Google’s Search Generative Experience (SGE): First Impressions from a B2B Content Marketer
My “artistic” rendition of Google SGE’s initial interface with search suggestions.
If you work in marketing and are even vaguely acquainted with the concept of a search engine, then you’ve heard that Google is testing brand new AI features. Ever since Google’s initial announcement at I/O a couple of weeks ago, everyone has been waiting to see their test “Search Generative Experience” (SGE) in action. First invites got sent out yesterday, and I was lucky enough to get an invitation from Google’s own John Mueller. (Thank you, John!)
Now that I’ve spent a couple hours playing around with Google’s new AI-powered search experience, here are my honest first impressions and takeaways for other content marketers and SEO professionals.
TLDR: Upon initial release, Google’s new Search Generative Experience (SGE), is a refreshing implementation of generative AI in search. Together with upcoming changes to the Helpful Content Update (HCU), SGE stands to shake up SEO and digital marketing more than anybody expected.
Please note that my marketing experience is skewed towards B2B tech, so I am not sure how much my thoughts might apply to other fields.
User Experience of Google’s AI Search
I did not expect to be surprised by any AI product that Google could put out. I’ve played with nearly every other big AI release on the market, and was disappointed in all of them.
But within the first 5 minutes of testing Google’s new AI search… I felt genuine surprise bubbling up within me. SGE didn’t feel like Bard, Bing AI, or ChatGPT. It didn’t comply with my expectations or validate my hesitations. When I first opened that Search Labs page, I was ready to write a complete teardown post. Yet, here I am, struggling to contain excitement.
So… why? What makes SGE different to other generative AI tools? To me, Google’s AI stands out because the UX still feels, fundamentally, like using good old Google Search. The AI is not trying to replace the search experience that we are all used to. These new features feel like they are enhancing other aspects of the search page, like salt bringing the flavor out of a freshly grilled steak.
Here is what an AI-powered Google Search result looks like when searching for our friends at SparkToro.
As you see on the screenshot above, the AI-powered SERP is almost entirely the same as Google’s typical results. However, at the top we get an AI summary answering together with some suggested follow-up queries and illustrated links for further research.
There’s also a small button directly under the search bar that simply states “Converse” with a blue arrow. That label doesn’t refer to custom Converse sneakers, (as much as I’d love some cool SparkToro merch) but instead takes you into a Bard-like chat interface to have a conversation with the generative AI.
The Top Result is Dead, Long Live the Top Result
Alright, here’s what most of you are waiting for: no, I don’t think AI is about to kill SEO.
I won’t deny the obvious: that top AI summary definitely takes up a lot of real estate on the SERP. My earlier screenshot was a relatively short example. Most of your results will look more like this:
Google might finally motivate more users to scroll. Or nobody will click on links ever again. Or something in between ;)
Warning: If there are any SEOs near you, give them a glass of water and remind them that they’re worthy of love. If you are the only SEO in the room, take a few deep breaths counting to five on the exhale. Return to this article only once you’ve recovered.
Yes, the coveted top spot on search results is probably dead. Ranking first or even third will no longer give you the advantage of capturing all the users who are too lazy to scroll down. It’s uncomfortable, and some of us might lose painful quantities of search traffic.
But I don’t think we need to worry about SEO ending just because the top result is now an AI generated summary. Because in actually trying out this feature as a user, I didn’t feel like summary replaced the need to check out the links themselves.
If anything, the AI summary feels like a more polished combination of existing featured snippets and cards. In fact, the AI interface is much better at featuring a wider variety of links instead of giving the spotlight to just one result. I find my eyes drawn to the illustrations on those featured links to the right side much more than to company and website names in the usual text results. Many visual-driven users will probably feel the same way.
Two Options for Viewing AI Search Results
This was a feature that I didn’t know I wanted: SGE lets you switch between condensed and expanded AI summary views.
For example, here’s the condensed AI result for “marketing software”:
The default (and condensed) view for the AI summary with links on the right.
In the default view, the AI summary is on the left side of the screen, while the links that the AI seems to have referenced to generate that answer are kept in a small gallery to the right. The focus is kept mostly on the text snippet, but the links grab your attention with featured illustrations.
Now, if you wanted to figure out where the AI sourced its information or get a better look at those links, you can click that small button in the top right corner of the AI result box. Your results will then look something like this:
The expanded view for AI results emphasizes links to relevant pages a lot more, interspersing them with the text.
In the expanded view, links take center stage. Again, we get featured visuals from each article but now they are accompanied by short summaries and expanded website names. And more importantly, these links are mixed with the AI-generated summary itself. This way, Google shows us where the AI got the information that it’s quoting and gives those original websites a fair shot.
To be honest, I kind of wish the expanded interface was turned on by default. I think it’s better for publishers and can make the summary easier to read for certain queries. But I realize that the expanded view might break up the flow too much for most users, so I might be in the minority.
Some extremely authoritative search results that should be highlighted as much as possible. (Cough. No bias here.)
Always Go Back to Search
One of my main frustrations with Bing AI is how much the chat interface tries to pull you away from the normal search page. Sure, asking complex queries with multi-step prompts is kind of cool. But if I’m on Bing, I’m there to search the internet. I don’t need an AI to take over and chaperone me on my trip to the world wide web. It feels patronizing and unnecessary.
So I was happy to see just how easily Google’s AI guided me back to the familiar search results.
First of all, as I’ve already mentioned – the main search page doesn’t even change beyond placing the AI summary at the top. So as you scroll past it, you will immediately be back to viewing the same results that were there before you ever tried the AI.
Once you scroll past that initial AI card, the rest of your search results will look the same as ever.
You can also enter a ChatGPT-style interface with this Google’s search AI in one of two ways:
Click the “Converse” button at the top of your results page
Select one of the “follow up” options below the AI summary.
Once you enter this window, you might feel like you are using Bing AI or Google’s Bard with the chat window taking up your entire screen:
The chat interface when you ask a follow up question feels a lot more like Bing AI.
However, when you scroll down, Google doesn’t force you to keep interacting with the chatbot. Instead, you are yet again back on the regular-looking search page (although I’ve noticed that the results here might be different from the full SERP):
When you scroll even within the chat interface, you get back to the normal SERP UI.
Some Interesting Behaviors & Quirks
Now, I’ve only used SGE for a couple of hours. I assume that I’ve missed a ton of interesting exceptions, types of results, and features that some of you reading might already be aware of.
However, I have noticed a few patterns within these AI results so I’d like to briefly talk about them.
Paid Ads Appear Above the AI Summary
Here’s an unexpected design choice: sponsored search results are still at the top of SERP, appearing above the AI summary card.
All of you working in paid search can relax. Your ads are still appearing at the very top!
This decision makes sense, since Google gets a fair amount of its revenue from showing these ads. Unlike Elon Musk, most tech executives know not to annoy advertisers by hurting their bottom line.
By keeping the sponsored results at the very top, Google can avoid any tensions with their search ads business. And any of you paid search experts will still get the full worth of your advertising dollars, perhaps even more so given how prominent those sponsored results look above the darker AI box.
Featured Product Reviews from Users and Publications
Google has said a few times that they want to feature real user voices and experiences more prominently in search results. And I can see how they are doing that when I search for consumer product reviews, like the iPhone:
Google’s AI cards for consumer products like the iPhone are extremely detailed.
Notice how on the left Google shows you key descriptions and a category like “good for gaming”, while on the right there are summary cards with highlighted quotes from linked of user and publication reviews. (Is this finally going to kill the listicle post?)
These summaries are pretty cool and filled with tons of interesting tidbits, but I do hope that Google might introduce them for more types of products than consumer goods. For example, I’d love to see a highlight reel for SaaS software featuring G2 and Reddit review snippets.
Competitors Will Outrank Your Branded Queries
This might be one of the biggest shakeups to SEO out of anything in SGE: most searches asking questions about your products won’t feature any posts from your own site.
For example, if you search for any reviews or questions related to how good HubSpot is as a product, none of the featured results will be from HubSpot’s own website:
When you search if a product like HubSpot’s software is any good, you will see summaries from external sources, not HubSpot themselves.
This is pretty cool. My guess is that Google is building off the Helpful Content System to show the most “helpful” and objective results, which necessarily means bumping down anything written by a company about their own products.
This means that if you want proper exposure for your software through SEO, you probably need to invest in some digital PR and get some publications to write about you. I think that’s going to scare a lot of marketers into trying out something new and attempting to build organic brand awareness.
Google’s SGE Doesn’t Want to Write For You… Kind Of?
If you know anything about me, you know that I don’t like most generative AI software.
In fact, we don’t use generative AI to write anything for Kalyna Marketing or our clients. And I stand by that assessment even with all the nifty features that I like about Google’s SGE so far.
We are still not going to implement AI into any content creation or strategy work. I think getting a large language model to write your content or do the thinking for you is a terrible idea, and Google did not change my mind.
However, I did test SGE’s writing abilities a little bit, with interesting results.
First, I asked it to write an email asking for a promotion, which the AI refused to do:
Damn, I guess I can’t write an email asking myself for a promotion.
On the other hand, when I asked it to write a blog post introduction about content marketing, the AI did produce a couple of paragraphs.
But, don’t cancel your ChatGPT subscription just yet. Because Google’s AI makes outsourcing content to it very difficult:
Just when you thought that the AI will write your blog post for you, the AI decides to give you writing advice instead.
The AI wrote a few paragraphs and then gave me tips on how I can write that introduction myself with my squishy human brain. It’s so sassy, and I absolutely adore everything about it.
Google seems to have taken the stance that SGE is a search engine, not a content mill, and I really hope they stick with that decision. I don’t think this content writing functionality belongs in a search engine, and specialized AI is much more interesting in every way.
Am I a Fan of AI Now? Not Really.
As I mentioned above, our agency won’t suddenly change our stance on AI and start using it in our writing or research work. I still don’t like AI writing tools, and I doubt that will change anytime soon.
However, Google introduced us to some pretty exciting possibilities on using generative AI for enhancing the search experience. I view SGE as a search tool, first and foremost. And because of that, these AI features fit in the same category in my head as AI-powered audio transcription tools. I’ve been a huge fan of Adobe’s automatic captions and Otter’s transcription for years. And I can see myself using these AI search summaries on a regular basis too. But in both cases - the AI is not there to replace 99.99% of my daily and work tasks or to assist with my creative thinking.
Google’s SGE reminds me that it’s a search engine at every step of the way. It doesn’t refer to itself as “I”, instead it speaks in third person kind of like Elmo from Sesame Street. The only aspect of SGE that has a “personality” is in how stubbornly it refuses to let me outsource my thinking to it.
So I wouldn’t call myself an AI convert, especially not for content. But I will say that Google’s SGE seems like the start of something very interesting both for SEO and digital marketing at large. We will have to wait and see.
