Google AI Mode: how to get cited in AI answers without losing leads

A purchasing director asks Google which surface treatment supplier to choose for aluminium components destined for the automotive industry. The answer comes in AI Mode, well written, with three companies cited and their links. If yours isn’t there, you’ll never see that contact, not even as a lost visit in Analytics. If it is there but the link leads to a generic page, you lose it all the same.

The short answer is this: to be cited by Google AI Mode you need an indexed page that’s eligible for a snippet, content split into blocks that each answer a precise question on their own, clear entities (who you are, what you do, for whom) and signals of authority beyond your website too. To avoid losing leads you need the second piece: pages that convert even with fewer clicks, a brand that people then search for by name, and a way to measure all this, which today is only partly possible.

We explained how the different engines reason in our guide to GEO. Here we stay with Google and a commercial question: how a citation becomes a request for a quote.

Where does Google AI Mode stand in Italy (October 2026)?

AI Mode has been available in Italy and in Italian since 8 October 2025, as a tab on the results page and in the Google app for Android and iOS. Since August 2026 Search Console has had a report dedicated to impressions in generative AI features, but without data on clicks and queries.

Google Italy announced it: conversational questions with follow-ups, search by voice and camera, and the query fan-out technique, which breaks the question down into sub-questions and runs many searches in parallel. According to Google, people using AI Mode write questions two or three times longer than a classic search.

On the data front, on 3 June 2026 Google presented the generative AI performance reports in Search Console, extended to every site in the world from 31 August 2026. They show impressions in AI Overviews and AI Mode by page, country, device and date, with a separate report for Discover’s AI features. Clicks, CTR and queries are missing, so you know which pages appear but not for which questions. Impressions and clicks from these features are also still included in the normal performance report, mixed in with everything else.

Classic SEO and optimisation for AI search: what changes for a B2B company?

The basics are the same: Google states that no additional requirements beyond those of search are needed to appear in AI Overviews and AI Mode. What changes is the unit that gets chosen (the passage of text, more than the whole page), the type of question and the way success is measured.

The Search Central documentation on AI features says so explicitly: to be a supporting link, the page must be indexed and eligible to show a snippet, and no special files, dedicated markup or “copy for AI” are needed.

With fan-out, a question like “how to choose MES software for an engineering SME with three plants” becomes a dozen smaller searches: costs, integration times, ERP compatibility, use cases. Each sub-search is a chance to be cited, even if your page doesn’t rank for the full question.

	Classic SEO	Optimisation for AI Mode and AI Overviews
What you win	A position among the results	A link among the sources of an answer
Unit of work	Page and keyword	Passage answering a sub-question
Type of query	Short, 2-4 words	Long, with context and constraints
Role of the brand	Helps CTR	Weighs on the choice of source and on the next search
Main metric	Clicks and position	AI impressions, citations, brand searches, leads
What happens next	A visit to the page	Often no click, sometimes a very qualified visit

So a B2B company needs fewer generic articles and more pages on the questions an engineer or buyer asks when they’re already comparing suppliers. You’ll find more on how to read these intentions in the article on search intent.

How to structure content so that it gets cited

A page gets cited when it contains a block that fully answers a sub-question and makes sense even when lifted out of the rest. A clear heading, the answer in the first few lines, details and data straight underneath.

Here the rule needs applying to the pages that matter for leads (services, products, industry solutions, comparisons), which are often the worst written, with promotional copy that doesn’t answer any question.

Take a company that makes industrial dust extraction systems. The “Solutions for joinery” page usually talks about quality, experience and service. A citable version answers the questions a production manager asks Google instead:

• what airflow is needed for a line with three CNC machines;
• which ATEX regulations apply to wood dust;
• how long installation takes and whether production has to stop;
• what maintenance is needed and how often.

Each question becomes an H2 or H3, with the answer in the first two sentences. Where it makes sense, a table (for example recommended airflow by type of machining) and a list of requirements. These formats are easy to extract, and they also help the salesperson when they send the link to a customer.

Two often-overlooked tips. Key data must be in the page’s HTML text: technical data sheets only in PDF and tables inside images are worth little as a source (the PDF can stay as a download). And each page must have a clear scope: five distinct industry pages are better than one “Applications” page with thirty lines for each.

Which authority signals count for AI Mode?

AI Mode relies on Google’s quality and ranking systems, so the same signals as search count: competent authors, a recognisable and verifiable company, mentions and links from industry sources. With few sources cited per answer, the margin is narrower. In the projects we work on, we act on four levels.

Real authors. Technical pages signed by a process engineer or the owner, with an author page showing role, experience and LinkedIn profile.

Your own data. A sentence like “in the tests we ran on 40 batches, average roughness fell from X to Y” is citable; “we guarantee the highest quality” isn’t.

Presence beyond the website. Articles in trade magazines, talks at fairs and conferences, listings with trade associations, reviews on Google Business Profile. A Pew Research study of the searches of 900 US adults found that Wikipedia, YouTube and Reddit were among the sources most cited by Google’s AI summaries, that is platforms where users and third parties do the talking.

Consistency. Name, address, industries served, products: the same information on the website, on LinkedIn, in directories and in the Google profile. Inconsistencies make it harder to understand that it’s the same company being talked about.

Sources, FAQs, entities: how to make clear what you’re talking about

A generative system is more willing to cite a page that uses precise terms, states the sources of its data and explicitly links the company to the topics it deals with. FAQs and structured data give shape to text that must already be clear on its own.

Sources. When you cite a regulation, a market figure or a technical specification, link to the primary source: the text of the regulation, the body, the manufacturer.

FAQs. At the bottom of service and product pages they work because each question resembles a prompt. Take them from customer emails, from the objections the salesperson hears and from queries in Search Console.

Entities. Google reasons in entities: companies, people, products, places, concepts. In practice:

• always use the same name for products and services, without alternating acronyms and different trade names;
• explain in one sentence what you do and for whom on the home page, the about page and the service pages;
• connect pages to each other with descriptive internal links, so that the site draws a map of its topics;
• use the structured data types Organization, Person, Product or Service, FAQPage and BreadcrumbList, consistent with the visible content.

Google points out that there’s no specific markup for AI features: schema is there to describe the page. In the article on GEO you’ll find the most useful markup types and how to validate them.

How to turn a citation into a lead

A citation in AI Mode brings few clicks, but anyone who clicks has already read a summary and arrives with a precise question. To avoid losing them you need landing pages that answer that question and offer a simple next step, and a brand recognisable enough to be searched for by name later on.

Pew’s figures give the order of magnitude: with an AI summary on the page, users clicked on a traditional result in 8% of visits, against 15% without a summary, and only clicked a link inside the summary in 1% of cases. These are 2025 US data on AI Overviews; Google, for its part, maintains that clicks from pages with AI Overviews are “higher quality”, with more time spent on the site.

Pages that convert even with few clicks

If the cited page is an informational article, whoever arrives finds the same answer they’ve already read and leaves. For every page that appears in Search Console’s AI report, ask yourself: what’s the next question someone arriving here will ask, and does the page anticipate it?

For example, an article on “how to calculate a plant’s compressed air requirement” should offer a calculator or a downloadable table, a link to the energy audit service page and a short form to request a site survey. The form asks only for name, company, email and one line of context.

Brand search as a deferred conversion

Many B2B buyers read the answer in AI Mode, jot down two or three names and come back days later searching for the company directly. In Analytics that visit shows up as organic brand or direct, and the credit for the citation is lost. That’s why searches for the brand name (in Search Console, filtering queries containing the brand) are one of the most useful signals. If AI impressions rise and brand searches rise in the same period, the citation is working even when there’s no click.

Lead magnets built for people arriving from AI

Anyone arriving from an AI answer is halfway through their evaluation. Resources that help them decide work well: a checklist for comparing suppliers, a spreadsheet for estimating costs and timescales, a technical guide with the regulatory requirements. An AI summary doesn’t replace them, and they give you a contact to follow up with a lead nurturing journey.

What changes for the CRM and sales

Without knowing where a lead comes from, you can’t judge which channel brings in customers. Two simple fixes:

1. add an optional “How did you hear about us?” field to the form with the option “AI answer (Google, ChatGPT…)”;
2. save the landing page and source of every lead in the CRM, so you can cross-reference the pages cited in AI Mode with leads and closed opportunities.

More detail in the article on B2B lead management.

Which KPIs should you track for AI Mode and AI Overviews?

There’s still no data point that directly links citation, click and lead. You cross-reference six indicators: impressions in Search Console’s AI report, citations on a list of questions, share of voice against competitors, referrals from AI assistants in GA4, brand searches and leads by landing page.

KPI	Where it’s measured	What it tells you	Limit to be aware of
Impressions in AI features	Search Console, generative AI performance report	Which pages appear in AI Overviews and AI Mode, in which countries	No clicks, no queries, AI Overviews and AI Mode added together
Citations on key questions	Manual checks or tools such as Ahrefs Brand Radar and Semrush	Whether you’re cited on the 20-50 questions that matter to your business	Answers vary from user to user and over time
Share of voice in AI	Same tools, compared with competitors	How much space you take up compared with other suppliers	Depends on the list of questions chosen
Referrals from AI assistants	GA4, custom channel group	Visits and conversions from ChatGPT, Perplexity, Gemini, Copilot	Clicks from AI Mode and AI Overviews arrive as Google organic and can’t be separated
Brand searches	Search Console, query filter with the brand	Whether visibility in answers generates direct searches	Also grows for other reasons (fairs, campaigns, word of mouth)
Leads by landing page	GA4 and CRM	Which cited pages generate leads and opportunities	Needs clean tracking of forms and events

A few practical pointers on setting them up:

• Search Console AI report. Every month, export the pages with the most impressions and compare them with the pages that generate leads. Pages that are seen a lot without conversions are the first to review.
• Channel group in GA4. Create a custom channel group with an “AI assistants” channel that collects sources such as chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com, placed above the referral channel.
• List of questions. Divide the questions by stage (discovery, comparison, choice of supplier) and repeat the checks every month using the same method.

Bringing organic data, campaigns and CRM together in a single picture is the job of our analytics and data service.

Mistakes to avoid

The costliest mistakes are decisions taken without data.

• Optimising only the blog. The pages that bring leads are the service and product pages. If they stay as brochure copy, citations land on informational articles, which convert poorly.
• Measuring only organic clicks. With fewer clicks per search, a drop in traffic can go hand in hand with more visibility and more leads. Look at AI impressions, brand searches and leads together.
• Excluding the site from AI features without data. Since 31 August 2026 Search Console has offered every site a control to exclude the site from AI Overviews, AI Mode and Discover’s AI features. For a B2B company that lives on visibility it rarely makes sense, and today there’s no click data to assess it calmly.
• Chasing files and markup “for AI”. Google says they aren’t needed. Time is better spent on content and conversion pages.
• Long forms and generic CTAs. A “Contact us” at the bottom of the page with nine mandatory fields wastes the click of someone who has already done half the work.

Action checklist for the next 90 days

1. Export the 50 pages with the most impressions from Search Console’s AI report and cross-reference them with the pages that generate leads.
2. Write down 20-50 real customer questions and check who is cited today in AI Mode and AI Overviews.
3. Rewrite the 5-10 most important service and product pages in question-and-answer format, with your own data and sources.
4. Add an FAQ built on real questions and check the structured data.
5. Put a clear next step on every cited page: a downloadable resource, a calculator, a short form.
6. Create the “AI assistants” channel in GA4 and the “How did you hear about us?” field in the form.
7. Set up a monthly report with AI impressions, citations, brand searches and leads by page.
8. Plan two or three appearances beyond your website, for example an article in a trade magazine.
9. After three months, compare the data with the starting point.

Want to know whether AI Mode cites you and whether it brings leads?

With our AI and SEO visibility audit we look at where you appear today in AI Mode and AI Overviews compared with competitors, which pages AI sees and how well they convert, and what’s worth fixing first across content, authority and tracking. If your market is still barely moving on AI search, we’ll tell you, and point out where your time is better invested.

You can book a call with Federico or write to us via the contact form. The details are on the GEO service page.

Frequently asked questions

Is Google AI Mode available in Italy?

Yes, since 8 October 2025, in Italian. You’ll find it as a tab on Google’s results page and in the Google app for Android and iOS.

How can I tell whether my site appears in AI Mode?

Search Console has the generative AI performance report, available for every site since 31 August 2026. It shows impressions by page, country, device and date, adding AI Mode and AI Overviews together, but not clicks or queries.

Do you need special markup to be cited by AI Mode?

No. Google says the usual SEO practices are enough: the page must be indexed and eligible for a snippet. There’s no schema dedicated to AI features.

Can you see AI Mode traffic in GA4?

Not separately. Clicks from AI Mode and AI Overviews arrive as Google organic traffic; you can only isolate visits from external assistants such as ChatGPT, Perplexity or Copilot, with a custom channel group.

Does being cited in AI Mode bring leads even without clicks?

Often yes, indirectly: whoever reads the answer jots down the name and searches for the company later. It’s worth tracking brand searches and asking in the form how people heard about you.

Is it worth excluding the site from Google’s AI features?

For most B2B companies, no, because you give up visibility that your competitors keep. It makes more sense for paid or restricted content.