Testing the Potential of Google’s Search Generative Experience (SGE)
Google announced some exciting news regarding its new Search Generative Experience (SGE). They are finally opening up access to Search Labs and enabling us to start testing Google’s generative AI powered Search experience. At Semetrical, we, of course, jumped at the opportunity!
We began testing the capabilities of Google’s new feature to see what it’s truly capable of, and where its downfalls lie. In this article, we share our honest feedback based on our experiences with Google’s Search Generative Experience. Just note that it’s still very early days and SGE changes every day!
Let’s get into it…
What’s in this article?
- What is Google’s Search Generative Experience?
- Experimenting with Google’s Search Generative Experience: our findings
- When does SGE not appear by default?
- What does the future hold for Google and how do we prepare?
What is Google’s Search Generative Experience?
The Google Search Generative Experience, or SGE for short, encompasses a range of search and interface functionalities that seamlessly integrate generative AI-powered outcomes into the query responses of the Google search engine. Instead of serving users a typical listing in the search engine results page with a link to a website, Google uses a generative AI engine that summarises information from multiple sources automatically and serves this to users. In this sense, SGE is very similar to ChatGPT. For the moment, SGE is still in testing stages with users in the U.S. via Search Labs and the release date is yet to be confirmed, however, it is likely to be at the end of this year or in 2024.
Experimenting with Google’s Search Generative Experience: our findings
We explored the capabilities of AI-generated search results and its potential impact on search engine technology. Join us as we unveil the insights gained and discuss the implications for the future of digital with Google’s Search Generative Experience. Here at the tests we ran:
- Recipes
- Local packs
- Google Maps
- Product review queries
- Property listings
- Recent news listings
Nonsensical recipes
We asked Google ‘how to boil eggs’ and it served us a very confusing recipe, that consisted of multiple different cooking times and non-chronological, repeated instructions! Of course, the time it takes to boil an egg is subjective, depending on whether you’d like a hard boiled egg or a runny yolk, which means different search listings will recommend different times. However, when summarised into one response by Google’s SGE, it returns a nonsensical recipe that is of no value to the user.
From this, it is clear that there is still some work to be done to improve SGE’s outcomes and ensure the results serve user intent and add value. By incorporating user feedback, refining the algorithm, and enhancing the clarity and coherence of instructions, Google can optimise SGE to deliver accurate, useful, and personalised responses. This will ultimately enhance the user experience and establish SGE as a valuable resource for accurate and relevant search results.
Integration with Google Maps (above the local pack)
When you search with a location-specific query, Google’s Search Generative Experience response will appear above the local pack. The results appear in a five-pack that looks almost identical to regular local packs, with a short description next to each result, star ratings, images, and descriptions.
We searched for ‘best clubs in Berlin’ to test this feature and this is what Google served us:
This means that local businesses must start to pay increased attention to optimising their local listings to ensure that they appear with the SGE results. It will be important to ensure that business information is up to date and accurate, but also their proposition is compelling enough to attract users amidst greater competition in this area.
Maps
Google’s Search Generative Experience tries to be helpful, but it doesn’t always go to plan. We asked for a map of Rome and this is what we got:
As you can see, it returned a helpful description of Rome, including some of the key attractions and history of the city. It also served us maps of Italy, which is not what we asked for. We asked for a map of Rome specifically. So, in this instance, SGE didn’t quite hit the mark.
Product review queries
When asking Google for the best strollers for 2023 to test product review queries, the results serve 14 reviews sites in a carousel of links, Unlike regular organic results, SGE does not appear to show Google shopping links, but this is likely to change on actual release. This will be coupled with paid ads that will appear above the SGE result, as while users are testing SGE, Google has disabled these for obvious reasons.
Whilst the inclusion of multiple review sites, offering a wide range of opinions and insights for informed decision-making is welcome, it will again make competition stiffer with the shifting results formats. Organic results will be further pushed below the fold, which becomes more prominent on mobile for these types of queries, meaning the impact on click-through rate (CTR) here could be considerable for a publisher if not cited as part of the SGE results.
Property URLs
We searched for ‘new homes for sale Watford’ and SGE was able to provide a list of new homes for sale in Watford, providing information on the house, number of bedrooms, and price. It was also able to provide an average property price in Watford for reference and the most sought-after areas. However, it did not provide links to where you can find these properties. Without these links to URLs, the user experience is significantly impacted and the results are not useful.
The SGE results for this type of keyword in the real estate market is especially lacking. There is zero reason for an AI-generated result apart from the semi-useful house price information. A user would almost always need to click onto a cited result to actually fulfil their search. And since not all cited results are taken from those ranking in the top position organically, publishers who do rank well are getting punished as the new SERP pushes them down the screen for a very unhelpful AI-generated response.
Recent news listings
When testing out the query ‘recent news in London’ in Google’s SGE. it served a bullet-point list of some recent news stories from London, and a few sites to click-through to. This included news sites such as The Sun and The Independent. Although, the stories on the list did not include links to URLs. On the whole, this experience is not terrible, however, it is not better than Top Stories in Google’s regular organic listings. The Top Stories feature is much more user-friendly, providing a host of relevant news stories from a plethora of news sources with links to URLs for each. This enables users to choose their preferred news sites and click-through to read the whole story.
When does SGE not appear by default?
There are a few instances where SGE does not appear by default, which is for the following query types:
- Highly political topics
- Some medical topics (but not all!)
- Some YMYL queries (financial queries)
- Sensitive/adult queries
- Branded queries (with option to click to expand)
- Quick answers, such as the weather or song lyrics
The exclusion of SGE in these cases is likely due to the complexity and sensitivity of the information involved, the need for accuracy and expertise, privacy considerations, specific brand intent, or the availability of quick and straightforward answers. These exclusions demonstrate Google’s effort to ensure the user experience remains informative, relevant, and safe across different query types. It’s also likely that after significant testing and user feedback, there will be more query types added to the list!
What does the future hold for Google and how do we prepare?
So, what is our verdict on Google’s Search Generative Experience and what’s next?
As SGE answers are not solely derived from top-ranking pages, it’s vital to identify the characteristics that make content more likely to appear in these results. By doing so, we can focus on creating and optimising content that is more likely to be featured in SGE, increasing our chances of capturing user attention and engagement. And to optimise our clients’ content for SGE, we need to identify the queries and questions users are likely to ask about our clients’ products or services. In this way, we will be fully prepared for when this tool is rolled out properly in the coming year.
Integrating credible external sources, such as third party mentions, interviews, and corroborations from multiple sites, bolsters the authority and reliability of our content. This step increases the likelihood of our content being considered for SGE, thus improving our overall search visibility.
Ultimately, the future of search results with the introduction of Google’s Search Generative Experience (SGE) holds great potential for both users and content creators. SGE represents a significant advancement in search technology, and its implications are likely to reshape how information is accessed and presented online. That being said, based on our experiments, it still requires some refinement in order to fully achieve this.
Similarly, as shown with our findings and as with any AI-powered system, there may be challenges with misinformation and biased responses. Google will need to continually refine SGE to avoid propagating false information or presenting biased viewpoints to users.
With SGE generating direct answers, users may find less need to click through to traditional search result pages. This could potentially impact website traffic and influence content creators to optimise for featured snippets and direct answers to remain visible. As such, this is something we need to keep on top of as there’s no doubt that Google’s SGE will have a huge impact on SEO strategies and performance in the near-future.
