the one against the competitor the models already pair with you. Don't guess - ask the engines your buyers' questions and note which competitor keeps appearing next to you or instead of you. That pairing is the comparison the models already want to make, so giving them the source for it is your highest-leverage first page.
Put this guide into practice - get your free visibility score in minutes.
