# AI Overviews and B2B manufacturers in 2026: which pages get cited?

> Which pages Google AI Overviews cite, how much organic clicks fall, and what B2B manufacturers should do, based on Google documentation and a field experiment.

- Page: https://chingeo.com/en/insights/google-ai-overviews-b2b
- Updated: 2026-11-12
- Source: ChinGEO (云端智荐)
- 中文版: https://chingeo.com/insights/google-ai-overviews-b2b.md

Author: [Huang Zheng](https://chingeo.com/en/about), Founder · Systems architect · Published 2026-11-12

**AI Overviews cite pages that Google has indexed, that are eligible for a snippet, and that answer a specific sub-question; Google says no special optimisation is needed. When an overview appears, organic clicks fall sharply. For manufacturers, finding a qualified supplier is an informational search, so aim to be named in the overview and let product and certification pages serve buyers who click through to verify.**

## The short answer

AI Overviews change where a buyer first meets you. They do not change the fact that a buyer has to verify you before sending an enquiry. That leads to three points:

1. **Clicks fall, mainly on informational searches.** In a randomised field experiment, organic clicks to external sites fell by 38% when an AI Overview appeared. Learning a product category, checking a standard and working out how to choose a supplier are exactly the questions overviews answer.
2. **There is no new trick for getting in.** Google states there are no additional requirements to appear in AI Overviews or AI Mode, and no special structured data or AI-specific files to add. Getting cited starts with doing standard SEO properly.
3. **Work on two things: being quotable and being verifiable.** Write certifications, standards and specifications as answers that can be lifted straight into an overview. Then make sure product and certification pages hold up when a buyer clicks through to check.

For definitions, such as what AI Overviews are and how nosnippet differs from Google-Extended, see the glossary entry on [AI Overviews](https://chingeo.com/en/glossary/ai-overviews).

## The data and its limits: fewer organic clicks when an overview appears

The best evidence so far comes from one randomised experiment. Its result is clear, but it covers a narrow setting.

In the working paper [Google AI Overviews and Publisher Traffic: Evidence from a Field Experiment](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6513059), Saharsh Agarwal (Indian School of Business) and Ananya Sen (Carnegie Mellon University) used a custom Chrome extension to randomly assign users to the normal Google results page, with AI Overviews, or to a version with the overviews removed. The abstract reports:

| Measure | Result (as stated in the abstract) |
|---|---|
| Organic clicks to external sites | Down 38% when an overview appears |
| Probability of a zero-click search | Up 33% when an overview appears |
| Ad clicks and number of searches | No effect found |
| Where the effect concentrates | When the overview sits at the top of the page, and on the highest-ranked organic results |
| User experience | Survey responses showed removing overviews did not make information harder to find or results worse |

Four limits matter:

- **It is a working paper** posted on SSRN in 2026 and has not been peer reviewed.
- **The users are in the US.** The authors' [registration](https://www.socialscienceregistry.org/trials/17393) in the AEA RCT Registry plans about 1,500 people who use only Chrome, with the intervention running from 7 January to 7 February 2026. The final sample is reported in the full paper. Browser extensions run on desktop Chrome only, so in our reading the result does not cover mobile search.
- **The 38% applies only when an overview appears.** It does not mean a site loses 38% of its traffic. Google says AI Overviews show only when its systems judge them additive to classic Search, and they often do not trigger.
- **The searches are everyday consumer searches**, not B2B procurement. We have found no public study that measures the same effect on engineering or sourcing searches.

Google offers the other side. Its [documentation](https://developers.google.com/search/docs/appearance/ai-features) and a May 2025 [Search Central blog post](https://developers.google.com/search/blog/2025/05/succeeding-in-ai-search) say clicks from results pages with AI Overviews are "higher quality", with users more likely to spend more time on the site. Google has not published the data behind that statement, so treat it as the platform's view rather than an established finding.

## For manufacturers, finding a qualified supplier is an informational search

Manufacturers feel the click loss earlier than consumer brands do, because of how buyers search. Before a buyer sends an enquiry, they have to understand the category, the standards and how to screen suppliers. Almost all of that is informational.

| What the buyer is doing | Example search | Intent | What you want |
|---|---|---|---|
| Learning the category and how to specify it | how to choose a stormwater detention module supplier | Informational | Your brand named in the overview or its sources |
| Checking standards and certificates | how to verify a CE certificate for industrial valves | Informational | Your certification or standards page cited |
| Comparing named suppliers | [brand] vs [brand], [brand] factory audit | Mixed | Consistent, checkable facts on your site and elsewhere |
| Finding and contacting you | [brand] datasheet, [model] price inquiry | Navigational or transactional | Product pages and enquiry forms one click away |

The first two rows are where overviews are most likely to appear and where clicks are most likely to stop on the results page. They also decide whether you make the shortlist. Only the last two rows are where buyers come to your site.

So the core judgement is this: **you do not need to win back every informational click. You need to be in the answer to informational questions, and to have the evidence ready when a buyer clicks through to check.** A buyer who sees your name and certifications in an overview may not click this time, but may search your brand name next. That is a navigational search, and those get clicked. This path is our inference and no public data yet measures it, which is why branded searches should be tracked on their own (see below).

## Which pages get cited: Google's conditions, plus writing that can be quoted

Being cited depends on a short list of official conditions. Everything beyond that is about how the page is written.

**Google's conditions**, from [AI features and your website](https://developers.google.com/search/docs/appearance/ai-features):

- The page must be indexed and eligible to be shown in Google Search with a snippet. There are no additional technical requirements.
- You do not need new machine-readable files, AI text files or special markup, and there is no special schema.org structured data to add.
- Meeting every requirement does not guarantee crawling, indexing or serving.
- AI Overviews and AI Mode may use "query fan-out": several related searches across subtopics and data sources, used to build one response.

What fan-out means for a manufacturer is our reading. A buyer asking how to choose a supplier for some type of equipment may trigger searches on certification requirements, applicable standards and material options. **The page that gets cited is often not your company profile, but the page that happens to answer one of those sub-questions.** One broad "why choose us" page does less than a page, or a clear paragraph, for each sub-question.

These page types are worth rewriting first:

| Page | Sub-question it answers | How to make it quotable |
|---|---|---|
| Product specifications | What are the key parameters? | A table with units, test conditions and the standard number each value relies on |
| Certifications and testing | Which certificates do you hold, and how can I check them? | Certificate name, number, issuing or testing body, validity, and where to verify it |
| Standards mapping | Which standard does the product meet in my market? | One standard per section: the conclusion first (class or grade met), then the test basis |
| Engineering or selection guide | Which model fits which operating condition? | The selection answer first, then the method, with clear limits of use |
| Application cases | Where has it been used, with what result? | Only project facts you can disclose and that can be checked |

This is the "trustworthiness" part of [E-E-A-T](https://chingeo.com/en/glossary/eeat): information others can check is information worth citing. For a page-level checklist, see our [6-point product page checklist](https://chingeo.com/en/insights/ai-supplier-search-product-page-checklist).

## How to measure once clicks fall

If you count only clicks, every time you appear in an overview without a click scores zero. Measure in four layers instead:

| Layer | What to look at | Where |
|---|---|---|
| Visibility | Impressions in AI Overviews and AI Mode, and which pages earn them | Search Console generative AI performance report |
| Search performance | Impressions, clicks and click-through rate by page | Search Console performance report (Web search type) |
| Brand demand | Searches for your brand and model names | Search Console, filtered to branded queries |
| Business results | Landing pages, enquiry volume and quality | Your analytics and enquiry records |

Three things about Search Console's data are worth knowing first:

1. **Clicks and impressions from AI Overviews are part of the overall Web data**, not a separate figure. Google's [counting rules](https://support.google.com/webmasters/answer/7042828) say an overview occupies a single position in the results, and every link inside it is given that same position.
2. **Since 31 August 2026, every site has a generative AI performance report.** According to the [help page](https://support.google.com/webmasters/answer/16984139), it counts impressions of links to your site in AI Overviews and AI Mode, broken down by page, country, device and date. The documented dimensions do not include queries, so it tells you which pages were shown but not for which questions.
3. **A Search generative AI control arrived at the same time.** According to its [help page](https://support.google.com/webmasters/answer/16908024), it can exclude a whole site from these features, after which the site gets no impressions or traffic from them. It does not affect other search results and does not govern model training, which Google-Extended handles. A manufacturer that wants buyers to find it has no reason to switch it off.

Treat rising generative AI impressions, rising branded searches and steady or rising enquiries as the signs that this work is going right, rather than organic clicks alone.

## What to do: four steps, in order

Secure eligibility, then fix content, then set up measurement, and only then divide work with other AI assistants.

### 1. Eligibility

- Use the URL Inspection tool in Search Console to confirm key pages are indexed and that the HTML Googlebot receives contains the main text.
- Check robots.txt, and check that your CDN or firewall does not block Googlebot.
- Look for stray nosnippet, max-snippet or data-nosnippet. Limiting snippets limits citation in overviews.
- Blocking Google-Extended does not take you out of overviews; it covers training and grounding in Google's other systems.
- Keep structured data consistent with the visible text.

### 2. Content

- List the sub-questions buyers ask (category, standards, certification, selection, delivery) and give each one a page or a clearly answered paragraph.
- Lead every page with the answer. Put specifications, standard numbers and certificate numbers in text, not only in images or PDFs.
- Write English pages in the trade vocabulary of your target market, using its conventions for standards and units.

### 3. Measurement

- Export the generative AI performance report monthly and look at the pages with the most and fewest impressions.
- Track branded searches and enquiries alongside organic clicks, in one sheet.

### 4. Dividing work with GEO

AI Overviews run on Google's index, so sound SEO is most of what eligibility needs. ChatGPT, Perplexity and other assistants retrieve from different sources and depend more on third-party coverage and consistent brand information. See [GEO vs SEO](https://chingeo.com/en/insights/geo-vs-seo) and the lesson [how SEO, paid ads and GEO fit together](https://chingeo.com/en/academy/seo/b-1-3).

## What not to do

These moves either have no basis or make things worse:

| Move | Why not |
|---|---|
| Buying a service that "guarantees" AI Overview placement | Google says indexing and serving are not guaranteed; nobody can promise a citation |
| Adding "AI markup" or AI text files for overviews | Google says none is needed, and there is no special schema.org markup |
| Adding nosnippet site-wide to "protect" content | You lose regular snippets and eligibility for overview citations at once |
| Hiding answers behind forms or downloads | Overviews cannot quote them, and buyers cannot see them |
| Mass-producing Q&A pages to cover every question | Google's [spam policies](https://developers.google.com/search/docs/essentials/spam-policies) list scaled content abuse as a violation; violating sites may rank lower or drop out of results |
| Stopping content work when clicks drop | First check generative AI impressions, branded searches and enquiries |

## Where to start

Start with the basics: are key pages indexed, can Googlebot read the main text, and does your site give a direct answer to each of your buyers' core questions? An [AI visibility audit](https://chingeo.com/en/audit) checks your site's technical foundation and how AI platforms answer about you, which helps decide which pages to fix first. If you want this done as a programme, see our [export SEO service](https://chingeo.com/en/services/seo).

## Frequently asked questions

### Do I need special optimisation to appear in Google AI Overviews?

No special markup or files are needed. Google says a page that is indexed and eligible for a snippet can be shown, and standard SEO practice applies; what remains is writing paragraphs that answer questions directly.

### How much do AI Overviews reduce clicks?

A randomised experiment with US users, published as a working paper and not yet peer reviewed, found organic clicks to external sites fell 38% when an overview appeared. That applies only to searches showing an overview and was not measured on B2B sourcing searches.

### How can I tell whether my site appears in AI Overviews?

Use the generative AI performance report in Search Console. It counts your impressions in AI Overviews and AI Mode by page, country, device and date.

### Does blocking Google-Extended remove my site from AI Overviews?

No. Google-Extended covers model training and grounding in other Google systems. To leave generative AI features in Search you would use the Search generative AI control or snippet controls, which a manufacturer that wants to be found should not do.

### What does it cost to respond to AI Overviews?

It depends on your site, target markets and how many pages need rewriting. Start with an audit and a plan, and assess the work by pages and deliverables.

---

ChinGEO (云端智荐): GEO specialist for Chinese B2B exporters: get your brand recommended by ChatGPT, Perplexity and Gemini, backed by websites, SEO and AI customer service. The site does not publish prices; costs are discussed through the project planner and a diagnosis call.

- Free AI visibility audit: https://chingeo.com/en/audit
- Site guide for LLMs: https://chingeo.com/llms.txt
- Southwest China Account Manager: Zhou Rongyue, +86 13880491401
- North China Regional Lead: Cui Yue, +86 13811954582
