Building a Map of AI Search Optimization

I think one of the most confusing parts about AI optimization right now is the pure volume of information that keeps coming out about it. I feel like I’m drowning in case studies and original research. And one of the ironic issues is that much of what I read seems legitimate and well-thought out.

And yet, I still am not really sure how the big picture of AI Optimization comes together. Like what are the big building blocks of how this shit works?

So I keep reading articles hoping it will all become more clear, then I wake up the next morning with more questions than I had the day before. But oh! Maybe if I read one more brand new article from this really trustworthy person it’ll all start to click…. And around the merry-go-round we go.

Compare that to SEO, which I’ve been working in for almost a decade, and have a relatively clear view of how it works. The big building blocks of SEO are content, links, and technical work. There’s a lot of nuance and overlap in each area, and the nuance matters, but the big picture is clear.

With AI, the big picture is the hazy part. And it’s made worse by the fact that everything about it seems almost – but not quite – the same as what I’m used to in the world of SEO. It’s like I’m looking at SEO through a crazy house mirror and it seems familiar but fragmented. I don’t know what’s real or what’s an illusion.

So after a few months of sitting in that confusion, and a trip to New Zealand to take some time off (woohoo 5th anniversary), I finally decided to sit down and force myself to create a map of the space that crystallizes in relatively simple terms this crazy AIO world I’m living in.

Today, I’m going to share that map with you. Fair warning, this probably isn’t like the other AI articles you’ve read. It contains no original research. And, it doesn’t contain a methodology that will save you.

Instead, it’s written so you can decide with common sense if anything I say is true. You will be able to read it and know the truth of what I’m saying just by reading.

For me, it was written so I can stop FOMOing and going down the rabbit hole with every new tactic I read about. Instead, it’s meant to help me to instantly place every ad, article, and LinkedIn or X post onto a map so I know if it helps me or not. If that sounds helpful to you, read on.

Is AI a Brand or SEO effort?

At the macro level, here are the two main POVs I see rolling around in the AIO space:

1. AIO is a brand effort

The first approach goes something like this:

AI is built on training data, which is basically an ingestion of the entire internet. The entire internet is too big for an SEO team to control and is more of a reflection of your brand overall. If the internet thinks your running shoes aren’t some of the best running shoes, that’s probably because you haven’t built the required brand to have that reputation.

2. AIO is just SEO

On the other side of this coin are the people saying “GEO/AIO/AEO is just another flavor of SEO.” and “yes, we know, SEO is dead, people have been saying that for years, can we please get back to work?”

They say getting featured in high quality publications across the entire internet has always been desirable for SEO, and besides that, AI will often search the web for things (they’ve told us they do, and we can see them doing it), so we still need to show up in search engines anyway. All of these building blocks look the same, and just have shifted a bit like SEO has been doing for the last two decades. We just need to evolve and stop acting like the sky is falling.

Bonus: Just try this one platform or tactic

There is a final bucket of content I see, but it’s not really a cohesive POV as much as a thick layer of noise sitting over everything else in our feeds.

Basically this layer of content is just a series of tactics that seem to work “really well” if you can just execute them correctly and at the right time. The implication feels like “hey just do standard SEO plus these other three things and you’ll have a leg up over everyone else.”

This is the llms.txt and the expanded FAQs, and the information dense-sentences, and the AI info pages, and all the SaaS platforms saying “see how our platform helped XYZ big brand get a bajillion more AI clicks in just 30 days or less”.

The info in this bucket usually is helpful and contains some nuggets of truth, but here is where the FOMO strikes the hardest. It’s hard to place these tools into the context of a greater campaign, and it feels like “if I learn how to use this tool I’ll really be learning how to do AIO, so why not buy it and use it as an on-ramp to get up to speed,” which is a super compelling starting offer, but usually ends in lots of wasted time and no real results unless you had a good strategy to begin with, which is what you were hoping the tool would give you, so it never works.

But the main problem that’s not included in any of these approaches is primarily that:

All of these POVs are accurate.

And equally: all of these POVs are bullshit.

Let’s walk through exactly how accurate and problematic each of these POVs is with a few real examples that are happening right now.

Sidebar: I’m only going to be focused on commercially valuable AI results in this article. That is, examples where AI recommends a product of some sort. I don’t really care how Google generates instructions on how to do a bicycle kick. Of course showing up here as a citation might have marketing value. But I’ll let other SEOs handle that.

Steph Curry’s AI Optimization

Enter Li-Ning basketball shoes:

First, let’s look at the “AIO is a brand priority” narrative.

When I asked ChatGPT to list the top basketball shoe brands, Nike and Adidas were consistently #1 and #2 across multiple iterations.

Li-Ning also showed up, at an average of #3. Still a strong showing.

Compare that to when I ask ChatGPT to recommend specific shoe models. I asked ChatGPT “what are the best basketball shoes” several times and Li-Ning didn’t show up once.

Notably, New Balance and Jordan both showed up multiple times despite losing at the brand level.

Kinda crazy right? #3 overall shoe brand doesn’t show up in product suggestions at all. Even after signing Steph Curry.

But what’s funny is if you look at the articles ChatGPT cites as sources for the product recommendation query, Li-Ning’s presence is surprisingly thin.

For example, ChatGPT cited a specific article by runrepeat.com multiple times.

For those who don’t know, RunRepeat is a well-known and trusted shoe review site. They started with running shoes but have expanded to become the self-proclaimed (and probably accurate) “#1 Athletic Shoe Review Site”. I read and trust their content personally. Reddit nerds recommend them as a resource. They cut shoes in half, take a dremel to them, and create plastic molds of the interior for every review. This is not your average affiliate site.

And, in RunRepeat’s best basketball shoes article (the one that was cited by ChatGPT repeatedly), Li-Ning shows up exactly zero times.

Matter of fact, if you look at all the basketball shoe reviews on runrepeat.com, Li-Ning doesn’t show up at all.

Google tells a similar story. If you Google “li ning basketball shoe reviews”, the search results are thin. It’s basically one well-known site, Reddit, and a few Youtube videos.

The Youtube section specifically includes four videos from the same channel:

Compare that to Googling “Nike basketball shoe reviews” which surfaces a wide variety of articles and videos from many different channels and publications.

Overall, it feels like Li-Ning’s brand presence is undeniable, but their product-level presence is thin.

So how could they fix it?

Honestly, a legit answer could be “just wait.” Steph Curry is one of the most popular basketball players on the planet, and their deal is days old at time of writing. After a year, RunRepeat will look silly if they don’t review Curry’s shoes.

Of course, Li-Ning could speed things up by reaching out to RunRepeat and saying “Knock knock, we just signed Steph Curry. Review our shoes please?”

Now pause for a sec. Who do you picture sending that email? Your SEO team? Probably not. More like your brand, affiliate or PR manager.

Another question: Do you think Nike or Adidas had to do any outreach to get their products considered, reviewed, and recommended by the most popular publications in the world? And thus to get recommended in ChatGPT? Again, not likely. If Nike didn’t show up in a basketball shoe prompt, you’d doubt the quality of ChatGPT, not Nike.

That’s the entire point of the “AIO is a brand priority” crowd. In a lot of cases, the SEO team are the wrong people to do what’s most important for AIO. And I say that as someone who has been “the SEO guy” for the last 8 years.

But of course, that’s not because SEO is completely irrelevant to AIO. In fact, there are some cases where you can get near-maximal AI results just from doing standard SEO.

To show you what I mean, let’s flip the script and examine the opposite end of the spectrum: a niche B2B SaaS product.

Trucking Management Software: SEO-first AI Optimization

I recently read a case study from an agency I know and respect about one of their clients in the “transportation management software” (TMS) space. TMS is software meant to help freight carriers organize and manage the movement of physical goods.

For trucking-focused companies like semi or dump truck contractors, sometimes this acronym is repurposed to “trucking management software”. As you can imagine, this is a niche software space.

The most-searched terms in the space – like “trucking management software” only get several hundred to just over a thousand searches a month, depending on the SEO tool you look at.

That’s a far cry from the 24,000 searches a month for “best basketball shoes”.

When I asked ChatGPT “could you recommend a trucking management software?” several times in temporary mode, one brand shows up consistently as a top recommendation: Truckbase.

Note that the results here were more volatile than the Adidas/Nike basketball dominance, but Truckbase was a consistent performer.

Now, of the 90 (!) sources ChatGPT cited, the overwhelming majority of them came directly from first-party content created by the brands themselves.

Of course, the first party articles weren't the only sources. And citations aren’t the full story anyway since AI models use training data.

And I was curious: what training data influenced this result?

I figured this was a small enough niche that I could try to build a pretty comprehensive profile of all the resources recommending Truckbase. So I scoured Truckbase’s Ahrefs backlink data and all the ChatGPT sources I could find and tried to find patterns. The main question I was trying to answer:

“Overall what is the playbook that’s helping Truckbase get recommended so consistently?”

I found some interesting stuff.

To start, two articles directly cited by ChatGPT were blatant SEO spam:

Similar site templates anyone?

These sites both reek of a “PBN” or Private Blog Network to me. So I did a little investigation.

Note: PBNs are age-old SEO schemes to build backlinks and pass authority to a target site. They’re built en-masse using AI (or article spinners or cheap overseas content), then carefully linked together to appear legit while being controlled by one central entity – often an agency.

Turns out, if you Google “wifitalentscom PBN” and “worldmetrics.org PBN” (the two offending domains) they both feature articles about “top PBN agencies.”

Any patterns? What’s that? “The Trust Agency” is listed first on both sites? That’s craaaazy.

This is my surprised face

Besides the PBN, multiple articles on fitsmallbusiness.com – a website I’ve seen featured in link-building circles - was actually cited by ChatGPT.

Bonus: fitsmallbusiness.com explicitly states in their footer that compensation affects how their content is written.

Now, not all the articles ChatGPT referenced are standard SEO slop. I also found two articles written on TechRadar.com and TruckingDrive.com – both sub-brands of media conglomerates that have public licensing agreements with OpenAI.

TruckingDrive in particular is industry-relevant. And, their list recommends Truckbase as their top pick.

But wait! The article recommending Truckbase is sponsored.

By Truckbase.

So, here’s the playbook we’ve seen so far in the TMS space:

• Write content on your own site
• Build a PBN recommending your product
• Buy sponsored posts talking about your product
• Get a couple HARO links

This is essentially an old school SEO playbook

However:

Think this playbook would work for basketball shoes? Think a sponsored post would get you to the #1 recommendation for basketball shoes? Yeah right. Your SEO team’s email would barely register as spam to the RunRepeat editors.

These are two wildly different industries with wildly different AI playbooks:

Basketball shoes AI visibility playbook

• Give Steph Curry $400M.
• Make a list of all the main publications in your space.
• Have your brand or PR team reach out to each one to get your products reviewed.
  • Mention Steph Curry.
• Hope your product team did a good job designing your products so they get reviewed well.
• Consider bribing editors or affiliate managers to give your product high ratings.
• Realize you’re in a big enough space that legitimate media publications control the narrative and can’t be bought.
• Go back to hoping your product team did a good job.

Even saying this is a “brand team” priority is underselling the effort. Your product team has to be killer, your CEO has to be cool with signing a check worth half a billion, and one of the most popular basketball players on the planet has to like your shoes.

Trucking management system AI visibility playbook

• Run the same SEO playbook the industry has been executing for 10+ years
• Tweak it slightly to start your research with ChatGPT instead of Google
• Call it “AI optimization” and tell everyone you’re an expert

Based on just these two examples, we can start to build a three-part framework for the entire AIO industry:

The Exponential Curve of AI Optimization

1. AIO is SEO. For niche and low-marketing-sophistication industries, you can just run the standard SEO playbook, tweak it a little, and get great results.
2. AIO is a company problem. On the other end we have hyper-competitive industries driven by brand power more than anything. Examples include basketball shoes in B2C or CRMs in B2B.
3. Messy middle: AIO is a marketing problem. This is a playbook we’ll explore in a future article. But briefly, you can imagine for an industry like mattresses, that the playbook could change considerably.

In mattresses, it’s not like one brand wins fully outright. Who’s the biggest? Sealy? Casper? Serta Sleep? If I named half a dozen more you’d probably be like “oh yeah they’re big too”.

In an industry like this where Brand strength isn’t monopolized, everything matters. This includes your site content, reviews on affiliate sites, Reddit threads, videos from any number of Youtubers, maybe even Insta influencers, etc.

To win this area you need a concerted effort across all those channels, and a strong enough brand to at least be in the conversation. This is where a dedicated AI strategist holds the most weight. One who can operate across affiliate, PR, SEO, ecommerce, and more.

Conclusion - written by AI:

So where does that leave us?

Remember the POVs from the top — AIO is brand, AIO is just SEO — and how I said each one is both accurate and bullshit?

Here's the resolution. Each is dead right for its slice of the curve, and bullshit the second you apply it everywhere else.

The "just SEO" crowd is correct — if you sell trucking software. The "it's a brand and company problem" crowd is correct — if you sell basketball shoes. Nobody's lying to you. They're each standing on a different spot on the same curve, describing the view in front of them, and mistaking it for the whole landscape.

What makes AI different from SEO is it’s not just about turning dials up and down on the same set of deliverables.

For SEO, if you’re in a competitive industry, you buy really good deliverables in all areas: links, content, and technical SEO. If you’re in a less competitive industry, you spend less. Same dials, different budgets.

For AI, the dials are not deliverables. They’re departments. For competitive terms you turn up C-Suite involvement. For less competitive industries you delegate to the SEO team. It’s a different dynamic, and requires a bigger picture understanding of how the entire company operates.

Which means the first question in AIO was never "what's the best tactic." It's "where does my industry sit on the curve?" Answer that, and everything downstream falls into place on its own — who should own this, what playbook to run, how expensive it's going to be, and whether that shiny case study in your feed even applies to you.

So next time a case study slides in promising a bajillion AI clicks in 30 days, don't ask "should I be doing this?" Ask two questions: what zone is that company in, and what zone am I in? If they don't match, close the tab. A basketball-shoe playbook will bankrupt a trucking-software company, and a trucking-software playbook wouldn't move Nike an inch. That's how you step off the merry-go-round.

Now — you probably noticed I sprinted past the most interesting part. The messy middle, where most real companies actually live, is where this gets genuinely hard: mattresses, supplements, the DTC brand with real momentum that isn't a household name yet. Too saturated for pure SEO, too winnable to walk away from. That's a full playbook of its own — affiliate, PR, SEO, ecommerce, and customer research all at once — and it's another article.

For now, the map is enough. You know the terrain, you know the zones, and you know the one question to ask before you spend another dollar or another ounce of FOMO on this stuff.

Where are you on the curve?