There were three recent articles that highlighted a major issue with AEO: it’s not driving traffic to your website.

I’ll go into more detail about what these articles say, but here’s the high-level summary: AI referrals are small relative to organic search, and the traffic is harder to track.

These studies identify real limits in AI referral traffic and attribution. Even with the rise of “zero-click searches,” SEO is sending materially more traffic to well-optimized websites than AEO.

But what if that isn’t how we should be tracking AEO?

We need to distinguish citations from recommendations and measure whether AI helps the right customers discover, consider, and choose a business.

What these three studies actually show

Marie Haynes: LLM referrals were 0.73% of Google organic volume and Google organic generated 27 times more conversions than all LLMs combined.

The SEO Works: Organic search generated 39 times more traffic than LLMs. The SEO Works: “Organic search delivered 39 times more traffic than AI assistants.”

Kevin Indig: Across 4 brands, he could not establish a reliable link between AI visibility and subsequent branded search or Direct traffic

Zero-click search continues to upend search

In March 2025, a Pew study that focused on the browsing behavior of 900 U.S. adults found that results with an AI Summary click on links about half as often as results without a summary. Additionally, links inside summaries received clicks on 1% of visits.

This does not mean SEO is irrelevant either, but it does speak to how people are developing new ways to research and make decisions.

Are AEO and SEO in conflict with each other?

Where the articles argue that traffic is overwhelmingly coming from SEO, that is not how decisions are made. SEO is part of the decision-making process, alongside social media, PR, brand, and more.

SEO has the benefit of being easy to track. There are tools to track which position you are in for each keyword, how much traffic you’re getting from search, and even how many leads originated from search. AEO can’t be tracked the same way.

This isn’t new, though. While the internet made attribution easy for many channels, channels like social media could work well while remaining difficult to connect directly to revenue.

While we accept this for certain channels, AEO is being compared to SEO because they are both “search.” But this is like comparing paid social to organic social. They have a lot in common and many companies use both, but you would never track them in the same way.

How does the user experience change from SEO to AEO?

If you look at how SEO drove revenue, it worked like this: Someone searched for information on a topic, and Google served relevant websites. If you worked hard enough, you became the first suggestion. Once someone was on your website, you could try to capture a lead, retarget visitors, and, most of all, control how you presented the answer they were looking for.

That last point is important.

If you owned the first position, you controlled the authority on a topic and were able to share your point of view on that topic. This was and is a highly effective form of advertising.

AEO works differently.

With AEO, someone is still searching for an answer, but instead of viewing your website, their entire experience is inside a chat window and every interaction is being personalized.

If you search for “What is the best CRM?” on Google, you and I will likely get the same results (albeit different ads). If we both enter the same prompt into ChatGPT, we will likely get different responses.

LLMs use context and memory for each interaction, so while a generic question will get you a generic response on Google, asking the same thing on ChatGPT may give you and me completely different answers because of the context it has about us.

For example, if ChatGPT knows you are working on a project for a Fortune 500 company, it’s going to consider different elements than if I’m working on a similar project at a growing startup. It may recommend Salesforce in one case and HubSpot in the other.

Being cited is not the same as being recommended

While the goal of SEO is to drive traffic to your website, AEO has a different goal. Yes, AEO can drive traffic to your site, but that traffic comes from citations. As the articles above outline, citations drive a marginal amount of traffic.

Coming back to the basics of SEO, you can rank on page one for a keyword and still get little traffic, either because you rank near the bottom or because you’re targeting the wrong keyword. This is similar to citations. You can be cited in an answer to a prompt that recommends your competitor.

The real goal is recommendations. You don’t need someone to come to your website if the AI tool recommends you directly in chat. You’ve made it to the shortlist.

But if you try to measure AEO in the same way as SEO, you will always undervalue the impact of those recommendations.

How do you measure recommendations?

There are many tools to track SEO and even citations, but there aren’t many ways to accurately track recommendations.

One way to do this is to turn off your history and memory and ask realistic buyer questions across the different tools and models. Keep track of the answers over time. This isn’t perfect, but it can provide insight into how you are being positioned today by these models.

Second, you need to use customer feedback to understand how new customers are finding you. Ask them about their buying journey to understand whether AI was part of it. This is considerably more work than looking at a report in Semrush, but when direct attribution isn’t possible, this is one of the best ways to understand what is truly influencing your customers.

Keep what works for SEO but change the success criteria

The technical aspects of AEO and SEO have a lot in common. The meta elements of your website work the same for both. You need to create content for both.

The difference is that, for AEO, your entire brand halo is important. While a Reddit post might not make it to page one of Google for your keyword, the same post could be instrumental in shaping an AI recommendation. The breadth is greater.

As marketers, we need to get comfortable again with not having full attribution while finding ways to track what is affecting our businesses.

In the end, you don’t need AI to replace every click you used to get from Google, but you do need to know whether it’s helping the right customers choose your business.