What are AI overviews? A quick recap

AI Overviews (AIO) provide a summary to certain search queries at the top of the search results, generated through Google’s LLM (large language model) Gemini. These summaries draw from information available across the web with the aim to provide instant, comprehensive answers to users’ search queries.

Each statement in an AI Overview has a link icon where you can view the source, and the ‘show all’ option to the right-hand side on desktop, and boxes underneath the AIO on mobile enable you to view all sources used for the summary. On average, there are ten sources referenced.

Since AIO started rolling out back in May, Google has been doing a lot of testing, and there’s still a considerable amount of volatility in what queries AIO are showing for, as well as what sources are used.

The latest research suggests that AIO are showing for 17% of queries in the US and UK, but that percentage varies significantly by industry. As it stands, Healthcare queries generate the highest number of AIOs by a significant amount at 63%, followed by B2B Tech at 32% and Ecommerce at 23%. However, this is changing all the time as Google establishes informational need and assesses user interaction with this feature.

What does this mean for the search landscape?  

With AIO providing answers to queries directly in the SERP (Search Engine Results Page), there’s likely to be an increase in zero click searches. As users become more comfortable engaging with these AI generated summaries, they’re likely to spend more time on the SERP and less time clicking through to various sites.

However, this will only apply where AIO are appearing. And based on the latest data, the majority of AIOs are appearing for informational queries. This could explain why we’re seeing more AIOs for certain industries, for example, searches around healthcare tend to be informational.

It’s likely Google will focus on generating AIO for longer tail, complex queries where users are in the early stages of exploring a product or service, rather than in the later stages when they’re ready to purchase. Whilst there has been some testing, Google may be cautious about making changes that could impact revenue.

Google search

How can you optimise for AI Overviews?

AIO continue to be very volatile and there isn’t yet a huge amount of data to conclusively inform how we might optimise for this feature.  However, based on the data that is available, there’s a few early conclusions we can draw:

  1. Rank well in organic search – there is a strong correlation between a site used as a source in AI Overviews and SERP position. Latest research in the US suggests sites ranking in position 1 have a 53% chance of being featured in AIOs, whilst content ranking in 10th have a 36.9% chance. Ensuring you have a comprehensive SEO strategy in place continues to be important.
  2. Optimise for related queries – there is evidence to suggest Google generates AIO based on the direct query but also on related queries to provide a more comprehensive summary. It’s almost as if they’re anticipating the next question a user may have after their original query. Ensuring a brand covers related, contextually relevant questions around a topic could help support their inclusion as a source in AIO.
  3. Demonstrate EEAT – demonstrating the principles of Experience, Expertise, Authoritativeness and Trust is more crucial than ever. To provide the comprehensive and factually correct summary that Google is aspiring to with AIO, the engine is likely to want to source information from credible, expert-driven sites that can be trusted. SEOs have focused on EEAT for years, but ensuring this remains a key focus of a brand’s strategy will remain important.
  4. Optimise product pages – to increase the chances of being pulled into AIO, including in a product viewer or product carousel, brands should optimise their product pages to have concise, informative descriptions that highlight key product features, incorporate lists into descriptions where relevant, include product reviews and use schema to highlight key details such as pricing.
  5. Diversify content format – given their inclusion in certain AIO, images and videos are likely to play a more significant role in influencing CTR. Brands should optimise images (ensure they’re responsive, compressed, have descriptive alt text etc.) as well as ensuring they’re relevant to the written content. Videos should be high-quality, with optimised titles, descriptions and structured data implemented.
  6. Build a reputable brand – building a brand that is reputable and well recognised will continue to be important. Investing in brand building strategies, and ensuring your business provides high quality products and services will continue to be important. Whilst initially content in AIO was sourced from lesser-known brands, for certain queries there has been a return to sourcing from top, well-known brands.
Video

What are the challenges for Google’s AI overviews?

There are a number of potential challenges to Google’s AIO that they’re likely to have to address over the coming months:

  • Trust – given the early volatility and inaccuracies that were appearing when AIO first rolled out, many users may be cautious of whether they can trust the information included in the AIO summary. Google is, and will continue to, work hard to ensure increased accuracy to build up that trust with users to demonstrate that they can rely on the information surfaced.
  • Bias – another watch out for Google’s LLM is ensuring that biases are not built into the model which could affect the content generated. Earlier in the year users found that Gemini’s image generator was producing historically inaccurate images, which Google then apologised for, and temporarily paused its ability to generate people. As with trust, Google will need to work to avoid bias, so users are confident they are getting impartial information.
  • Copyright – while the list of links to where the information is sourced from does appear in the AIO, it is quite vague on what information is being pulled from where. This could put Google at risk of potential copyright issues. In 2023, the US Copyright Office launched an initiative to examine copyright law and policy issues related to AI. This included assessing the scope of copyrighted materials and AI training. The Office plans to publish a report in several sections analysing the issues sometime in 2024.

Final thoughts

The introduction and continued rollout of AIO provides a significant shift in the search landscape. Whilst there’s still a lot of volatility, there’s certain SEO best practices that we can continue to follow, as well as potentially a few new considerations, to optimise strategies and give brands the best chance of Google drawing on their content and featuring their site as a source in the summary.

Google themselves face a number of challenges around trust, bias and copyright that they will need to work though to convince users as well as official bodies that AIO can provide the comprehensive and accurate answers to queries they’re striving for.