Skip to content

Will People Click Links in SGE Results?

By Matthew Edgar · Last Updated: April 24, 2024

Google is currently testing the inclusion of generative AI responses in search results. This search generative experience (SGE) is placed above the list of traditional search results. The concern is that this will deter clicks—why would people click a link contained in an SGE response if the answer is already provided?

To find out more about the impact SGE would have on clicks, we asked 197 people to evaluate SGE responses. The primary question was if they would click on the links in the SGE response. We also asked how much they trusted SGE responses, how much detail was provided, and if they would ask follow-up questions.

Overall, we found that 58% would click on links contained in SGE responses, 22% might click, and 20% would not click. That means, around 20-42% of clicks are at risk of being lost. This is similar to a recent study from Gartner that found that organic search traffic could decrease by 50% or more by 2028.

Will people click on links in SGE responses? Our study found that 20-42% of people would not click.

SGE Clicks by Search Intent

Every search query is a bit different. One of the best ways to understand those differences is to evaluate the search intent of the query. People may be looking for information (informational intent), people may be researching products they want to buy (commercial intent), or people may be searching for ways to complete a purchase (transactional intent). SGE responses can appear for any of these intent types. To find out if there could be differences in whether people would click on SGE links, we assessed SGE responses for different search intentions.

How likely it is people would click on links in SGE responses change based on query intent

There were notable differences across different query intents. When reviewing an SGE response to an informational query, 28% of respondents would not click on any links in that SGE response but only 17% of respondents would not click for transactional queries. Commercial queries ended up in between with 20% of respondents saying they would not click.

There were also differences in how many people said they might click. Only 16% of respondents said they might click on a link in an SGE response to an informational query, but 23% might click on links in an SGE response to a commercial query. That suggests a possibility of gaining more clicks from an SGE response.

Encouragingly, 56% of respondents said they would click on links in the informational query’s SGE response and 62% of respondents said they would click on links in a transactional query’s SGE response.

While this seems like good news, keep in mind that the websites listed in the SGE response can be quite different than the websites listed in traditional search results. A study from Authoritas found that 93.8% of links in SGE responses were not in the top ranking domains. In the following example, only Investopedia and Budget Your Trip are included as links in the SGE response. These are the websites that would get the clicks. In contrast, the traditional search results include Kayak, Travelocity, Costco Travel, Expedia, and more. Those websites are only mentioned in the SGE response but are not included as links.

Comparing an SGE result to a traditional search result

SGE Clicks by Topic

Click behaviors may also differ based on the search topic. We know people have different expectations for YMYL topics, like queries related to finance or health, in non-SGE results. Do we see these differences in SGE as well?

We evaluated three topics: software, travel, and finance. The results clearly show that clicks on SGE responses would change based on the topic. Searches about software saw the fewest clicks, with only 51% of respondents saying they would click and 28% saying they would not. Travel saw the most with 63% of respondents saying they would click and only 18% saying they would not click. All three topics had similar rates of “maybe” responses, with about a fifth of respondents saying they might click on a link contained in an SGE response.

SGE Clicks change by search topic

Important Note: The study only evaluated a few queries within each topic. So, this should not be taken to mean that SGE responses to travel queries will always get more clicks than software or finance queries. An evaluation of more queries would be needed to draw those types of conclusions. This study’s data suggests that click rates do change based on the search topic.

How Much Do People Trust SGE Results?

Generative AI is not always reliable and may hallucinate responses. As people reviewed SGE results, we asked them how much they trusted the generated response. Most people (53%) trusted the text provided in the SGE response and 39% of people said they trusted the generated response at least somewhat. Only 8% said they did not trust the generated response.

Do people trust SGE responses?

Trust does influence clicks on links in SGE results. The more people trusted the SGE response, the more likely it was they would click on a link in the SGE response. Of those who said they trusted the SGE response, 59% said they would click on a link. Of those who said they did not trust the SGE response, only 44% said they would click on a link.

People who trust SGE responses are more likely to click on links

Do SGE Results Provide Enough Details and Context?

One of our hypotheses was that if people thought the SGE response contained sufficient details and context, they would not need to click on links in the SGE response. If the SGE response provides enough information, why would you need to go anywhere else?

There is a difference, but it was smaller than we thought it would be. We can break respondents into three groups: those who thought the SGE result contained enough details and context (the “Enough Details” group), those who thought the SGE result contained some but not enough details or context (the “Some Details” group”, and those who did not think the SGE result contained enough details or context (the “Not Enough Details” group).

The “Some Details” group would click the most, with 63% indicating they would click a link. The “Enough Details” group would click 56% of the time. However, of the “Not Enough Details” group, only 53% would click. That is counterintuitive—if the SGE response does not provide enough details, we expected the “Not Enough Details” group to click more.

There are also differences in the “might click” responses. The “Some Details” group might click 24% of the time, while the “Not Enough Details” group would only click 14% of the time.  Those in the “Enough Details” group would click 18% of the time.

What about those who would not click? Of the “Not Enough Detail” group, a third would not click on any links. In comparison, 26% of the “Enough Detail” respondents would not click and 14% of the “Some Details” respondents would not click.

While some of this seems counterintuitive, this suggests that a moderate amount of detail may encourage more clicking. Too much detail or too little detail deters clicking. In other words, it appears to be a Goldilocks situation: people will only click if the details and context are just right.

Details/Context of SGE Response Compared to Clicks

Do People Ask Follow-Up Questions?

One of the unique features of SGE is that you can converse with the response and ask follow-up questions. This is considerably different than refining a query in traditional search results. With a follow-up question, the generative AI powering SGE will present new, refined, and possibly personalized information. This means that one user may see multiple SGE responses.

Overall, 50% of respondents said they would ask a follow-up question. Of those who would click on a link in the SGE result, 73% said they would ask a follow-up question. Of those who would not click on a link in the SGE result, only 30% said they would ask a follow-up question. This suggests that asking a follow-up question would not deter clicks. It also suggests that maybe asking follow-up questions could encourage clicks, though we did not test that question specifically.

Follow-Up Questions Compared to Clicks

Of those who would ask a follow-up question, 48% thought the SGE response only contained some of the necessary details and context, while 36% thought the SGE contained enough details. Of those who would not ask a follow-up question, 51% thought the SGE result contained enough details. This further supports the idea that there is a “just right” level of detail and context to encourage some type of user engagement.

Takeaways and Final Thoughts

The main takeaways from this study include:

  1. SGE will lead to a decrease in clicks. The decrease in clicks will affect all types of search queries.
  2. Informational and commercial queries may be more at risk for lost clicks than transactional search queries.
  3. Some topic areas will have a greater decrease in clicks than others. The way SGE affects SEO performance will not be uniform.
  4. If the SGE response contains too much detail or not enough detail, people will be unlikely to click.
  5. Asking follow-up questions with SGE responses does not appear to deter clicks.

Now is the time to evaluate your search results and determine what your users and customers would do. Where does SGE currently appear (or where could it appear)? Does your website appear in the SGE responses—or at least websites like yours? If your website appears, would your users and customers click on your website from the SGE response? Are the details provided in the SGE response “just right” to encourage clicks?

Get Help With SGE

If you want help doing this type of evaluation of your company’s search queries and answering those questions, please contact me today. This study is based on some of the work we are doing to help companies prepare for SGE.

Respondents Overview

  • Respondents ranged from 18-65+, 2/3rds of respondents were between the ages of 25-44.
  • All participants were paid for their responses.
  • Responses were collected in April 2024.
  • Respondents viewed desktop SGE results.
  • All respondents were in the United States.

You may also like

What Are Generative AI Hallucinations?

One of the biggest problems with generative AI responses is hallucinations. What are hallucinations? Why do they happen? More importantly, as generative AI starts to dominate search results, how do we factor hallucinations into our future SEO strategies?

Understanding Prompt Types

How do we optimize for prompts and make sure our companies show up in generative AI search responses? How is this different than optimizing for search queries? It begins by understanding what types of prompts people use.

Search Intents & Generative AI

How will SGE change SEO? It likely depends on the search query intent. Learn how to adapt your SEO strategy and prepare for SGE.