AI search is still a brand new industry and those of us who are focused on how to get recommended by large language models like ChatGPT and Google's AI mode are learning as we go I've been doing my own experiments and learning from others for the past few years now I'd like to share a few things I've learned and talk about one thing that I think a lot of people are still
getting wrong I know a lot of people are thinking AI SEO is the same as SEO I've been doing SEO for 17 years and yes a lot of it is the same but we're gonna focus on what's different cause that's where the opportunity is so how does AI search work someone goes to ChatGPT or Google's AI mode they type in a prompt and then the large language model is going to
perform a series of searches this is often called the query fan act from those five to 12 searches it's going to retrieve about 100 pages and then it's going to cite about five to 10 of them and then the final response to the user is going to be a summary of those cited pages along with any primary bias it had before doing a search and so if you want the large
language model to recommend your brand you need those pages that it's citing to recommend your brand and that's all undisputed and I think anyone who's paying attention would agree with that assessment now I would argue that most SEO's are not in fact paying attention but those of us who are understand that if you want the large language model to recommend you you need to be recommended by those cited pages and
so the strategy seems obvious right identify the pages that are being cited and then find some way to get a recommendation for your brand on those pages problem is that's really hard to do typically we can easily get a recommendation on about 20% of these pages these are like commonly cited Reddit threads directories without too much competition and pages from your own website after that you're lucky if you can get
a recommendation on 5% of these highly cited sources and this is the mistake that I think most people in the Geo space are making right now they're putting about 80% of their efforts into getting recommendations on these pages however there's a much easier way instead of targeting the pages the large language models are already citing it's much easier to just create new pages that are better optimized to what the large
language models are searching for and just replace those pages that the large language model is citing with your own pages and then you can place the recommendation yourself the easiest way to do this is to create articles on your own website you can also post articles to third party platforms like LinkedIn or medium you can create your own third party websites and put content there or find websites related to your
industry that are willing to post content in exchange for money or a reciprocal recommendation from your website trust me after about six months of banging our heads against the wall we now spend about 80% of our time creating content for our clients and only about 20% trying to get recommendations on pages that are already being cited
