# New Claude Skill: AI Search Content Optimizer 5 min read Okay, at Nectiv I've been playing around with different skills and automations to improve content structuring for AI search. Over the past year, there's been a ton of research by thought leaders like Kevin Indig, the AirOps team, Chris Green, and others on how to best structure content. Last month we posted an image resource called "[The Perfectly Optimized Page for AEO](https://nectivdigital.com/blog/the-perfectly-optimized-page-for-aeo-ai-search)" that contains best practices on structuring content for AI search. That infographic was based on actual data studies from many of these thought leaders on how to improve content structuring. While this resource is really useful for people, we wanted to make it a little more actionable. So what we did was take many of these data studies and then implement them into an actual Claude skill. The goal is a resource that SEOs, marketers, content engineers, whatever people are calling themselves these days can use to take a page that might not be very well optimized and enhance its performance in AI search. So here it is, an actual resource that you can use to improve your content formatting on existing resources. [**Download The "AEO Content Checker Here"**](https://github.com/Nectiv-Digital/nectiv-claude-skills/tree/main/aeo-content-structure-optimizer) While I won't go into detail about every single check that the skill does (I believe you should customize it to your own use case anyway) here are some of the top things that it looks for when evaluating a given piece of content. ## 1\. Summary Overview At the top of each Google doc, the skill will provide a summary overview that shows the before and after changes for each suggested optimization. You can see adjustments to key recommendations below such as entity density, statistics, definitive language and more. This will show you the areas to consider focusing on the most when evaluating which things to adjust. ![](https://firebasestorage.googleapis.com/v0/b/nectiv-website.firebasestorage.app/o/public%2Fmedia%2F1793392001253-SummaryOfChanges.png?alt=media) ## 2\. Key Takeaways ![](https://firebasestorage.googleapis.com/v0/b/nectiv-website.firebasestorage.app/o/public%2Fmedia%2F1793391387257-KeyTakeaways.png?alt=media) Always remember that 44.2% of ChatGPT citations come from the first 30% of a page. The skill checks that the page opens with a Key Takeaways list of 3–6 bullets and that the main answer shows up early. Source: [The science of how AI pays attention, Kevin Indig](https://www.growth-memo.com/p/the-science-of-how-ai-pays-attention) ## 3\. Entity Density ![](https://firebasestorage.googleapis.com/v0/b/nectiv-website.firebasestorage.app/o/public%2Fmedia%2F1793391463125-ABM_Platforms_Entities.png?alt=media) LLMs like ChatGPT hate vague language. The skill flags vague phrases like "many tools" or "the platform" and swaps in real brands, products, numbers and dates.The goal is to find instances where your content isn't specific and replace it with entity-rich language. After all, LLMs are prediction models that are heavily entity-based. Source: [The science of how AI pays attention, Kevin Indig](https://www.growth-memo.com/p/the-science-of-how-ai-pays-attention) ## 4\. Verified Statistics ![](https://firebasestorage.googleapis.com/v0/b/nectiv-website.firebasestorage.app/o/public%2Fmedia%2F1793391544166-Verified_statistics._.png?alt=media) Throughout many of the data studies I've read, we've seen continued focus on statistics. This is also something that many marketers I've spoken with are emphasizing. The skill will find examples of text that could be verified with statistics, perform web searches, and pull in relevant statistics that you can then insert into the content. ## 5\. HTML Tables Earlier this year, the Nective team did a study where we compared URLs performing in ChatGPT against URLs performing for the same set of queries in Google. We found that ChatGPT citations were 2.3 times more likely to include a table than an individual Google result. When you think about content density, chunking, or whatever you want to call it, it's clear that structured content like tables becomes really imperative. As a result, this skill will look for opportunities to insert one into your content where appropriate. Source: [ChatGPT Citations Are 2.3x More Likely To Include A Table Than Google Search, Chris Long](https://nectivdigital.com/blog/new-data-study-chatgpt-citations-are-2-3x-more-likely-to-include-a-table-than-google-search) ## 6\. Definitive Language ![](https://firebasestorage.googleapis.com/v0/b/nectiv-website.firebasestorage.app/o/public%2Fmedia%2F1793392077703-Definiitive.png?alt=media) This is one that I think a lot of people miss, but in Kevin Indig's study, he found that definitive language is actually really important for whether AI will cite content or not. He discovered that 36.2% of cited text uses definitive language, compared with 20.2% of text that isn't cited. This Claude skill will actually find where you've used softer language that isn't definitive in nature, such as "may consider" or "can help," and replace it with much stronger language. I could use this because I use a lot of this soft language in my own writing. Source: [The science of how AI pays attention, Kevin Indig](https://www.growth-memo.com/p/the-science-of-how-ai-pays-attention) ## 7\. FAQ Content ![](https://firebasestorage.googleapis.com/v0/b/nectiv-website.firebasestorage.app/o/public%2Fmedia%2F1793631547959-FAQ_Content.png?alt=media) Among the most popular tactics I've discussed with marketers over the past year, adding FAQs to content is almost certainly at the top of the list. Nearly every conversation I have, the SEO mentions that this is one of the key initiatives they're working on on-page. This makes sense, as FAQ content is naturally dense and answers questions very directly, but it's also backed up by data. In Chris Green's controlled test, Q&A formatting matched queries best and dense prose. The skill breaks up dense paragraphs into lists and short, self-contained sections, so each part still makes sense when AI retrieval pulls it on its own. Source: [How Content Structure Matters for AI Search, Chris Green](https://www.chris-green.net/post/content-structure-for-ai-search) ## Conclusion Hopefully this helps everyone out there have ideas on how to structure their content better for AI search. The goal of this skill is to integrate into your existing workflows and parse out what you think works best for you. Once again, you can find the full skill on our Github below: [**AEO Content Optimizer Skill**](https://github.com/Nectiv-Digital/nectiv-claude-skills/tree/main/aeo-content-structure-optimizer)