Comparing search results: Google, Perplexity, Gemini, and ChatGPT.

It seems like discussions around doing SEO and marketing strategies for AI search journeys is picking up steam.

That’s likely because usage is not only increasing for generative AI in general and as a search replacement specifically, but we’re also getting word of more search engine integrations into ChatGPT and other chatbots like Mistral AI.

I still consider myself an SEO strategist, but I also broadened my title and consulting business to do holistic marketing, because I believe the future is less about rankings and more about brand storytelling and visibility through AI search journeys.

In this short post, we’ll take a look at the search results for Google, Perplexity, Gemini, and ChatGPT, through my eyes as a 10-year veteran of SEO. 👀

Our goal is to understand how users might interact with the search results differently, and what that means for measuring content performance in the age of AI.

As for our query, we want to search something with a business case. I used to work with a lot of plumbers early in my career, so I tend to use those searches as examples.

Let’s use, “steps to take to fix a leaky sink faucet.” 💧

Google

The first thing I notice is the truncated AI Overview followed by a featured snippet:

Years ago, the featured snippet would be worth a lot in terms of SERP real estate.

Nowadays, I’d skip past it to expand the AI Overview, where the top results mentioned don’t match the featured snippet:

That’s when we get to individual steps in the process. I can click on each step to find the article I need:

In short, Google is removing friction in its SERP, so I only have to click on one result for my question, and only if I’m unsure of the information.

Having said that, I now have increased trust for the brands mentioned in the AI Overview. If I ask a follow up or need plumbing services, that could be a differentiating factor.

Perplexity

With Perplexity, we get what I’d consider Google’s AI Overview but expanded.

Information from the sources is written out more in the answer itself:

If we do need additional information, we can hover over a source to find it:

However, as I’ve said in my article on how Perplexity works, I don’t think the citations are there for referrals as much as legitimacy.

That said, Perplexity does a nice job with follow-up questions:

I personally use these for content topic research, but they could also be a way to filter users toward more precise information.

That all said, I see users as being less inclined to visit a website here, but they may remember the brand for particular steps.

Gemini

I cancelled my Gemini Advanced subscription, so let’s see what the free version with Gemini 1.5 Flash gets us. 🤗

What’s interesting is I recall Gemini Advanced showing more sources or even images from the web, but I don’t see that here.

In fact, we get a text-only answer:

While the information seems helpful, the user is left with no sources to follow up with.

This is why I think the future is more search-based, like AI Overviews, Perplexity, or SearchGPT (shown last below).

ChatGPT GPT-4o

ChatGPT gives us a text-based answer, similar to Gemini:

The information seems valid, but again, it leaves us with no sources or follow up options.

I will say the formatting of GPT-4o is much nicer than Gemini 1.5 Flash in terms of readability.

ChatGPT SearchGPT

I don’t have access to SearchGPT yet — Update: I found out how to access it! — so I’ll quickly show some screenshots shared by TestingCatalog, then introduce our query.

In the below query for “Detroit Tigers,” we can see the answers are summarized with branded citations mentioned at the end of sentences:

We can also see a list of traditional search results representing highlights from an ALDS Game 5:

For a different branded query of TestingCatalog news, we can see the sources expanded, showing both citations as well as “Search Results”:

Overall, the SearchGPT experience reminds me closely of Perplexity.

That’s why, when we’re anticipating the impact SearchGPT may have on user journeys, Perplexity is the best analog, in my opinion.

Let’s see what we find when testing with our query …

We can see how it gives us a similar answer to ChatGPT but now with sources at the bottom.

We can hover over the sources to view their page titles and branding and click them to visit the webpage:

Admittedly, this experience looks a little different than TestingCatalog’s, so maybe there are more tweaks to come.

That said, I don’t see a lot of reason to click on any sources in the answers we reviewed across all of the search experiences, unless there’s a particular step that’s of interest.

From a content production standpoint, it’s an interesting opportunity:

• Construct content in chunks that can be processed as sections of AI-generated answers.
• Ensure content quality is high (i.e., helpful content) to get inclusion and brand mentions.
• Focus on metrics like time on page or conversions to measure content performance.

Thanks for reading! I plan to make some improvements and expansions in the coming days, so stay tuned!

Until next time, enjoy the vibes:

Thanks for reading. Happy marketing! 🤗