The author’s views are solely his own (except for the unlikely event of hypnosis) and may not always reflect the views of Moz.
Discover key statistics about Google’s Search Generative Experience and its impact on SEO, ads, and user experience. Dive into the latest trends shaping the search landscape in this insightful Whiteboard Friday with Tom Capper.
Click on the whiteboard image above to open a high-resolution version in a new tab!
Happy Friday, Moz fans. I’m here today to share with you some quick stats, four quick stats about Google’s search generative experiences.
So if you’ve never heard of this or seen this before, this is a new type of search result, a new type of SERP feature. I assume Google started testing earlier this year. Currently this is not widely rolled out. You can only see this if you are activated in Google Labs. You’re logged in, using Chrome and have a US IP address, although VPNs work.
So maybe this is a small hint of a direction that Google might consider. Maybe not. I may do a few blog posts about whether I think they’ll stick with this plan, but I still think it’s interesting to take a look at what they’re doing now and yes, see what we can observe .
The structure of Google’s Search Generative Experience
So the basic structure of the SGE is that it is placed at the top of a search engine results page, above all other types of results and SERP features. And you have a little warning. Generative AI is therefore experimental. Then a block of content, usually text, then some questions. So ask for a follow-up and then some suggested questions.
For example, I searched for Mozcon. One of the suggested questions is: Where is Mozcon? And this would only result in a follow-up search. And then there are these three links on the right that look like organic portrait results, but are actually supposed to be the articles that this AI-generated text is based on.
Key SBU statistics
What statistics would I like to share with you?
Left in the SGE
Well, I find these three links very interesting. One might reasonably assume that these are just the top three organic results or something similar, but that is actually not the case. In only 13% of the 100 SGE SERPS I examined, all three of these links were actually present in the top 10 organic links.
I should say something about the data I use here. So these are the top keywords from each MozCast vertical. In total, a hundred SERPs are recorded on the desktop in the USA. So the study is relatively small, but it’s enough to get an idea of what’s going on here. So yes, 13% of SGE SERPs had 100% overlap between these three links. Then it also appears in organic form underneath, which I think is surprisingly low. And 41% of the time, there was no indication at all that none of those links actually appeared in the organic results, which, yeah, I find that very surprising. It shows that this is a different system. This is not based on organic.
Advertisements are subject to SGE
The second statistic I want to share with you concerns ads. So you’ll notice that I haven’t placed any ads above the SGE here and that’s not a coincidence. Not once have I seen ads, when they exist, fall below the SBU, i.e. pushed all the way to the bottom, and only 30% of the SERPs I looked at even had ads? Which I think is pretty low when you look at the top competing MozCast keyword terms, but also think about how far down they’ve been pushed. So it wouldn’t be a particularly good situation economically for Google if they continued like this.
The third statistic I want to share with you addresses these questions. So, like I said, you need to make another request and then ask some suggested questions here. Now it looks a bit like people are asking questions too. And in fact, the questions you see here often appear in a box with other questions below. I find that a bit strange. I think it’s very uncharacteristic of Google to have such dual functionality and have two SERP features that essentially do the same thing. I also asked both people somewhere further down the box. And then these questions often even with the same questions. So 83% of the SGE SERPs had people asking for a SERP feature as well, which like I said, I think makes it seem like they rushed it a bit. You know, if they had thought about it a little more, maybe they would have done it. We’ve changed the SERPs below to not contain overlapping features. I have some more stats on this that I will share later.
But that also happens with cards. So if there are cards in this block then usually or often I should say there is also a local package further down that basically has the same functionality twice. It’s not really very elegant how these things look.
What is actually included in the SBU?
And that brings me to the last statistic I want to share, which is what you actually have in the SBU. So 70% of the time it’s just raw text, like a featured snippet, except it’s obviously generated by AI. The second most common are locations, i.e. location lists. And when that happens, it’s a little strange because those three links are usually just links to the Google Maps results or local Google results that are also listed on the map. Here too it seems a bit clumsy. Then you can get products here in 7% of cases. When they first demonstrated this on Google IO, I didn’t notice that they had ads in the AI generated block. I’ve never seen this in the wild. I’m not sure if they actually have this functionality. It was probably just a model or something like that, or maybe a statement of intent.
So yeah, those are four quick stats about SBUs. The main thing I want to take away from my experience collecting this data is that it’s a somewhat half-baked feature at the moment. So they probably rushed it out to respond to pressure from competitors and investors. I can imagine that there is still a long way to go. Anyway, I hope you enjoyed it. I hope you found these statistics interesting. Thanks.
Video transcription from Speechpad.com