Google’s AI Search Features: An Investigation for Tourism & Hospitality

by | Sep 3, 2025 | Expert Views

Google Search has been changing rapidly in the last 12 months in the UK, from the introduction of AI Overviews to public search in August 2024, to the rollout of AI Mode to UK users in July of this year. And there are no signs that it will be slowing down; also in July, Google announced another AI Search experiment, called Web Guide.

In the SEO community, these features have turned the industry on its head, and SEOs around the world have been keeping a close eye on all of the new developments and how these affect websites of all kinds. While there is a lot of information out there about all of these features, it’s hard to draw real conclusions and actionable insights and takeaways for small businesses, especially in the travel, tourism, and hospitality niches. Inspired by research done by SE Ranking, I decided to launch a much smaller-scale investigation into AI Mode, AI Overviews, and their relationships to one another and organic search results. With this investigation, I wanted to take a closer look at how these tools function on a hyper-local destination scale for the Lake District, Cumbria.

Before we dive into that, however, let’s look into what AI Overviews and AI Mode are and how they work.

What are AI Overviews?

AI Overviews were first introduced as “Search Generative Experience,” all the way back in May 2023, when it was slowly rolled out into beta testing to opt-in users in the US. In May 2024, Google began rolling it out to all US users, and this extended to the UK and other countries in August 2024.

AI Overviews typically appear at the top of a Search Engine Results Page (SERP) when a search is done, and generally provide a short summary of multiple sources in an attempt to answer the search query. They can incorporate dynamic elements, including plain text, bulleted lists, images, videos, and more. AI Overviews make use of Google’s search index, a massive database of all of the information on the internet, to pull what it determines to be the most relevant and helpful sources.

Over time, AI Overviews have changed drastically in response to users’ behaviour and feedback. Currently, AI Overviews usually appear minimised, with a “Show more” button for the user to expand and explore further. There is a list of sources along the right side of the overview, as well as link icons throughout the response, linking specific portions of the response to specific sources. AI Overviews can now also appear further down the SERP, or even within other SERP Features, such as in “People Also Asked” boxes.

Screenshot of an AI Overview for “where is the lake district”

Image: AI Overview for “where is the lake district”

Not every search generates an AI Overview, though according to SE Ranking’s May 2025 study (for US users), it was appearing in as many as 30% of their searches. The vast majority of AI Overviews appear for informational search intent, where the user is trying to find more information about a subject, whereas they are least likely to appear for navigational search intent, where the user is trying to navigate to a specific web page or website (i.e. “facebook login”).

What is AI Mode?

AI Mode is Google’s newest public “search experience,” which also uses Gemini, to provide immersive, detailed AI-powered answers. Instead of only showing website links, AI Mode collects information from across the web and presents it in a formatted, conversational response, similar to ChatGPT. AI Mode appears as a separate tab on the SERP, or as a separate button on Google’s homepage.

Screenshot showing the location of AI Mode on a SERP and Google’s homepage

Image: Location of AI Mode on a SERP and Google’s homepage

AI Mode was first announced and launched to opt-in beta testers in the US in March of this year, reaching a full US rollout in May, and then coming to the UK at the end of July. Much like AI Overviews, AI Mode also makes use of the search index; however, AI Mode goes a bit further, by using the sources it finds to generate entirely new responses and summaries of information. AI Mode’s interface is very different from a typical SERP, and can include much longer responses than AI Overviews, especially for complex queries.

Another major difference between AI Overviews and AI Mode is that AI Mode can generate multiple responses within the same search session, and it can reference and build upon its previous responses. For example, if you launch an AI Mode search for “what is there to do in the Lake District?”, and then, after reading its response, you decide you’d like to further narrow your search, you can follow with further queries, like, “what about in Keswick?” or “on September 15th?”

Screenshots of AI Mode searches for "what is there to do in the lake district" and "On September 15th?"

Image: AI Mode searches for “what is there to do in the lake district” and “On September 15th?”

Methodology

For this investigation, I selected 60 keywords from lists generated using Cumbria Tourism’s Semrush account. I included a variety of search intents, lengths of query, and search volumes ranging between 12,100 monthly searches down to just 480. The keywords selected also included a mix of “questions” (i.e. “where is the lake district”), as well as more generic, informational queries (i.e. “langdale pikes”). I tried to include a variety of sectors within the scope of this investigation, including queries directly related to accommodation, activities, attractions, and events.

For 25 of the keywords, I decided to run 3 independent, cookie-and-cache-cleared, incognito mode searches, recording all URLs cited in AI Overviews, AI Mode, and the Top 10 Organic links. In the interest of time, I decided to only use the sidebar-cited links for AI Overviews and AI Mode, and I did not include any other SERP feature links, opting only for the “true blue” Organic search results. As time was short, and I wanted to gather all of my data within one working day, I made the decision to only do 1 search for each of the remaining 35 keywords. In total, I did 110 searches for 60 keywords.

Disclaimers

Due to the nature of this investigation, the data had to be gathered manually, and interrogated using an Excel spreadsheet. As such, I was highly limited in the number of searches I could do and URLs I could record in a single day, so the data are quite limited in scope. Additionally, the data do not account for differences in time of day or location of search (all was done from my home office IP address), which may factor in results. Additionally, due to the ever-changing nature of algorithms and updates, the results may not be able to be replicated exactly on future occasions.

Due to the initial selection process including Cumbria Tourism’s Semrush account and reports, there is a strong bias in some of the search results to favour Visitlakedistrict.com.

In future investigations, I would love to expand the number of keywords searched, as well as to explore the volatility of AI Overviews and AI Mode between multiple searches of the same keyword. Additionally, it would be very interesting to observe AI Overviews’ and AI Mode’s behaviour around branded search terms for local businesses, accommodation and activity providers, attractions, and more.

How many sources do AI Mode and AI Overviews contain?

Chart showing the URL Count Distribution (AI Mode vs. AI Overviews)

Chart: URL Count Distribution (AI Mode vs. AI Overviews)

AI Mode

On average, each AI Mode answer included 13.83 URLs (sources). Across all 110 searches in AI Mode, 1,521 URLs were included as sources for its answers, 1,147 of which were unique (75.4% unique URLs).

I found the following distribution:

  • 16-20 URLs were the most common at 28.2% of all searches falling in this range.
  • 14.5% of all searches had over 20 URLs cited.
  • Only 7.3% of searches had 5 or fewer URLs cited.
  • The range of URLs cited was 25 (2 to 27 URLs).

AI Overviews

On average, each AI Overview answer included 15.25 URLs (sources), which is on average 1.42 more URLs than AI Mode. Across all 110 searches using AI Overviews, 1,678 URLs were included as sources for its answers, 1,071 of which were unique (63.8% unique URLs).

I found the following distribution:

  • 11-15 URLs were the most common at 30% of all searches falling in this range.
  • 22.7% of all searches had over 20 URLs cited.
  • 14.5% had 5 or fewer URLs cited.
  • The range of URLs cited was 58 (3 to 61 URLs)

Summary

AI Overviews had more URLs cited on average and by frequency, and had a long tail of citations and a slight skew towards less URLs cited; AI Mode had a tighter distribution range overall.

Neither AI Mode nor AI Overviews generated any responses with 0 or 1 citation, meaning that for all of the searches, Google always made an effort to cite at least 2 sources for its AI answers.

AI Mode was more likely to cite unique URLs over AI Overviews, meaning that AI Mode responses wouldn’t duplicate the same URL citation across answers or searches.

How much overlap is there between AI Mode and AI Overviews?

For the 60 keywords selected for this research, I looked at all Domains (or websites) and URLs generated to analyse the overlap between Domain and URL citations for each keyword, for a total of 1,813 URLs (including duplicates).

Domain-level overlap

Looking at domains, there was an average of 14.3% Domain overlap between AI Mode and AI Overviews, or an average of 2.55 Domains per query.

  • Only 6 keywords had 0% overlap between AI Mode and AI Overviews.
  • 20 keywords had ≥ 20% overlap between AI Mode and AI Overviews.
  • 13 keywords had between 0% and 10% overlap, and these were mostly for commercial search intent, such as:
    • “best pubs in lake district”
    • “best area to stay in lake district”
    • “things to do in kendal”

URL-level overlap

Overall, there was an average of a 3.9% URL overlap between AI Mode and AI Overviews, or an average of 0.6 identical URLs per query.

  • 37 (61.67%) keywords had 0% overlap between AI Mode and AI Overviews.
  • 4 keywords had ≥ 20% overlap between AI Mode and AI Overviews, all of which were informational search intent:
    • “highest point in lake district”
    • “how many Wainwrights are there”
    • “where is barrow in furness”
    • “killington lake”

Summary

This shows that, while AI Mode and AI Overviews sometimes would use the same Domains as sources, they very rarely would cite the same pages.

How much do AI Mode and AI Overviews overlap with Organic results?

Looking at the unique domains for the Top 10 Organic results for each of the 60 keywords, and comparing it with the unique domains cited in AI Mode, there is an average of a 14.7% domain overlap.

Comparing the unique domains in the Top 10 Organic results with those cited in AI Overviews, there is an average of a 15.9% overlap.

Other observations

  • AI Mode cited a Google domain (including links for Maps, Shopping, etc) 94 times, making it the most cited domain in AI Mode in this investigation, followed by Facebook.
  • User-generated content (UGC) domains did very well in AI Mode, with the top 20 most-cited including the likes of Facebook, AllTrails, TripAdvisor, and Komoot.
  • OTAs, cottage companies, and accommodation aggregators did very well overall, with the top 20 overall domains cited including: sallyscottages.co.uk, lakelovers.co.uk, sykescottages.co.uk, lakelandretreats.com, and expedia.co.uk.
  • Youtube was the overall winner in AI Overviews by far (with 103 total citations, with the next-highest being sallyscottages.co.uk at 54 citations), showing the importance of UGC and video-form content.
  • AI Overviews showed far more URLs for commercial queries than AI mode, with all of the top 10 keywords by URL count in AI Overviews being commercial.
  • Although there is not enough data to say conclusively, initial investigations showed AI Mode to be more volatile than AI Overviews. Looking at the 3 searches done on the same 25 keywords, AI Mode returned a range of 42 in total URLs cited (313-355), whereas AI Overviews were slightly more consistent with a range of just 21 (332-353).

Takeaways & Actions for Businesses

While this investigation is very small-scale compared to some of the studies done by SEO agencies and leaders, there are still a lot of meaningful insights and takeaways for tourism and hospitality businesses.

  1. AI Mode’s most-cited domains were Google-owned, like Maps and Shopping, which also includes information from businesses’ Google Business Profile like opening hours, address, and contact information. To prevent spreading incorrect information to potential visitors, it is important that the information on your Google Business Profile is kept as accurate and up-to-date as possible. You can read more about how to do so in this Google article.
  2. Google has published guidelines for how to write and publish high-quality content, including a guideline referred to as E-E-A-T, which stands for Experience, Expertise, Authoritativeness, and Trust. As shown by the prevalence of UGC and social media platforms in AI citations, it is clear that AI Mode and AI Overviews prioritise the Experience aspect of E-E-A-T, preferring to link to content where users and influencers are reviewing or visiting a business or area. This highlights the importance of your brand being present in these spaces, either through your own social media content, working with influencers, or ensuring that your visitors post helpful, positive reviews on sites like TripAdvisor.
    1. If you’d like to learn more about working with influencers, consider signing up for our upcoming Working with Social Media Influencers Webinar on 13 October!
  3. OTAs and other aggregators were cited many times in both AI Mode and AI Overviews, especially for commercial search intents (i.e. “best area to stay in lake district” or “weekend getaways lake district”). If you are listed on any OTAs or other distribution channels, it is worth checking that your content, photos, prices, and reviews are all kept as up-to-date as possible.
  4. There was only about 15% overlap between the Organic Top 10 results and AI Mode or AI Overviews, meaning that having a good organic ranking doesn’t necessarily mean that you will appear in AI results for the same searches. However, the inverse is also possible, where your website may appear in AI answers for queries that you do not rank well for.
  5. Ensure that your website and content are AI-ready. This can take multiple forms, such as including FAQ sections, implementing structured data where relevant for your website, and ensuring that important information isn’t kept behind Javascript-heavy elements.

Do you want to learn more? Sign up for my upcoming webinar on 10 September, AI is changing SEO: Is your business ready?, where we will cover all of the above and much more, including specific examples of actions you can take.

If you’d prefer a more hands-on approach, I’m also running an in-person workshop at Cumbria Tourism’s Staveley offices on 11 November, which you can register for here. At this workshop, we’ll go through some searches in-person and explore first-hand how to evaluate content with an eye for AI search.