Checklist: Get Cited in Google’s AIOs and LLMs Using On-Page SEO Techniques

What can I do to get cited in LLMs or Google’s AIO?

By Colette Nichol, Story Strategist and SEO

To get cited by LLMs or Google’s AIO, the best long-term thing to do is to create a brand that is strongly associated with the topics for which you want to get cited.

You need to build out brand properties on platforms like YouTube, LinkedIn, Instagram, Spotify, and X. Make sure that what you’re posting is strongly related to your core topics. You can also invest in PR that will help you build your brand through earned media and paid media. And you can even try doing straightup press release blasts. The goal is to get multiple mentions across the internet that connect your brand to the core topics that will help you drive revenue.

In other words, if you sell pink widgets (and you need to rank and get cited for pink widgets) then you need your brand to get mentioned in tandem with pink widgets in multiple places around the web.

In fancy SEO speak, you need to comingle your brand entity with your expertise entity across the web.

But, if you just want to try your hand at getting cited the old-fashioned way aka. with on-page SEO, try out my checklist.

This is what I use to get my content to rank. And so far, it has been working well. I’ve seen posts and sites recover through this use of the strategies on this checklist.

Notes about LLMs

I tried to post this on Reddit and got banned from r/SEO. So, here it is for you.

The questions:

• How do LLMs surface answers?
• Are they like a search engine?
• Do they correlate with search?
• Are they completely different?
• And, is SEO dead?

The answer:

LLMs don’t behave like search engines.

They’re probabilistic text generators that sometimes ground their outputs in search. They don’t “answer” questions like Google. They’re not trying to file away the best info so they can surface it and be helpful. 

Instead, they’re trying to predict the most “probably correct” next words. 

Because they’re all trained on different data and have different levels of filtering, every LLM behaves differently.

• Each has its own training data and filtering.
• Even when they access the same SERPs, they don’t interpret them the same way.
• Filters (and training data) can warp results, leading to answers that don’t align with search (or reality).

LLM Errors

I’ve tested ChatGPT, Grok, Gemini (weakest by far), and Google’s AIOs.

All make errors that don’t align with the SERPs or reality. Some are egregious.

Example:

Grok insisted the federal estate tax exemption went into effect on July 4 (bill signing) instead of Jan 1, 2026 (effective date). The SERPs were unanimous, and ChatGPT got it right.

But Grok just couldn’t spit out the right answer. When I gave it the gov’t URL with the correct info it even went so far as to say, “I’m sorry, I cannot assist with that request as it relates to unauthorized access.”

Um, I’m sharing a public link Grok. Calm down.

That said, ChatGPT makes mistakes too, spitting out “facts” that contradict the SERPs.

• Sometimes LLMs mirror the SERPs. Sometimes they don’t.
• SEO and GEO/AISEO are connected, but it’s not one-to-one. 
• Winning in Google often helps, but not always.
• LLMs also remember user behavior and prompts. That makes testing on your own account unreliable. Personalization is way deeper on the LLMs than on Google.
  

In terms of on-page that helps you get cited in the AIOs, this is what’s worked for me so far:

• Tables and structured data (very important)
• Custom graphics that contain useful lists/answers (can really help you get into the AIO)
• Case studies (feeding the model fresh info!)
• Answering the query in the first 100 words
• Being as clear as humanly possible and front-loading sentences

So, in sum, SEO still matters (obviously) and is driving most of our traffic.

But LLMs are an important additional target.

For now, they should be approached together.

Plus, only enterprise-level businesses will have the budget to do SEO and GEO separately. For most SEOs, we’re going to be doing both.

---