AI Mode is a new experimental feature, currently available in Google Labs. It’s quite possible that AI Mode becomes the regular way that people search. 

Let’s have an early look at AI mode, understanding that it is currently an experiment and will change over time.

It turns out that the websites shown in AI mode often differ from traditional search, or are shown in a different order. Let’s talk about why this happens and brainstorm on what we can do for our sites to improve our chances of being recommended by AI Mode. 

AI Mode uses a "query fan-out" technique unlike traditional Search

In traditional search, a user submits a query, and Google retrieves results from its index of web pages that are likely to be relevant, helpful and reliable. AI mode acts in a different way. It provides an AI Overview, similar to what we currently see in Search, but uses what Google calls a “query fan-out” technique to decide which websites are shown. This technique means that Google is bringing together the results from multiple related searches into one comprehensive answer. 

Instead of finding results relevant to one query, the system is searching for multiple queries and merging results together.

At Google I/O in 2024, Google told us that soon AI would do the searching for us. This is what’s happening with the search results presented to you in AI mode. 

Google gives us this example query: “What’s the difference in sleep tracking features between a smart ring, smartwatch and tracking mat?” Google’s current AI Overview gives a different answer to this, and references different websites than does AI Mode.

Traditional Search: The AI Overview discusses each of the sleep tracking options and it is up to the reader to explore further and make their decision.

 

 

AI Mode: With AI mode, the AI answer pulls the information together for me, summarizing it all and giving me much more helpful advice to assist my decision making:

 

 

This seems to me to be similar to the agentic Deep Research offered in the Gemini chatbot, albeit much faster.

It’s worth noting that the websites shown in the AIO in traditional search and the AI mode answer are different:

Traditional Search:

There are several other websites listed as well.

AI Mode:

If I click “Show all”, AI Mode then gives me a list of blue links that start with those three options and show me more.

What I think is most interesting about this is that the websites shown are different. Because AI mode is summarizing results from multiple queries, it means that AI Mode ranks websites differently than traditional search.

Here is another example:

What is the latest news from OpenAI?

In traditional Search, I get a list of websites I can browse through. I can click on OpenAI’s blog, read a story from the Verge, or choose from multiple top stories. In order to get a recap of the latest news stories, I need to browse, click, return to search and click some more.

 

In AI Mode, Google does the searching for me to give me a summary of the latest news, which is quite different.

Again, different websites are shown as we see in traditional search. Many are the same, but the order is different.

I do think the information above is accurate. However, some of it is not completely up to date. For example, AI Mode tells me OpenAI will soon release GPT4.5, yet this has already happened.

Takeaway for SEOs: AI Mode uses a different way of ranking results. We will need to spend time studying which pages are shown for our clients’ queries. However, this will be difficult because “queries” have turned into conversations. AI mode gives us the opportunity to ask followup questions. It would surprise me if we see these as impressions and clicks in GSC. 

How do we optimize for AI mode?

We will need to do a lot of analysis and testing! It seems to me that having a presence in AI mode comes down to the following:

  • Being known as a go-to source or authority on your topics.
  • Thoroughly understanding the questions your audience has and their intent.
  • Answering these questions in a comprehensive way using verbiage that is easy to cite or summarize in AI answers.

I do think it's possible that we can, to some extent, use a knowledge of vector search and perhaps even play with cosine similarity to create content that's more likely to be used as an AI mode reference. But ultimately what will be most important is that you are consistently a source that people prefer and enjoy for your topics.

Will AI mode replace traditional Search?

I think it will. And AI mode does as well.

I think that Google will eventually add much more to AI mode including searching with images and voice and then make it the default, giving people the option to return to "classic search." From my own brief time having access to AI mode I find it very helpful. Also, it does seem that AI mode is more accurate than the answers given in AI Overviews. I'll share more about that in another post.

I am certain this was Google’s plan all along - to have an AI driven answer engine. 

If you enjoyed this, you'll love my newsletter!

Join us in the free SEO & AI News Area of the Search Bar