AI visibility / primary bias

You need to stop thinking about how to rank in traditional Google search and start focusing on how to get recommended. In AI search.

Cause pretty soon all search will be AI search. I think it's gonna happen by the end of this year.

And to be clear, this will still be mostly happening on Google. It's just gonna be an AI giving you the results.

I want to talk about the most fundamental shift, cause it seems like most people are struggling to understand this. In traditional Google search, a human performs a search and then they click on one of the top ranked pages.

In that paradigm, the only thing that really matters is how well you rank for those keywords. In AI search, on the other hand, the human performs a search or a prompt, and then the AI breaks that down into a series of searches.

And then instead of considering one or two results where a human would, it will typically consider about a hundred results. And my guess is as these models get more efficient, they're going to start considering more and more results...

you do need your page to be cited. So let's talk about the most important factors for being cited and ultimately recommended.

The first factor is the hardest to influence. It's what's known as primary bias.

This is just what the model already thinks about your brand before performing a search, and it's based on its training data. There's really no easy way to influence this.

But the more content there is online talking about your brand in a positive light, the more likely AI is to site your website. The other badges are much easier to influence.

First of all, you just want to make sure that the information the AI is looking for is clearly displayed on the page. Title of the page should include the most likely term the AI would search for when looking for that information.

And then the h, twos or sub headings should include any secondary terms. And then immediately after the heading you want to directly answer the question.

Maybe the most impactful optimization you can make for being cited in an AI search is having high fact density in your content. What does that mean?...

u're on your own website. Don't say we or our.

Say your brand's name. And then immediately after recommending your brand, back it up with two or more facts.

Share numbers you're proud of, awards you've won, or testimonials from customers. Now, you don't want to turn the entire article into a sales letter, but I recommend doing this at least once at the top of the article, right after addressing the main question, and then in one or two other places if it makes sense.

There are a handful of other factors that affect citation rate and recommendations, but in our experience so far, these are the easiest ones to influence.

Yesterday I read the best overview I've ever seen on AI search. It was from Kevin Indigo's newsletter Growth Memo, and he went over everything it takes to be recommended by large language models like ChatGPT or Google's AI mode.

He broke it down to 11 factors to determine which brands AI Search recommends. And it was so good that I thought I'd do a video on each factor.

He starts by explaining how AI search works. When you ask a question to a large language model, it will usually perform a search on the web.

This starts with a retrieval of relevant web pages. It will then site or look at some of those web pages, and finally it'll decide which of those pages can be trusted.

So for now, we're just focused on retrieval, and we're gonna start with the first factor, which Kevin calls selection rate and primary bias. Primary bias refers to the bias that large language models have before they perform a search.

As you probably know, large language models are all pre trained on data from across the internet...

cusing on hyper specific terms. They tend to be much less competitive.

If you want to be recommended for competitive terms as a smaller brand, you're going to have to get recommendations placed on those websites that are already seen as an authority in your space. So what can you do to improve your primary bias and therefore improve your selection rate?

You need more information about your brand across the internet. This isn't something you can fix overnight.

Make sure you're consistently using the same brand name and the same words to describe your brand, words that people might use when searching for the products or services you offer. Join as many conversations as possible, create as much content around your brand as possible, and get people talking about you online.

Tomorrow, get into Kevin's second factor and by the end of this, you'll know everything it takes to rank an AI search.