Early Impressions Of Google Search Results Generative

In the blog post about SGE Early Impressions of Google search results with generative AI we are going to be going over the changes that Google is bringing to search because Google has just released early access to Google SGE, Which is the search results with generative AI inside that will give you a bunch of answers instead of clicking on websites.

I know it’s a scary moment for SEO, So I wanted to break it down and tell you exactly what it is because there are a lot of things that we can observe, as we test it ourselves that were not shared in the Google I/O presentation.

Here I recommend you to follow my other post about SEO. In This post, you can learn How you index and rank your pages easily on Google Generative Search results. I know it’s a scary moment for SEO, So I wanted to break it down and tell you exactly what it is because there are a lot of things that we can observe, as we test it ourselves that were not shared in the Google I/O presentation.

Before coming forward Let’s check what Google is Saing about SGE or Google Search Generative Experience, the Impressions Of Google Search Results with generative AI.

An overview of SGE By Google

“SGE is an early step in transforming the Search experience with generative AI. When using SGE, people will notice their search results page with familiar web results, organized in a new way to help them get more from a single search.”

10 Takeaways About New Google Search Generative Experience

As you know Recent Google announced its new Search results named Google (SGE) Search Generative Experience. It has the ability to generate AI-powered text with the help of Artificial intelligence like Chatgpt-4 and other chatbots. Here are 10 takeaways for New Google Search Generative results SGE that you must know.

Takeaway 1 -AI Answer Takes 8-10 Seconds to Show up In Google SGE

The mechanics And I think one of the main mechanics that has not been explained in the Google I/O presentation is how the AI answers are generated and the mechanics behind that. In reality, when you are typing a search query in this new version of Google, it takes about 12 seconds for the AI results to pop up.

And if you do nothing, that’s exactly what happens. The results pop up, they push down the search results, and people have the choice to read that instead of clicking on the website. But during these 8 to 12 seconds, there is just a small blinking box that doesn’t really take much space on the SERPs where the primary call to action is clicking on the website link.

Now, if you start scrolling down, these AI results are basically hidden. So even though they keep loading above, they actually won’t push the search results down. So as soon as people hit their scroll mouse a little bit and go down a little bit on the SERPs, it basically won’t displace the search results any more and it’s kind of like normal SEO as we know it today.

That’s a really interesting mechanic because essentially it gives the power to the user to decide whether they want an AI answer and they want to wait for like 8 to 12 seconds or if they want to click on a website by just slightly scrolling down a little bit and stopping the AI generation.

Now, Google is also experimenting with another two-step mechanic where they don’t load the AI answers at all, when you search a query and you have a generate AI answer button that then generates the AI answer. We’ll see which one they pick in the end.

I really like this mechanic because it really lets the user choose what they want. Do you want an AI answer and then you don’t scroll or do you want a website and then you just scroll subtly a little bit and you don’t have AI coming in the way anymore?

This means that light browsers that are not really that deep into the topic will probably pick the AI answer and that’s probably the traffic that will drop from websites, whereas the people who are more engaged and want a deeper solution or a deeper resource, will go for websites.

That means that while ad websites will definitely feel the blow by the lower search traffic overall, the sites that sell affiliate products or the sites that sell products on their site will probably not feel the drop in conversion as much because most of the traffic that will hit their website will be even more targeted.

Takeaway 2 – Long Snippets Are Cut Short In Search Generative Experience

AI answers are cut short Another observation that I’ve made is that quite often Google will cut short the AI answer with a show more button.

Now, the way they actually size it is quite often based on the size of your browser and of your screen because I’ve tried it on different browser sizes and the page would often layout in a way where the AI answer would take most of the space, but they would always show the first result regardless of the height of my browser.

I think it’s a really good thing that most of the time you do see the first organic result because in a way it reminds you that you can always scroll down to get websites and not just rely on the AI answer to get your info. No, there are no ads in this beta.

So I don’t know how it’s going to behave when we get as heavy SERPs, but my expectation is that they will probably shrink the AI answer even more because that’s kind of the adjustable variable in the SERP layout like nothing else really moves.

So I’m kind of hoping that the AI answer will take the heat, not the organic results, but let’s see how it rolls out. The third observation that I’ve made is,

Takeaway 3 -Not All Results Will Trigger An AI Answer

Did AI generate content for all queries? that not all queries generate an AI answer. Now don’t get me wrong, most queries will. And I was quite surprised by the range of queries that Google is willing to dip their toes in, including a lot of ymyl queries, but they still cover their tracks a little bit.

So for example, when I Googled a query of high interest to me – how to prevent hair loss – I got an answer.

But in the end, I got a disclaimer trying to essentially take away the liability of Google that says this info does not constitute medical advice or diagnosis.

But there are some keywords that straight-up showed no answers. So, for example, when I Googled best credit cards, there was no answer at all. It was just the classic good old organic results.

I’m sure NerdWallet is going to be very happy about that. The problem is, I can see these AI answers work quite a lot like featured snippets, where essentially one day you might have no featured snippet and the next day you may have one. And these things might change dynamically over time as Google refines the way they generate AI answers.

For SEO, essentially the results are really slow. Ranking a page is not something that often happens overnight. It’s going to be quite difficult to react to these things. And dodging the queries that have no AI answers is going to be a little bit difficult, I think. So let’s see how it rolls out.

Another thing that I’m wondering about is the liability of Google on ymyl queries, especially because they are answering medical queries, they are answering financial queries with AI.

I’m really thinking, what’s going to happen the day someone follows the advice they find on the SERPs and something bad happens? They may just be testing that on the beta, but not rolling it out live. I think that’s something that could really happen.

Takeaway 4 – Are featured snippets And PAA (People Also Ask) still here?

The fourth thing that I’ve noted is that featured snippets and people also ask are still here. In the Google I/O demonstrations, they did not show any of them in the SERPs, but in a version that they gave us, they’re all here.

And quite often that’s going to lead us to weird situations. So when you have an AI answer that takes the same sources as the featured snippet, they will often say exactly the same thing. The AI may be a little bit longer and more developed, but really it leads to a lot of redundancy on the SERP, which really doesn’t look good for Google.

So, for example, I searched for the keyword how to lose weight fast. And on top of that, you get an AI answer where the NHS is taken as one of the sources.

It looks like the SEO team is doing a better job than the emergency rooms these days. The featured snippet is also taken from the NHS and you get a lot of items in the list that they give you that are the same, like don’t skip breakfast, eat regular meals or eat plenty of fruits and vegetables.

And you can also see the limits of the AI model here, because quite often, as it takes info from different sources, it will give contradictory advice. So, for example, one of the list items here is don’t skip breakfast. But they also recommend intermittent fasting, which for most people means skipping breakfast.

So which one do you do if you want to lose weight? That’s why this reinforces my belief that these AI answers are kind of okay, but anyone who really cares about the topic is probably going to have to look a little bit deeper to achieve their goals.

Takeaway 5 – AI answers vs featured snippets that quite often the AI answer is worse than the featured snippet.

Another weird situation that happens with featured snippets and AI answers is AI answers vs featured snippets that quite often the AI answer is worse than the featured snippet. So, for example, if you search for how to paste on Mac, the featured snippet tells me that you need to press command V or go into the Edit and Paste menu in my menu bar.

But Google insists on generating a pretty long answer that basically tells me the same thing and gives me a bunch of fluff that I don’t need to know how to paste.

So I really think Google needs to kind of like scale it back down and remove a bunch of AI answers from situations where the featured snippet literally does a better job. It’s going to save them money, it’s going to drive more traffic to websites, and it’s going to make everyone happy.

Takeaway 6 – AI Will Straight Up Without Attribution

Now, talking about awkward situations, there are times when AI will straight up steal the content from your site and give you no attribution and no link whatsoever.

So, for example, if you search for the keyword how to mew, you will get an AI answer that’s word for word the same content as the featured snippet and the top-ranking site. Talk about redundancy and wasted space on the SERPs.

It doesn’t just suck for the user to just have two times the same answer on the SERP, but straight-up stealing content may spell trouble for Google. Now, the same thing happened with the query, how long does it take to get over a breakup? The first paragraph was straight up lifted from Women’s Mag without a citation in the main view.

Thank you Lily Ray for reporting this one, it was a really interesting one. So I’m expecting Google is use the beta to find these issues and iron them out eventually, but really, we could have a lot less space wasted, and websites a lot higher on the SERPs while making the experience better for the user.

Takeaway 7 – Sites in AI Snippets are often Very different from top-ranking sites: Good news for small publishers

The 7th thing that I noted, and that’s finally good news for small publishers like you and I is that the sites used in the AI answer and featured on the top right or in the carousel mobile are often very different from the top-ranking sites.

It looks like Google heavily favors specialized websites in these kinds of AI answers, instead of ranking first for every inquiry. So, for example, if you search for the best protein powder, The top organic results will show the usual suspects – Forbes, Healthline, Food Network, et cetera.

But the AI snippet will feature more specialized sites like Total Shape, AGN Roots, or proteininworks.com. All of them are small publishers like you and me.

This is a huge opportunity for small publishers to finally start showing up again for these big commercial queries where big publishers have mostly taken all the traffic at this point and probably get some traffic back from these queries that you may lose on long-term queries that AI takes away.

Now, if you’ve read the Google I/O notes correctly, you also noted that they announced a big shakeup in the helpful content system and I’m kind of hoping that these AI snippets are running on that already.

This is foreshadowing Google giving more spotlight to specialized sites over big publishers that rank for everything. Now, if we go back to the carousel on the top right of the AI answer in Google I/O, they only showed us three resources shown in there. But actually, now it’s a carousel that will show up to six or seven websites that you can scroll horizontally.

It’s not very intuitive on desktop, but on mobile, this carousel is actually right below the AI answer and above the number one result. So I expect the sites that are featured there to get quite a bit of traffic from mobile, maybe a bit less from desktop, but that is quite encouraging, especially if they push smaller sites. Now there is a button in the top right that will break down the AI answer and show you the source for each statement that it makes.

But I don’t expect many people to use this one, to be honest. It’s more like a cover for Google to say that they’re quoting sources than something that people really use.

Takeaway 8 – SGE SERPs have been Heavily redesigned

The 8th thing I’ve noticed is that the SERPs have been heavily redesigned and I think that is as big of a change as the AI itself. So I’ve taken the regular version of Google that I’m running on my regular browser and the new version of Google that I run in the beta on Chrome and I’ve played the Seven Differences game and here’s what I found.

The first thing is that all search results show favicons, which wasn’t the case in the previous version of Google, which makes you much more aware of the kind of site that you’re clicking on and you recognize the sites that you know a lot more.

The brand is featured a lot more. So I expect search to become much more of a brand exercise than it is today because it is more featured and people are going to click on the resources they trust the most. And as we all know, a higher click rate tends to lead to a higher ranking in Google.

The second thing is that the SERPs layer is a lot more arid, there’s a lot more space between results. The results themselves take more space and I’ve actually looked at the pixel height of each result and on the old Google the pixel height is 120.89 pixels.

On the beta of Google, the pixel height of a single result is 146.89 pixels, which means that each result is taking on average 21% more vertical space.

What that means is the further down you are on the SERP, the more scrolling it’s going to take compared to the current version of Google. So I expect the click distribution to be a lot more top-heavy because these top results take more vertical space.

The third big change is that all SERPs are now infinite scroll. Now we already had that on mobile most of the time and sometimes on desktop, but there’s literally no page two of Google at all in the beta version. So I think that is a change that they’re making for good.

When Google rolls out AI there will be no more page two ever on Google. The fourth change, which is a pretty big one in my opinion, and I don’t like it, to be honest, is the links to the website.

So the title tag that you see, or the rewritten title tag by Google is now black instead of blue, which means they blend a lot more in the SERP. What I expect it to do is it will shift the CTR more towards Widgets like images, videos, products, et cetera. But I guess with the refocus on the brand, I guess they’re going more for you recognizing the brand than clicking on something that looks like a link.

But still, it is a pretty big visual change to the SERPs, and I’m wondering how people are going to react to that if they roll it out. I really like these changes in terms of branding. I think it’s cool. It will help sites that actually have an audience do better in terms of CTR and the sites that don’t build an audience not to do so well in terms of CTR, which I think is a good thing. But I wish they went even further where they would let me follow websites.

And then when I Google a query that is related to that website, if they have a piece of content that matches, they rank that piece much higher than they would otherwise so that website owners could focus on growing followers and growing their organic traffic from their followers. That would be cool.

Takeaway 9 – Link outs?: AI Answers don’t link even when they should.

Now my 9th observation, and to get back to the AI snippets, one thing that’s very interesting and weird is that they almost never link out. Actually, they really don’t link out at all. So let me give you an example of a query and show you how weird that is.

If I was searching on Google for the best subreddits for SEO, I’d get a big list of subreddits I could join if I wanted to connect with the SEO community.

But what’s my option now? I can’t click on them. The only solution would be to take them one by one, Google them in another tab and go on them individually. Whereas if the AI snippet was linking, I could click on them one by one and bam, I’m exactly where I want to be.

But there is one case where Google does an excellent job linking out and that is when you do a product query and they can plug Google Shopping in. So if you search for, say, the best bikes for kids, then here you’re going to get a beautiful comparison table that looks a lot like an affiliate site.

Thank you, Google, for lifting that out of our sites as well. And people can click on each bike where it opens a side panel showing them the prices, and the retailers, and you can price shop right in there.

Now, they only do that for Google Shopping stuff. So if, for example, your search for the best VPN, you won’t get that, you’ll just get a text list and then you need to search for stuff or you need to click on search results, which in a way is better for websites that are comparing and doing affiliate reviews.

Now, when we talk about linking, we’ve been putting some pressure on Google over on Twitter. We’re really trying to get Google to link inside the snippets to the sources that they’re taking info from.

Takeaway 10 – New conversation mode.

The 10th thing I want to talk about is the new conversation mode and how not very good it is. Now at the bottom of the AI answer, you have a prominent button that says that you can ask a follow-up query to the search engine.

When you click on it, it triggers a new interface that feels a lot like a chatbot if you’ve been using ChatGPT or Bing or something like that. But it really doesn’t work the same way. It’s a lot worse, to be honest. So to go back to my example, if I Google best bikes for kids, I get my first answer.

I click on the button to ask a follow-up question and I say, oh, my kid is one meter 30 and he’s seven years old, what’s the best option?

I expect some kind of AI answer that will match that. But all Google does is edit the search query and regenerate the AI answer. So, for example, with my query here, they edited the query to say, best bikes for kids one meter 30, 7 years old, and it gave me the AI search answer.

But it’s based on search results that are not that granular. There’s no language model that tries to understand my query, so it feels very underwhelming and I honestly think nobody’s going to use that because the results are really not that much better when you refine your queries.

I think most people will stick to using Google the same old way, going back to the top search bar to do a follow-up query if they need to.

So that’s basically all I have noted for Google. I know all these changes can be quite scary and it’s quite difficult as an SEO person to both keep an eye on what’s coming and also keep an eye on what’s working right now, which are very different things.

My recommendation is to still focus on what’s working right now and evolve as they actually roll these things out because we don’t know what the final form of these things would be.

According to searchenginejournal article about Google Search Generative Experience: A Look At SGE With 12 AI Overviews, in their final words they stated:

“When a generative AI response appears in search results on a desktop or mobile device, it pushes SERP features like featured snippets, images, videos, people also ask, knowledge panels, and organic search results below the foldWhile some SGE answers seemed to provide less value than typical SERP features, it’s important to remember that this is just the beginning of the future of AI search”

Now, as for my personal impression, after using this version of Google for over 300 queries, at this point, I can’t write anymore and tell you traffic is not going to drop.

That’s not the case. Google is giving an alternative to people. Instead of going to a website, they can use the generative AI answer to get the answer to their queries and some of that traffic that used to go to websites will use that. But I don’t think it’s much worse than the time when they implemented featured snippets, where we went from having zero answers to the SERPs to quite a lot of answers added in the SERPs.

And because they are implementing a two-step process where the user can signify if they want an AI answer by either not scrolling or by clicking a generate AI answer button, I think a lot of people will still choose to go to websites when they are invested enough in a topic to know that the AI answer will not be enough for them.

So I really like this kind of like subtle feedback system that the users can use to signify to Google they don’t want this AI stuff. But even if traffic drops, I think SEO is not going anywhere. The reason why it’s something that people often forget is that the power of SEO is not as much in the volume of traffic that it can deliver as it is in the intent of the visitors.

Read more: Decode How User Search Intent Works?

So if a person is searching for how to lose weight fast, they are looking to lose weight right now and they care enough about it to do a search. And in the future when they click on the website, they will care enough about this to know that a quick list on an AI answer is not good enough for them.

And that tells something about how dedicated they are to solving that problem because they are looking for an in-depth solution. The intent of that person is a lot higher than someone that’s like swiping reels on Instagram or that is browsing Pinterest because they expressed at this very moment in time that they want to take action.

And that’s why 100 visits from SEO will always overshadow 100 visits from Twitter or 100 visits from Facebook because of the intent. So all in all, I think SEO is going to change a lot.

But I don’t think there’s another traffic source that’s yet ready to overtake it in terms of conversion rate. So that’s pretty much the notes I wanted to go through for WordPress SEO Tutorial 2024. I think this post will help you to understand SEO in a better way.

You may also like:
Common SEO mistakes made by beginners 2024
Check why your pages aren’t ranking high on Google

FAQs