How AI Search Is Changing the Way People Find Websites

This is a guest post by Sophia Nomicos, founder of Loopla, a UK family events marketplace, and author of the Agentic Formula newsletter.

The way people find websites is changing.

Some searches still start on Google. Others now start in ChatGPT, Perplexity or Gemini. And even on Google, more answers are being shown directly in AI summaries, before anyone clicks a result.

That means fewer searches are turning into website visits. SparkToro found that 68.01 per cent of Google searches in the US ended without a click in the first 4 months of 2026.

So if your brand relies on search traffic, the changes could be having a big impact.

While search result clicks are getting harder to win, another kind of traffic is starting up: people arriving from AI tools and AI-generated answers.

AI referral traffic is still small but the early conversion data is interesting.

Across several early benchmarks, AI-referred visitors appear to convert around 4–6x better than traditional organic search visitors, though results vary by business and sector. One B2B tech benchmark from Opollo found that AI-referred visitors converted at 14.2 per cent, compared with 2.8 per cent from Google organic.

So the question is no longer just: how do we rank on Google?

It is also: how do we become one of the sources AI tools trust enough to mention, cite or send traffic from?

Some teams are starting to run their own experiments into what improves a brand's authority in the eyes of LLMs.

One example is Glasp. Its founders increased daily sessions from ChatGPT from 517 to 19,129 in 4 months — roughly 37x more AI traffic. The figure counted real people clicking through from ChatGPT, not bots.

So what can businesses do now?

The field is still new. The research is early. But a few patterns are starting to emerge.

It is also worth pointing out that this does not replace SEO. The companies doing this well are usually building on top of good SEO, not abandoning it.

How to increase your chances of being cited by AI tools

There is no single checklist that guarantees AI visibility.

AI tools do not work exactly like Google. They can pull from search results, product feeds, knowledge sources, social platforms, reviews, forums, media coverage and their own model memory.

That makes the work broader than classic SEO.

For marketers, the goal is to make the brand easier to find, easier to understand and easier to trust across the web.

AI tools look beyond your own website

AI tools look beyond your website to understand your brand.

They also pull information from the wider web, and some sources seem to carry more weight than others.

Peec AI analysed 30 million AI citations and found that Reddit, YouTube and LinkedIn were the most-cited domains across ChatGPT, Google AI Mode, Gemini, Perplexity and AI Overviews. Wikipedia and Forbes also appeared in the top five.

This means that appearing on those sites will likely give your brand more authority signals outside your own domain.

For most businesses, that means focusing on the platforms AI tools already seem to use: LinkedIn, YouTube, Reddit, review sites, directories, comparison pages, partner pages and industry publications.

A software company might focus on LinkedIn, YouTube, G2, Capterra, Gartner Peer Insights and comparison pages.

A local business might focus on Google Business Profile, review sites, local directories and local press.

A consultancy might focus on LinkedIn, podcasts, trade publications, partner pages, case studies and client testimonials.

The goal is to make sure your brand appears in the places your market already trusts, and where AI tools are likely to look for context.

Tell the same story everywhere

The next common thread that came up was making your brand easy to understand.

AI tools are not only reading your website. They may also pull from your LinkedIn page, YouTube channel, Reddit mentions, review listings, directory profiles, articles and other places where your brand appears.

When those places describe the company differently, the brand can be misunderstood.

Semrush found that when a brand is described differently across blogs, social media, YouTube and Reddit, AI tools can get confused. In the same report, 30 per cent of marketers said AI tools had given inaccurate descriptions of their brand, and 29 per cent said AI tools had shown unclear brand positioning.

So if your website says one thing, your LinkedIn page says another, your YouTube description is out of date, and your directory listings use old positioning, it creates a messy picture of what the company does.

For example, a company that calls itself a coupon site in one place, a review site in another, and a software directory somewhere else is harder to understand.

A company that uses the same clear description across its website, social profiles, directories and press mentions gives AI tools a cleaner signal.

As a practical takeaway, draft one concise 2–3 sentence description of the brand. It should cover what the company does, who it helps and where it fits in the market. Then use that same description, or a close version of it, everywhere the brand appears.

Reddit and LLM search

Reddit is now one of the main places AI tools can use to understand what people think about a topic.

That is because Reddit is full of genuine user content and opinions. People ask questions, compare products, complain about problems, and explain what worked or did not work for them.

Reddit says it has more than 100,000 active communities and 116 million daily active unique visitors, which means there are public discussions across most categories.

Google has also put more weight behind Reddit. In 2024, it expanded its partnership with Reddit to access Reddit's Data API. Google said Reddit has "an incredible breadth of authentic, human conversations and experiences." AP reported the deal was worth about $60 million.

For marketers, this does not mean using Reddit as a promotion channel.

It means understanding the conversations already happening in your market.

Look for the subreddits where your customers spend time. Watch the questions they ask, the products they compare, and the language they use when they describe problems.

Post useful and helpful comments in the relevant chats. Do not use Reddit to promote your brand heavily, as promotional posts are often removed and accounts can be banned.

However a presence on Reddit means that LLMs see your brand when they search for answers.

Set up Bing Webmaster Tools

Most teams check Google Search Console. Fewer check Bing Webmaster Tools.

That now matters more because Bing sits inside Microsoft's wider search and AI ecosystem, including Copilot and Bing's AI-generated answers.

Bing has also added an AI Performance report inside Bing Webmaster Tools. It shows when your site is cited in AI-generated answers, which pages were cited, and how that citation activity changes over time.

It will not show everything happening across ChatGPT, Claude, Gemini or Perplexity. But it does give businesses one official place to start checking AI visibility, rather than relying only on guesswork.

Bing also recommends using sitemaps and IndexNow together. Sitemaps help search engines understand the structure of your site. IndexNow lets search engines know when pages have changed, so updates can be found faster.

So as a starting point, set up Bing Webmaster Tools, submit your sitemap, check that Bing can crawl your site, and review the AI Performance report if it is available.

List products inside ChatGPT

For ecommerce and marketplace businesses, AI search is not only about getting articles cited.

Product discovery is also moving into ChatGPT.

When someone asks a shopping-style question, ChatGPT can show product options with images, product details and links to buy. For some eligible products and merchants, it may also show Instant Checkout, so customers can buy without leaving ChatGPT.

Merchants can also share product data with OpenAI. OpenAI says most merchants start with product feeds, which help control how products appear in ChatGPT. The feed can give ChatGPT more accurate information on names, descriptions, prices, availability, images and seller details.

That means product data is becoming part of AI visibility. Clean product titles, descriptions, images, prices, availability and shipping information now matter beyond your own website or Google Shopping.

OpenAI says product results are not ads and are not influenced by partnerships. The product appears when ChatGPT sees it as relevant to the user's request.

For now, this matters most to businesses that sell physical products. But it points to a wider shift: AI tools are moving from answering questions to helping people choose and buy inside the chat.

Build proof outside your own site

AI tools do not only look at what your website and profiles say about you. They can also cross-check what other sites say about you.

That can include mentions of your brand on Reddit, Wikipedia, YouTube, LinkedIn, blogs, G2 and other trusted sites.

So one of the clearest ways to build authority is to get more reviews on trusted third-party platforms.

For software companies, that might mean G2, Capterra, Gartner Peer Insights, Software Advice or TrustRadius.

For local businesses, that might mean Google Reviews, Yelp, Tripadvisor, Trustpilot or local directories.

For ecommerce brands, that might mean product reviews on marketplaces, Google Reviews, Trustpilot, YouTube reviews and Reddit mentions.

The point is simple: do not only keep reviews on your own website. Get them onto the sites your buyers, and AI tools, can check.

Write content for conversational searches

There are also ways businesses can optimise the content on their own website for LLM search.

This includes sales pages, product pages, service pages and blog content.

People do not use AI tools in the same way they use Google. Google searches are often short, like "vibe coding definition". AI searches are more likely to sound like a full question or a command: "How do I start vibe coding if I'm not technical?" or "Summarise this for me."

That means website content needs to include the longer, more natural phrases people are likely to use inside AI tools.

These can be added to sales pages and product pages as FAQ sections. They can also be used as subheadings in blog posts, guides and comparison pages.

The answer below each question should be concise and complete. It should make sense even if it is read on its own, away from the rest of the page.

That gives the LLM more context and a cleaner answer to work with.

What marketers can do next

The AI search landscape is still early, but the first steps are practical:

Check how your brand is described online. Make sure your website, LinkedIn, YouTube, directories and review profiles all explain the business in the same clear and concise way.
Show proof in more places. Get reviews, case studies, media mentions and product comparisons onto the sites your buyers already use and trust.
Build a presence on Reddit, YouTube and LinkedIn. These platforms keep appearing in AI citation research, so they should be part of your visibility strategy. Focus on useful answers, clear profiles and content your buyers would genuinely trust.
Add natural questions to key pages as subheadings or FAQs. Use the questions customers are likely to ask in ChatGPT, Perplexity or Gemini, then answer them clearly.
Clean up your product data. If you sell products, make sure titles, descriptions, images, prices, availability and shipping details are accurate and up to date.
Set up Bing Webmaster Tools. Google Search Console still matters, but Bing now gives another view of how your site may appear in AI-generated answers.

Together, these steps give your brand a stronger baseline for being seen as an authority by LLMs.

Sophia Nomicos is the founder of Loopla, a UK family events marketplace helping parents find classes, clubs, camps and days out. She also writes the Agentic Formula newsletter on what is working for businesses with AI.