Skip to content
Go back

The AI SITE Framework: 6 On-Page Tweaks to Get Cited by ChatGPT and Google AI Overviews

Getting cited by ChatGPT and Google AI Overviews is the new SEO game. And unlike traditional link building, you don’t need a massive domain rating to win.

Jesse Schoberg, CEO and Co-Founder of DropInBlog, shared a framework at a recent conference that helped 60% of their 2,000 customers get more AI mentions within a single week. The changes are all on-page and doable in an afternoon.

The framework is called AI SITE, and each letter represents one optimisation you can apply to your existing content right now.

Table of contents

Open Table of contents

What is the AI SITE framework?

AI SITE is a six-part on-page optimisation framework designed to make your content more likely to be cited by AI systems like ChatGPT, Perplexity, and Google AI Overviews.

The core insight is counterintuitive: AI models are worse than Google at interpreting unstructured content. The more you spoon-feed them clear, structured answers, the more likely they are to cite you.

“GPT is like dumber than Google as far as analysing stuff. The more structured your data can be and the more you can spoon-feed it the answer, the more likely it is to give it.”

Jesse Schoberg, CEO & Co-Founder of DropInBlog

This means old-school SEO fundamentals — proper heading hierarchy, structured data, clear formatting — matter more than ever. And critically, domain rating is less important for AI citations than it is for traditional search rankings.

Here are the six elements:

LetterElementWhat It Means
AAnswer FirstLead with the direct answer, not a fluffy intro
IIntent-Matched HeadingsUse descriptive H2s/H3s that match search intent
C (first)Clear StructureFormat content as lists and tables, not paragraphs
I (second)Index SchemaAdd FAQ, HowTo, and other structured data markup
TTrusted SourcesLink out to authoritative sources to back up claims
EEntitiesCreate named frameworks and proprietary concepts

A — Answer First

The old approach to blog writing started with a long introduction before getting to the point. AI models want the opposite — they want the answer immediately.

Start your post by directly answering the query in the first sentence or paragraph. Then expand with the full post below.

For example, a coffee brand ranking for “does decaf coffee dehydrate you?” opens with: “In short, no.” Then the detailed explanation follows. That post now gets cited in both Google AI Overviews and ChatGPT.

One DropInBlog customer, My Keys to Music (a DR6 site), uses both a “TL;DR” and a “Short Answer” section at the top of every post. The result? They consistently beat much larger sites for first-answer citations in ChatGPT.

I — Intent-Matched Headings

Stop using generic headings like “Introduction”, “Step 1”, “Step 2”, and “Conclusion”. Instead, make every H2 and H3 descriptive of what the section actually covers.

AI models rely heavily on heading structure to understand what a section is about. If your heading says “Step 2” the model has to read the entire section to understand the content. If your heading says “How to choose the right yoga mat thickness” the model can immediately match it to a query.

Schoberg shared an example of a hair product site (DR16) that uses descriptive headings throughout their table of contents. They consistently outrank established cosmetic brands in AI citations.

C — Clear Structure

This is what Schoberg calls a “silly easy win”. If any part of your content contains a list, comparison, steps, or ranking — format it as a bullet list or data table instead of burying it in a paragraph.

AI models are far more likely to cite content that is already structured as a list or table because it requires less interpretation.

One customer runs a home yoga app and was trying to rank for “best app for beginner yogis”. They created a comparison table on one post and a feature list on another. Both posts now show up in ChatGPT responses alongside VC-funded competitors — and ChatGPT cites them as the primary source.

I — Index Schema

Add structured data markup (schema) to every post where it applies:

Most CMS platforms have plugins or built-in support for this. The pattern is the same as the rest of the framework: you’re feeding AI models pre-structured data so they don’t have to do the analysis themselves.

The My Keys to Music example is again instructive here. With FAQ schema on every post, this DR6 site became the primary source ChatGPT cites for queries about Nord music courses — ahead of sites with far larger SEO teams and domain authority.

T — Trusted Sources

If you cite authoritative sources in your content, AI models are more likely to cite you in return.

Link out to credible references — studies, official documentation, industry reports — to show where your data comes from.

Schoberg shared an example of a supplement brand that links out to PubMed studies for each ingredient they discuss. Their content is now the one getting cited in AI Overviews for ingredient-related queries.

This aligns with what we know about E-E-A-T signals in the age of AI search. Demonstrated expertise and sourcing build the kind of trust that gets you cited.

E — Entities (Create Named Frameworks)

This is the most creative element of the framework. If you create a named concept, methodology, or framework and structure it clearly on your page with proper schema, AI models will pick it up and cite it.

Schoberg practised what he preached here. He created the “AI SITE” framework, published it as structured content, and then asked ChatGPT: “Are there any frameworks for getting mentioned by AI?” The result? AI SITE appeared as the number one result, above GEO.

“You put that together and you put it into a nice structured content on your page and you put your schema on that blog post… and there we are. Number one, AI SITE, above GEO.”

Jesse Schoberg, CEO & Co-Founder of DropInBlog

The action plan

Here’s what Schoberg recommends doing next week:

  1. Identify your money articles — focus on posts that mention your brand or product, not generic informational content. Think “best yoga apps” not “what is yoga?”
  2. Add lists and tables — find anywhere content is buried in paragraphs and restructure it
  3. Add FAQ sections — and make sure the FAQ schema validates
  4. Push citations — link out to authoritative sources throughout your content
  5. Fix heading hierarchy — ensure H2s and H3s are descriptive and follow proper nesting order (H2 then H3, not random)

These are all updates to existing content. You don’t need to write new posts. Just restructure what you already have.

Bonus: Repurpose your top articles as YouTube videos

Google owns YouTube. As AI Overviews take over the top of search results, Google faces the same problem everyone else does — where do the ads go?

The solution: Google is now transcribing YouTube videos and using those transcriptions as sources for AI Overviews. This means your money articles should also exist as YouTube videos.

Schoberg’s own company does this — they create a video for every integration tutorial. For the query “how to add a blog to Thinkific”, the AI Overview citations pointed to their YouTube video rather than their blog post.

Creating YouTube versions of your best content gives you a second shot at being cited, and Google has a financial incentive to favour YouTube as a source.

Why this works (and why DR matters less)

Every example Schoberg shared makes the same point: domain rating matters less for AI citations than for traditional SEO. A DR6 site can outrank established brands in ChatGPT. A DR16 site can beat major cosmetic companies in AI Overviews.

The reason? AI models prioritise clearly structured, well-sourced, directly answerable content over raw domain authority. If your content is formatted better than the competition — lists instead of paragraphs, answers before introductions, proper schema — the AI will pick you.

SEO isn’t dead. It’s just rewarding different things now. And the good news is that most of these changes take an afternoon, not a quarter.


Back to top ↑