What answer engines publish about getting your pages cited

Common belief: The site needs seven specific page types before an answer engine will recommend it.

The correction: The only stated requirement is a page Google can index and show with a snippet

That list, capability, application, industry, material, certification, comparison and proof, comes from one marketing agency's blog post about manufacturing websites, repeated since by people quoting the post. Google's own documentation names no page type at all.

Google publishes one requirement instead. For a page to appear as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to appear in Google Search with a snippet, and Google states there are no additional technical requirements beyond that.

That puts the question on the pages you already have rather than pages somebody would build. Are they indexed, and is anything on your side keeping them out of Search or out of a snippet? Both readings take an afternoon. The seven-page build takes whatever the quote says it takes.

Common belief: A page gets used because it answers the exact question the person typed.

The correction: The query that pulls your page in is not the question someone typed

Google describes query fan-out as generating concurrent related queries to fetch additional results that address subtopics of the user's question. One typed question becomes several queries running at once, and you see none of them.

Google then grounds the answer, which it defines as using core Search ranking systems to retrieve relevant, up-to-date pages from the Search index, then using those pages to generate a response with clickable supporting links. Retrieval runs on the ranking systems that already order ordinary Search results.

Microsoft describes the same gap from the Copilot side: the search query Copilot generates and sends to Bing is different from the full user prompt.

Your page has to be findable for a subtopic query you never see, fetched by ranking systems already in operation. Writing a page to mirror the phrasing of a prompt aims it at a string that may never reach the index.

Common belief: Adding an llms.txt file is the entry ticket for AI answers.

The correction: Google says it does not use llms.txt for Search visibility

Google's generative AI optimization guide, the same page that defines fan-out and grounding, says Google does not use llms.txt or other special AI text files for Google Search visibility.

Publishing the file buys nothing in Google Search visibility on Google's own account. Cross it off the list. If an llms.txt file is the first deliverable on a proposal for AI visibility work, ask about that line.

Common belief: Getting named as a source brings back the traffic a top result used to bring.

The correction: A citation is not the click a blue link used to be

Pew Research Center measured clicks on Google results pages. Users clicked a traditional search result in 8% of visits when an AI summary appeared, compared with 15% of visits when no AI summary appeared.

That figure counts clicking on Google results pages with and without a summary. It carries no number for how many visits a citation in any AI answer sends to a nonprofit site. Being named as a source still puts you on a page where, in Pew's data, users clicked a result less often, so read the settings and the reports before you commit budget to new pages.

Common belief: One line in robots.txt is the AI policy, and it covers ChatGPT.

The correction: Blocking GPTBot is a training opt-out; OAI-SearchBot is the one that carries ChatGPT answers

OpenAI documents GPTBot as the crawler used to crawl content that may be used in training its generative AI foundation models, and says disallowing GPTBot indicates the site's content should not be used for training. That line is a training opt-out.

ChatGPT's search features use a different crawler. OpenAI documents OAI-SearchBot as the crawler used to surface websites in ChatGPT search features, and says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though they may still appear as navigational links.

Open your own robots.txt and read which of the two names appears. Disallowing GPTBot leaves the search crawler untouched. Disallowing OAI-SearchBot is what keeps the site out of ChatGPT search answers.

Common belief: None of this is measurable, so there is no point looking.

The correction: Three reports already show whether any of this is happening

Search Console carries a Generative AI performance report. Its impressions count how many times links to your site were shown to a user in a generative AI feature on Google Search, broken out by Pages, Countries, Dates and Devices. That count says nothing about the impressions column in your Ad Grants account. A separate article covers a campaign that reads Eligible while showing no impressions.

Bing Webmaster Tools carries an AI Performance report. Total Citations counts the times your content was visibly referenced or shown as a source in AI-generated answers during the selected date range. Grounding queries are the key phrases the AI used when retrieving the content it referenced, and Bing states these are not full user questions or prompts.

ChatGPT referrals arrive in your analytics already labelled: OpenAI says ChatGPT referral URLs include utm_source=chatgpt.com, which sites can use for analytics. That count is arrivals, not mentions, the opposite of what Bing's Total Citations measures.

Common belief: Nothing on our side decides whether the site shows up in AI answers at all.

The correction: There is a switch in Search Console, and it may already be off

Search Console has a Search generative AI control. It lets you include or exclude your site from Search generative AI features: AI Overviews, AI Mode, and generative AI features in Discover. Exclusion keeps your site's links and content from appearing in those features or grounding responses there.

Two items, then, for this week. Open that control and confirm whether your site is included or excluded, then open robots.txt and read which OpenAI crawler names appear. Both are readings, not projects.

Another article covers Google's removal of the campaign-level language setting from Search.

Two readings before any new pages

The published requirements are short: a page that is indexed and eligible for a snippet, crawler access you have not quietly blocked, and reports you can read. Open the Search generative AI control, open robots.txt, then read what Search Console, Bing Webmaster Tools and your referral tag already show. If a quote for new page types still makes sense after that, you will be buying it on evidence.

The qualification review is free. All we need is your organization's name and website. Start the free quiz