Crazy Rich Algorithms - The Remix: From SEO to GEO to GPT Checkout

Promotional still from Crazy Rich Asians (2018), courtesy of Warner Bros. Pictures.

What Happens When the Funnel Collapses?

The arc of SEO to AI’s impact on advertising is already playing out like an opera. I made my first LLM-influenced purchase about two months ago. In late June, I made some predictions about when LLMs would begin to charge for native ads. And then by mid-July, OpenAI skipped ads altogether by piloting check-out capabilities within ChatGPT. Today, NYMag is now also declaring “SEO is dead” (an oft mentioned refrain, but too early to be saying really).

Here is the full opera in its 3 acts.

Act I: From Yellow Pages to ChatGPT

I wrote an earlier post about my first online purchase that could be credited to ChatGPT. TL; DR I was playing with selfie analysis and health assessment with ChatGPT, and what ensued was a diagnosis of my reaction to niacinamide/retinol along with some K Beauty alternatives. (Fast forward to today, my skin is looking good again, thankfully).

In late April, OpenAI released a new feature for shopping, allowing businesses to register their products to show up accurately and up-to-date in chat suggestions.

ChatGPT commerce or “conversational shopping” is here. It might be nascent, but it provides a clue to where things are headed for our use of LLMs in daily life.

The term GEO (Generative Engine Optimization) began to surface in late 2024, a fitting name as it sounds close to SEO and marketers love a three-word acronym. Not four, and not hard to say like “LLMO” (Large Language Model Optimization) which may be more accurate but sounds like reciting the alphabet.

Let’s rewind to understand where we’re headed.

SEO vs. SEM vs. GEO

SEO (Search Engine Optimization) arrived in the mid-1990s, soon after search sites such as Yahoo, AltaVista came onto the scene. Website owners understood that search engines were effectively ranking their sites, and that to be at the top of the rankings meant better placement (at the top, first page) in search results. This actually predates Google’s launch.

How did these website owners improve their rankings? They optimized their sites, for example to include specific keywords, and submitted their websites to directories. In effect hoping to make Yahoo a “smarter” Yellow Pages, delivering their site first rather than lower in the sequence.

(For anyone whose birth year begins with a “2” the Yellow Pages was a thick, yellow-pages directory of every local business or professional)

A few years later, in 1999, a system named GoTo.com allowed advertisers to bid on keywords in order to show up higher in search results. They are credited as the first to create the PPC (pay-per-click) or Paid Search model which has become a cornerstone of performance marketing today. (SEM stands for Search Engine Marketing but most just say “paid search”)

Skipping ahead 25 years (and the advent of social media, more on that later), we now have Generative AI. The LLMs such as ChatGPT, Gemini, and Claude have burst onto the scene. ChatGPT boasts of 1 billion messages sent daily and 300 million active weekly users. Here

Meanwhile, it was reported in January that Google’s share of search dropped below 90% for the first time since 2015. Ask anyone from the millennial and Gen Z cohort, and the majority are using other ways to find what they want online. Be it TikTok, Pinterest, or Instagram - each now offers some version of paid search.

I generally think diversification is good. And it’s almost unbelievable that Google would enjoy 90%+ market share for a decade uncontested.

So is GEO going to beat out SEO? Not just yet. ChatGPT accounts for just 1% of search traffic - making Google roughly 373 times larger. And when will we see something like GEM (cute, since it’s an actual word - might be the next acronym) make an appearance? Meaning when will marketers have to pay to show up in chat suggestions, rather than just relying on data quality and site relevance?

Act II: From Free to Fee

A personal, professional frustration of mine is when executives focus on social media following, rather than engagement or paid social effectiveness. Or in other words, prioritize what is visible in organic social media, thinking it will have instant commercial impact. Why? Because, for over a decade and a half now, social media algorithms have only shown a teeny tiny percentage of followers a brand's organic content. It always made me wonder how the leadership of a publicly traded company could think that another publicly traded company (Meta) would allow for free advertising.

It’s all a bit black box, but estimates put organic reach at 2-6% of a brand’s followers. If a brand has 100K followers on Instagram, that’s 2-6k that will see a post. For anyone wondering why likes and comments are low, this is why. And organic reach is rumored to be notoriously low on Facebook, it may be a tad higher on Instagram. Any time the algorithm changes? All that a brand has been doing to “show up first in the Yellow Pages” needs to be dusted off again. Facebook Advertising, after all, has been in place since 2007 with many algorithmic changes (and declining organic reach) since.

Why so much on social and where’s the prediction? I think the history and ramp up of social, from organic to paid social dominance is a sound parallel to what we’re seeing with LLMs and future monetization efforts.

And right now? We’re still in the “free” phase. But not for long.

Prediction: Later this year, 2025, into early next year, 2026, we’ll see OpenAI testing ad placements in chat sessions.

The arc of social media looked something like this:

1. User growth, no clear revenue model (From The Social Network recall when Justin Timberlake says “You don't even know what the thing is yet. How big it can get, how far it can go" - and also it's not time to sell ads yet)

2. Early display ads, changes to the news feed (Who misses chronological feeds? I do)

3. Programmatic ads, hyper targeting via data (Ads show up seamlessly and well-timed to intent)

4. Ad dominance, creator monetization model (Anyone else see a TikTok analysis of Alix Earle’s likely monthly TikTok pay? It’s staggering)

For LLMs, we’re soon entering part 2 of the arc.

What could this look like to users?

• You ask a question on how to make cacio e pepe. A sponsored result shows up first “Here’s a recipe brought to you by NYTimes Cooking”

• In researching a destination, you ask about hotels. The first chat response is from Expedia, inviting you to browse and book via their platform

• “I need a gift for my best friend.” Answered by a Shopify-assistant

Now there might be a little resistance from users to start. My K Beauty transaction was swift because it did not include any advertising, which made it feel authentic. But otherwise, within an LLM is a perfect place to browse and shop. If you’re asking, you have high intent and are giving specific context to your needs. It’s nicer to receive a recommendation in a conversation, rather than on a banner ad you’ll never ever click. And for now, it feels like an unsponsored recommendation.

Which leads to a question for marketers - how to prepare?

Act III: Collapse of the Funnel

Well, that prediction came true faster than expected.

OpenAI is already testing native checkout inside ChatGPT - skipping past ads entirely.

That means users won’t just get product recommendations - they’ll be able to buy directly in the chat.

To revisit my earlier example about shopping K-beauty:
My intent to buy could’ve been captured without ever leaving ChatGPT.
No Google. No TikTok.
Discovery → Recommendation → Purchase — all within an LLM.

The move is reportedly happening via Shopify, allowing OpenAI to take a cut of in-chat purchases. See coverage from Reuters, OpenAI Working On Payment Checkout System.

Why it matters

I was just explaining to my brother-in-law why his daughters prefer shopping through TikTok. Instead of a sterile page of links or a row of PLAs from Google, they get:

• A review

• Animated visuals

• A sense of context

Basically, an organic sensory ad.

This is similar - but compressed even further.

We’re witnessing the collapse of the traditional commerce funnel in real time.
The old model: intent → ad → link → checkout is being replaced by a single interface that remembers you, advises you, and sells to you

This changes everything:

For Google: OpenAI eats search intent before it ever hits a keyword auction.

For Amazon: OpenAI cuts to the sale without a marketplace middleman.

For brands: The question shifts from “Where are we placing ads?” to “Are we inside the AI making the recommendation?”

The real 5-carat emerald ring?

Not just recommendations.

Persistent memory + intent-rich queries + seamless checkout.

Let’s stay with my K Beauty skincare example:

Maybe next I ask ChatGPT for makeup with anti-aging properties similar to my last purchase. Or a beach tote for an upcoming trip - one that fits the vibe ChatGPT has already helped me plan.

Ideally, it remembers that I:

• Have rewards set up with Olive Young

• Think Loewe is the definition of cool

• Want brands I’ve already approved

That’s not just AI-assisted commerce. That’s a personal shopper, data analyst, and storefront in one.

This isn’t a pivot to ads. It’s a hurdle over them.

Because AI won’t just replace parts of the consumer journey. It can restructure the entire path.

And it raises the real questions:

• How much do we trust these platforms with purchase power?

• How will brands stay visible in a world without search?

• Who owns discovery when the interface is invisible?

Final thoughts:

• GEO isn’t replacing SEO - it’s shifting the balance.

• Brands must write for humans but structure for machines.

• Voice and tone now matter more than backlinks.

• Early registration with OpenAI is still free - but that window won’t stay open forever.

And remember GoTo.com, the company that pioneered paid search? Its founder, Bill Gross, is back. He’s now building ProRata.AI and Gist.AI - new platforms aimed at embedding ads into AI-generated answers and monetizing the content LLMs ingest.

From pay-per-click to pay-per-crawl. We’ve come full circle.

So… would you let ChatGPT shop for you?
Where would you draw the line - groceries? Clothes? Healthcare?

I’ll be watching. It’s moving faster than even I guessed.