In part one of this series, AI and the Future of Search, we discussed the roots of Google’s engagement with Artificial Intelligence, from the acquisition of DeepMind Technologies in 2014 to its integration into various facets of the platform. In 2023, we witnessed a pivotal moment in this during Google I/O, marked by the introduction of generative AI. Notable developments included the “Help me write” feature in Gmail, offering time-saving email composition assistance, and the advent of Google’s Generative AI-powered Search Experience (SGE), promising an enhanced search experience.
In part two, we are going to share our very own SGE research, as well as our top observations about SGE results. It is important to note that as this is an experimental feature, results may vary over time as updates are being made.
What is SGE?
Google’s SGE integrates generative AI into the search experience. When a user carries out a search, SGE will give them the option to generate an AI response. This will then produce an AI-powered snapshot that will provide a concise overview of the user’s query. SGE has been trained to give non-emotional or opinionated responses, instead giving neutral responses backed up by sources on the web. These sources always appear to be referenced in the results so that users can see the source, and also click through for more information.
SGE is powered by a number of large language models (LLMs) and has been trained to carry out search-specific tasks such as identifying high-quality web results. Google states that SGE is used alongside their core ranking systems. By focusing SGE on particular tasks, such as corroboration, Google can effectively reduce certain known shortcomings of LLMs, such as hallucinations or inaccuracies. However, this does not mean that SGE is always correct.
SGE results vs organic search results
SGE is likely to have a big impact on the world of search, and people are understandably worried that this could affect their rankings in organic search. We have conducted in-house research in order to investigate how similar SGE results actually are to standard results in order to evaluate whether or not SGE appears to process search queries in a similar way to the organic results we’ve all become accustomed to.
Using a VPN for California, USA (January 2024), we compared standard organic search results and SGE-generated results associated with a variety of search terms and phrases. Although our sample size was relatively small, we thought the results were worth sharing as they show a strong trend that there is an overlap between what is currently displayed on the organic search results page and what is referenced by SGE. We found that 55% of our searches had a URL that was also referenced for the same query in SGE.
Overlapping search results vs. results with no overlap:
Being powered by AI, SGE searches are more than likely going to generate results that are different from those produced by standard organic search. By nature, AI will be continuously learning and developing search results – and whilst Google’s algorithms are now largely AI-based, they aren’t generative in the way SGE is. Standard search algorithms can use AI to analyse content and keyword focus before mapping said content to relevant search terms – whereas SGE can generate original, unique content based on learnings taken from a multitude of different sources.
As such, it comes as no real surprise that there are SGE and standard organic results containing no overlap. However, at 45%, the percentage of results without any overlap might be smaller than expected.
What do these research results mean for SEO?
As you can see from the chart above, the number of keyword searches we saw with no overlap between standard organic search listings and SGE-generated results was very similar to the number of results where there was at least one common link. So what does this mean for SEO?
For us, it is positive to see that the majority of searches have some kind of overlap. If all results were void of any overlap, we might be concerned that the current SEO best practices are unlikely to have any impact on SGE search results. As it stands, it is reasonable to assume that there are familiarities between ranking factors being identified and favoured by Google’s standard search algorithms and ranking factors being championed by SGE.
As SGE is taking up valuable real estate within organic search results, it is important that we understand how and why these results are generated so that we can work with our clients to help maximise their website’s potential visibility throughout SERPs. It is also important to ensure that this is done without negatively impacting existing rankings. As SGE is still in beta and constantly changing, our goal at Varn is to stay ahead of the curve and innovate our SEO processes and tools in line with the latest AI and SGE developments. There is still much to learn, and we hope that you follow us on our journey. You can keep up to date with our latest insights, observations, and learnings by signing up for our newsletter and by regularly revisiting the Varn Blog.
Keep an eye out for more upcoming SGE blogs, including ‘Our Top 5 SGE Observations’ and a review of the Google SGE patent, where we discuss how the Search Generative Experience works. In the meantime, if you have any questions about Google SGE, AI, or how to protect your organic rankings, get in touch with the SEO experts at Varn. We would love to hear from you.
