#23 - Kevin White

Hey all,

‍

This week, I sat down with Kevin White, the Head of Marketing at Scrunch AI, and asked him the question on the top of every marketer’s mind:

What’s changing about search in the age of LLMs?

GEO — generative engine optimization — isn’t just a trendy way of referring to the same SEO best practices we already know. The rules are changing as more and more users lean on LLMs to answer their questions.

Here’s what Kevin’s learned about GEO so far.

‍

‍

What we know about GEO search

LLM algorithms are black boxes, but one thing is obvious: LLMs don’t give a list of links like a traditional search algorithm. They answer prompts directly, referencing links as citations for their response.

To be cited as a source in an LLM response, a few things need to be true:

1. The answer to the prompt needs to be high-quality. LLMs are looking for credible responses, so backed-up data, human experiences, and proven facts are prioritized.
2. The structure of your page needs to be scannable. LLM crawlers scan pages looking for answers, so all the SEO best practices around clarity, structure, and load speed remain. 
3. The information on the page needs to map directly to the query. Rather than prioritizing aligned keywords, LLM crawlers are looking for content that maps closely to the prompt.

This is a key difference between SEO and GEO. Search engine optimization relies on keyword research, while LLMs demand what Kevin calls “prompt mapping.” 

Rather than plugging in keywords like “CRM software,” marketers should be focusing on the questions a user might query into an LLM. That means incorporating long-tail phrases like, “Which CRM integrates best with HubSpot workflows?”

For brands willing to adapt their content strategy to prioritize GEO, Kevin says the changes are good news. Sites no longer have to rank #1 on search engine results pages to be cited in LLM responses.

“You can earn citations and brand mentions even if you’re lower on the page,” Kevin said. “That’s the arbitrage opportunity right now.”

So… should I use prompt maps to make a ton of low-quality content?

No.

As with every new marketing channel, the knee-jerk reaction is to automate. Spin up thousands of AI pages. Try to brute-force your way into relevance.

Kevin’s take: that’s a losing game.

“You’ll see traffic,” he said. “But I’m skeptical of the quality of the audience. Top-of-funnel awareness with no intent is just noise, and you’re probably going after conversions.”

Not to mention, platforms are clamping down on sites stuffed with low-quality AI content. It’s risky to crank out AI garbage, because your entire website could be penalized.

Kevin’s recommendation is to focus on material that blends educational insights with human perspectives. Incorporate long-tail prompt phrases into content while also plugging in original thoughts, anecdotes, stories, and experiences.

The one exception he makes for automation is when you’re scaling proprietary information. At his last company, Kevin’s team built AI-generated pages using proprietary data — account plans for the Forbes 100 — to showcase what the product could do. 

“It worked because it was real product output,” Kevin said. Original, proprietary, and specific. 

So yes, use AI. Just don’t outsource the thinking.

Kevin’s playbook for getting your content GEO ready

If you’re trying to show up inside LLMs, start here:

• Map prompts, not keywords. Export your existing keyword list and turn it into real questions people might ask ChatGPT or Perplexity.
  
  
• Answer those questions clearly. Write pages that provide specific and in-depth answers. Make sure you’re using reliable sources to create your content. (LLMs pulling from other LLM-created pages do not make quality content).
  
  
• Make your site readable for bots. Use clean structure, simple markup, and load speeds that won’t choke a retrieval crawler, and consider adding FAQ sections that allow you to include entire prompts as questions.
  
  
• Track what’s actually happening. Watch for brand mentions, citations, and self-reported “found you on ChatGPT” traffic.
  
  
• Keep a human in the loop. Use AI for scale, but build ideas from lived experience, product knowledge, or proprietary data.

GEO rewards what SEO pretended to: real expertise, clear structure, and honest answers to questions prospects are actually asking.

This chat left me convinced we’re in the middle of the biggest shift in search since keywords. To learn more about Kevin’s perspective on where content marketing is headed, check out the full podcast recording below.

‍