The AI Effect: How ChatGPT and Gemini Are Changing Search

Remember when search queries looked like, “best time LinkedIn post”?

Now, it’s:

“What’s the best time to post on LinkedIn if I’m targeting HR professionals in Canada and my posts are mostly thought leadership content?”

And while that might just look like more words at first glance, there’s so much more to it. The new search query reveals a more complex thought process, a more nuanced expectation, and a very different kind of user. And much of this is by design.

Think about a visit to Google or Bing. The page is usually empty, with a small search box and a few buttons. There’s no prompt to get you started; the whole thing is very functional. It doesn’t visually encourage you to converse or input a large query.

You input your normal, basic query into the box and are instantly presented with various possible answers. If it’s done its job right, you’ll click on the result that seems to do the best job of telling you what you want to know, and you’re off. You never have to see Google again for that query.

Google has used classical conditioning to train us to search this way, too. If you enter a query into the search box and receive a bunch of useless blue links in return, the bad results act as a negative reinforcer. You instinctively know to reformulate your query and try again.

ChatGPT Pushes Users to Be Different

Now, think about the design of ChatGPT. While it lacks the classical conditioning and behaviours that have become instinctual to anyone with an Internet connection, it’s designed in a way that doesn’t require the same classical learning.

ChatGPT has a large input box with all sorts of options, encouraging you to customize and interact in a format that works for you. Essentially, it’s meeting you where you’re at, attempts to work the way you do, and encourages conversation through preemptive and post-interactive text.

“Certainly! Here is your…,” it says before delivering (what you hope, at least) is the information answering your most burning questions. When it’s done, it asks if it can find/create/build/tell you about…,” encouraging you to explore and direct the search in the way that works for you. 

In this way, LLMs have taught us to ask better, fuller, and more personal questions. And that means content must now respond like a conversation, not just answer like a textbook.

LLMs make exploration effortless, changing our expectations of search entirely. You don’t have to click five tabs, cross-reference content, or wade through fluff. You ask a question and get a tidy little summary. (Whether these summaries are accurate or not is questionable, but I’ll get to that in a minute.)

As you might expect, it has birthed a new type of searcher: The Passive Explorer.

Today’s LLM searchers want to learn, not commit. They skim. They bounce. And they’ll maybe come back later if they feel like it. So, if your content greets them with a pop-up, gated PDF, and three blinking CTAs? They’re out.

The rise of LLMs means your site is no longer the first place users learn about a topic. They’re not landing on content intended for the awareness or consideration phases of the buying process.

Instead of being the star of the show, your site is now the follow-up. And this means you’re no longer delivering the awareness or the curiosity spark. Instead, you’re expected to deliver the deeper, more useful part of the journey.

And THAT changes everything. Or, at least, it should.

People aren’t just asking “What is X?” anymore.

They ask, “How do I use X to fix my problem?”

LLMs have made us more strategic. Users expect frameworks, workflows, and personalized advice. That means content today needs to function as a mini strategy session, not just an info dump.

What’s the new gold standard? Think: Playbook meets therapist meets explainer.

And given that LLM searchers have already jumped in and out of the awareness and consideration phases, they will likely land on your site a lot farther down the conversion path than they would have previously. When ChatGPT users land on your site, they’re looking for what we traditionally described as Bottom of Funnel content, at the point of conversion.

That point of conversion might be the actual sale of a product or service. But it might also be the “sale” of information, where the user can go deeper into a topic, be more specific, get a unique story or insight, or learn about nuances that LLM tools just aren’t equipped to handle.

Evidence, while only anecdotal at this point, already points to this change in user behaviour. Traffic from ChatGPT and other AI search tools is arriving on-site farther down the funnel, viewing fewer pages, learning less about whatever it is on your site, and going straight to the conversion. So, while there’s less traffic sent by generative AI overall, it’s proving more valuable.

A new study by XFunnel further proves my point. Their report found product-related content makes up to 70% of all the sources referenced in their responses and 56% of Top of Funnel searches.

For you, this means providing the information users search for at earlier stages of the buying process on conversion pages to get included in AI responses and ensure consistency across platforms to bring the user to the conversion point. Publishing other pages about your industry just isn’t as useful, particularly the high-level, generic overviews you’re used to creating.

In a world where ChatGPT summarizes everything and doesn’t always show sources, users have stopped caring who said it or where the information originated. Today, LLM search users care about how it’s said.

Clarity over credentials.

Relevance over reputation.

Flow over flash.

And while it’s concerning and even enraging for publishers, the obfuscation of sources is by design and done for a reason. It introduces friction into the fact-checking process, pushing users to accept the technology’s responses at face value (for lack of a better term).

Why?
 

You’re more likely to use the technology if you trust it, and if you don’t go anywhere to fact-check, you’ll spend more time using the platform. From ChatGPT’s (Gemini’s or other platforms) perspective, this is just good business. All they have to do to make you keep using the technology is to provide answers that are “good enough” to get you by and not cause any significant harm.

This is bad news for publishers who cannot inject themselves into the process. But it also means anyone creating content is going to have to adapt. Your brand might be a household name, but if your content reads like it’s trying to impress an algorithm instead of help a human? It’ll feel cold. Untrustworthy. And yes, your audience will notice.

A concerning side effect of the move toward AI and LLM search is that users confuse their own confidence in the accuracy of the information they receive. The information AI provides looks correct and authoritative. This technology even makes up supporting evidence and plausible explanations to make darn sure the answers look legitimate.

And users buy into it. There are a million examples of people utilizing ChatGPT’s outputs in everything from legal arguments to government policies without ever having fact-checked or scrutinized the content. Sometimes, they don’t even bother to remove the tech’s friendly introductions.

The confusion between confidence and accuracy is even more concerning because hallucinations are a maladaptive trait of the technology; it may not be possible for LLMs to stop making things up. This means that the number of users relying on inaccurate information will only grow. And with no feedback loop built into the process, someone might only recognize the technology has lied to them when something goes horribly wrong.

On the flip side, LLM hallucinations can also be very bad for publishers and website owners. To explain what I mean, let’s walk through an example:

Pretend you’re a B2B marketer who has just asked ChatGPT about CRM software for an hour. It’s told you about all the wonderful things CRM software can do these days, including a new core feature that all good CRM have these days: It provides you with a list of all the software a contact has on their computer. Just what you need to make more sales!

So, you dig a little deeper and ChatGPT tells you all about how this magical feature works. And if you want to use this fantastic feature, ChatGPT tells you that JackSmartCRM is your best option. So, you go over to the JackSmartCRM site, make the costly investment, get it all connected, and set up only to discover that it doesn’t have the lists of software you were expecting. In fact, this CRM is worse than the other one you were using. Now, you’re out a lot of money, upset at JackSmartCRM for not doing what you were told it would do, and have to try to get a refund.

While this may be a made-up example with some exaggeration, users are already experiencing this on a grand scale. And while it’s not JackSmartCRM’s fault that ChatGPT lied, it won’t stop consumers from blaming it anyway, or at least seeing the company in a less favourable light.

While the situations I’ve shared here are more of a cautionary tale for publishers and marketing professionals of all stripes, they provide us with hints of how to adapt to the new world of search that’s barreling towards us. Most importantly of these is the power of credibility and trust.

There is no longer any room in the digital world for information and sites that fail to make their content appear credible and trustworthy. 

If a user asks Google a question and the new AI-powered search result already gives them a decent overview, why should they click on your site?

Your content has to earn the click. And if it’s going to do that, it needs to offer:

• More depth.
• More specific relevance.
• A unique story or insight.
• A tone that feels human, not just accurate.

If your blog post or landing page can’t beat ChatGPT’s summary in usefulness or emotional resonance, it won’t survive the scroll. It’s just that simple.

AI is helpful. Sometimes even brilliant. But it’s still emotionally tone-deaf. If your content doesn’t feel like it was written by a human who’s been there, worried that, tried that, and failed that, then you’re leaving your strongest advantage on the table.

Empathy is your SEO edge and marketing differentiator. It’s also what earns trust in a world where LLMs dominate the surface-level stuff.

You evolve. You stop writing for the algorithm because the keyword tool told you to and start writing for human readers who are looking for a guide or a conversion point of some kind.

You make your content:

• Conversational – Because that’s how people search now.
• Exploratory – So users can skim, scroll, or dive deep on their terms.
• Emotionally intelligent – Because AI can’t connect like you can.
• Strategically valuable – Because people want to act, not just learn.

And if you’re trying to wrap your head around how to actually do that?

It’s a deep dive into how real people make decisions, what makes content resonate, and how you can build marketing strategies that actually work in today’s AI-shaped world. Think: empathy-driven SEO, modern funnels, keyword research with a human lens, and content that earns trust, not just clicks.

Because in the age of ChatGPT, the best content isn’t the smartest.

It’s the most helpful, human, and honest.

Want to be the brand people trust after they close ChatGPT?

Start here.