@tjrobertson52

If you want to rank in large language models like ChatGPT, you need to be using software to track the sources that ChatGPT is using. I spoke with the product manager of one of these pieces of software yesterday and he says that they currently have 92 competitors, 92 other companies doing the same things they are. Over the last two months, I've looked into about a dozen pieces of software that does this.

I've used or done demos of about half of those. I wanna tell you what you need to look for if you're interested in starting to use this software and then I'll tell you which one I'm using. And this is not a sponsored post.

No one's paying me to say this.

There are two main features you're looking for in software like this. You wanna be able to track how often you're recommended by these large language models versus your competitors. And more importantly, in my opinion, you wanna track what sources the large language models are referring to before making that recommendation.

Getting your brand mentioned in those sources is the primary way you increase your visibility in these large language models.

So what you're actually tracking is a set of prompts, prompts that you think your ideal customer might be typing into these large language models. And we really just have to guess here. While there is some data about what people are typing to large language models, pretty much every prompt is unique.

This isn't like keyword research where you can see search volume for different terms. The search volume for every prompt is one. So what most of these tools will do is they'll look at your brand and they'll guess the kind of prompts people might type into a large language model before it recommends you or your competitor.

It then needs to run those prompts multiple times across the most popular large language models, and I say multiple times because the results are gonna be different every time. You can't just run this once like you would when you're checking keyword rankings, and you want a tool that can at least check ChatGPT, perplexity and Google's AI overviews. If it checks other models, great, but you don't wanna water down the results with models that people rarely use.

Here's one thing that a lot of people don't realise: most of these software companies are getting the results through the API of the large language model, and that's actually an issue because you get very different results through the API and most people are not querying the API, they're using the product directly.

So it's important that the software you're using is looking at the same results a user would get when using the product directly, and then most importantly it needs to be able to aggregate all the sources that the large splicing models are looking at to show you which sources are used most often, and not just the domain but the actual page that was referenced. So these are gonna be directories, listicles, discussion threads, the specific Reddit thread that ChatGPT is looking at before making a recommendation.

Blog posts, YouTube videos. This is content that you might want to replicate on your own channels or website.

This is the real value in the software. It'll tell you exactly where your brand needs to be mentioned or what kind of content you need to create on your website to increase your visibility. Now most of the software that exists right now is still incredibly expensive or not quite working.

The one that I found that I really love is peak dot a I P e e c dot a I. While the software still isn't quite complete, it's already been incredibly useful for us and it's not nearly as expensive as some of the competitors like profound.