Latest / Found in AI: AI Search Visibility, SEO, & GEO / How Can You Make Your Content Show Up in AI Search Results?

Transcript

How Can You Make Your Content Show Up in AI Search Results? · Found in AI: AI Search Visibility, SEO, & GEO

• 0:41
  (Transcribed by TurboScribe.ai. Go Unlimited to remove this message.) Every content marketer is asking the same question right now.
• 0:45
  What actually works in AI Search?
• 0:48
  If we haven't met yet, I'm Cassie Clark.
• 0:51
  I'm a content strategist who helps startups and B2B brands get seen.
• 0:56
  And now that AI engines are part of how people search, that means creating content that shows up in AI-generated answers.
• 1:03
  In today's episode of Found in AI, I'm joined by Alex, co-founder of SEO Forge.
• 1:09
  He's been deep in the trenches testing how chat GPT, perplexity, and other LLMs serve as content, and what startups can do to actually earn that high intent traffic.
• 1:21
  We dig into what's different about AI Search, why unique insight matters more than keywords, and how to structure help centers so LLMs pick up and cite your brand.
• 1:31
  Here's part of that conversation.
• 1:33
  So I've seen a lot of people talking about like the traffic that comes from these AI search engines, I guess is the right way to call them, that those are like your high intent customers, like the ones that will convert.
• 1:48
  Did you just flat out ask them how they found your brand?
• 1:51
  Like, how did you know that they came from chat GPT and not from Google or wherever else?
• 1:57
  Yeah, so we did two things.
• 1:59
  So at first, we just saw in Google Analytics, we had first analytics as well.
• 2:04
  So we could see people directly coming from chat GPT.
• 2:07
  And we had pretty good tracking setup.
• 2:10
  So we could see, you know, if someone came from this source, would they sign up? Would they convert?
• 2:16
  So we directly saw who those people were.
• 2:19
  But secondly, we also added a question in our onboarding, where we asked people, how did you hear about us?
• 2:25
  And we just added an option immediately that said, chat GPT.
• 2:30
  Honestly, we thought about saying just like LLM, AI search, but we saw only traffic from chat GPT at that time.
• 2:36
  So we just added an option for chat GPT.
• 2:38
  And that's how we knew they're coming from chat GPT.
• 2:42
  Yeah, so chat GPT seems to be the one that I use the most.
• 2:45
  Are you seeing traffic from Perplexity and all the other places?
• 2:49
  I would say Perplexity is the second one, but it's still very far away from chat GPT in terms of just quantity of traffic.
• 2:56
  And then I've seen Gemini as well.
• 2:58
  And when it comes to Cloud from Anthropic, I have seen traffic from it too, but only really towards deep tech startups and companies.
• 3:09
  So not really like, let's say, a healthcare company.
• 3:12
  I haven't really seen Cloud traffic to them.
• 3:15
  So that must be industry specific.
• 3:19
  Exactly.
• 3:21
  Yeah, yeah.
• 3:22
  So tell me about how SEO Forge works.
• 3:25
  Like, I know Google Analytics, it seems similar, but this is with AI search and not just Google.
• 3:33
  So tell me how it works.
• 3:35
  Yeah, so what we've done, you know, is we've taken the experience we've gathered over the past seven months into understanding, like, just outside any platform, like what strategy works, what kind of content do you need to push?
• 3:49
  Is it different from just normal SEO and so on?
• 3:52
  And we took that and then we combined it with the fact that we had a few platforms that we've developed before, where basically people, you know, they can talk with multiple LLMs at the same time.
• 4:04
  So like chat GPT, Gemini, and so on.
• 4:07
  And we kind of understood how people use LLMs to get their questions answered.
• 4:14
  We could see whether they click citations or whether, you know, they just copy the name of the company and then they paste them to Google.
• 4:21
  So we had some insights into that.
• 4:24
  And the platform that we came up with is basically we create a set of prompts for you.
• 4:30
  And we do that by, you know, scraping Reddit.
• 4:32
  So, you know, finding user generated prompts.
• 4:36
  We look at LLM conversations and try to get actual prompts and questions that people ask from there.
• 4:42
  And we also try to tie it in with, you know, keywords from Ahrefs and long tail keywords and so on.
• 4:47
  And we track those on a weekly basis.
• 4:50
  And the way we do this, we don't do it through the APIs.
• 4:53
  We actually try to, you know, see how a user would see it on the interface.
• 4:58
  And then when we give you the responses back, we actually try to analyze the responses and tell you there's, you know, maybe you have a content gap, like why is this competitor mentioned and you're not mentioned here?
• 5:09
  Or why is this competitor citing or not?
• 5:12
  So that's kind of the initial thing we do.
• 5:15
  And once you ramp up content, if you don't have any content or if you already have content, we actually try to create a set of prompts just around that piece of content to try and understand what's driving traffic to that piece of content specifically.
• 5:29
  And we try to adjust the content based on responses from LLMs and best practices.
• 5:34
  We just wrapped up a study on about 500,000 user queries and we executed them on ChatGPT and Google Search as well.
• 5:44
  So we found some differences between what ranks on Google and what ranks on ChatGPT.
• 5:48
  So we usually give our users just the list of instructions where, hey, those are the things you need to do.
• 5:55
  We want to get more traffic from ChatGPT, so we just try to help them using the platform to try and get more traffic using those strategies.
• 6:04
  So are you seeing a big difference in what's working for Google and what's working for ChatGPT or the other LLMs?
• 6:12
  There are differences, I would say.
• 6:14
  What's interesting to me is, and this has been circulated around a lot, but on Google you see 10 blue links, right?
• 6:24
  And then on ChatGPT, you just see a long piece of content.
• 6:29
  Sometimes there could be 16 citations, sometimes I see five, so it's really different.
• 6:34
  But most of the people, I would say they get what they're looking for directly within the content from ChatGPT, so they don't need to go beyond it.
• 6:43
  And what we found works best when it comes to getting more traffic from ChatGPT, but not necessarily from Google, is having unique insight.
• 6:52
  So imagine on Google, someone could ask a very broad question or just type in a keyword that's quite broad, and then you have a bunch of pages ranking just because they've optimized for so long directly towards that keyword.
• 7:04
  But in ChatGPT, we've noticed that the longer a conversation goes, the more specific the questions get.
• 7:11
  So if someone asks very, very specific questions, that could mean they have very high intent in finding a solution.
• 7:17
  And if you have that insight that could help them, it doesn't matter if you've optimized towards that keyword.
• 7:22
  If ChatGPT crawled you, you know, if the crawlers from OpenAI crawled your content, ChatGPT is very likely to cite you.
• 7:30
  So we usually push people to really have unique content and to bring unique perspective to content, so we don't advertise or push for AI slop or nothing like that generic content.
• 7:42
  And apart from that, there's also, but this has been circulating in the SEO world for a long time, but doing technical SEO properly, so using schema, having proper H1, H2, H3 headings, having an FAQ, I saw your post, by the way, that mentioned that.
• 7:59
  We also push for, you know, a larger FAQ is just the help center.
• 8:03
  So we've seen great results from help centers too.
• 8:06
  So we push people to do those things and to bring unique data and insights forward.
• 8:12
  And if they don't have it, to go and try and get it from their users or carry out surveys, you know, things like that.
• 8:18
  And that's been working really well for ChatGPT.
• 8:20
  But of course, for Google, it takes a bit longer to rank just because, you know, you can't just have a one day old domain that just does really well on Google.
• 8:29
  Right.
• 8:29
  So there's a couple of different things I want to hit on there.
• 8:32
  So I read a post, I don't remember who posted about it, where they were changing the name of their blog from just linking the blog at the top to like content hub or knowledge user base or something like that.
• 8:46
  So that's interesting, but it's still linked to their blog.
• 8:50
  So you would suggest just having a whole separate entity on your website that was just FAQ with like deeper things, like here's one article on, I don't know, how to use our product.
• 9:05
  And then you go into another one.
• 9:07
  So it's almost, in my mind, it's almost like you have two blogs, but one is just dedicated straight to just how to.
• 9:14
  Am I thinking about that correctly?
• 9:17
  Yes, you are.
• 9:18
  So basically, the way the Help Center is structured differently to just the blog, let's say, is that in the Help Center, all of the titles are just questions.
• 9:28
  And there's only one question per article.
• 9:30
  So there's no like, we don't have, we haven't incorporated H2s or H3s.
• 9:34
  It's literally just that one long form answer.
• 9:38
  And it's basically things that you wouldn't normally answer within a blog post.
• 9:42
  So if you want to dive deep into what your platform is capable of, or if you want to, let's say, talk about the user pain point in detail, and let's say you don't really cover that on a blog post already, but you want to have that content quickly, we recommend adding it to the Help Center.
• 9:57
  That's a really neat idea.
• 9:59
  And I'm thinking about how I can do that for my own website.
• 10:01
  I'm not sure that'll take some more planning.
• 10:04
  But you're seeing like the clients that you work with who are doing this, they're seeing results from that already?
• 10:11
  Yeah, that's been working really, really well.
• 10:13
  The only thing that I would say to keep in mind, for you and for anyone who will watch this, is there's lots of tools out there and platforms that allow you to build Help Centers.
• 10:24
  But most of those tools force you to host the Help Center on a subdomain.
• 10:29
  So like help.domain.com.
• 10:31
  And that actually, while the Help Center leads to quite a bit of tragedy traffic, that traffic is not as useful because you can't really convert people from going to the Help Center to your website or to your intended button or action.
• 10:47
  So we usually recommend, even if it takes a bit more time, having that Help Center directly on your website, similar to how you want to have a blog deeply integrated into your website to convert more people.
• 10:58
  That conversation gave me a lot to think about.
• 11:01
  Especially the way Help Centers can pull in high intent traffic if you structure them the right way.
• 11:07
  Over the next few weeks, I'll be adding in some of those help desk style articles to my own website and testing how they perform an AI search.
• 11:16
  I'm Cassie Clark, a fractional content strategist who builds AI search ready content systems for startups and B2B companies.
• 11:24
  If you want to talk about how we can make your content work harder in this new search landscape, connect with me on LinkedIn or visit cassieclarkmarketing.com.
• 11:33
  Next week, I'll share what I know about how AI search picks the answers to service.
• 11:38
  Stay tuned, I'll see you then.