Generative Engine Optimization - with Max Sinclair from Ecomtent

This talk focuses on generative engine optimization (GEO) and its implications for SEO in the age of AI. The speaker discusses the technological shift similar to early 2000s, emphasizing the rise of large language models like ChatGPT and Amazon Rufus in search functions. They reference the historical context of search engine algorithms, particularly Google's page rank and how companies optimized for it, using TripAdvisor as a successful case study. The speaker offers practical advice on auditing brand mentions in AI systems and creating custom GPTs. They also highlight the importance of product listings with context-rich attributes to optimize visibility in AI-driven searches, citing science papers from Amazon that outline best practices for content optimization.

Hi guys, last talk of the day. How are we feeling? Been a long day. Well, hopefully we'll make this as fun as we can. Um, so yeah, I'm going to talk about generative engine optimization. Who's heard of a GEO, AEO? Anyone heard of this stuff? This guy has. Okay. So, it's the future of SEO. Um, and yeah, I'm going to get I'm going to get into it.

So, I spent um six years at Amazon before I found any content added content. We help e-commerce sellers to optim optimize for AI powered search. So, that's chat GPT, Amazon Rufus, uh, and this kind of thing. So we are now today in a similar inflection point as we were in the early 2000s. In the early 2000s, Google went from having a tenth of Amaz of um Yahoo

search volume to 10 times it in a couple of years. And the big innovation that drove this explosion was, if the slides will work, the page rank algorithm. So historically, Google would um it worked out the best way to understand how good a a page was was to rank um the relative authority of all the pages pointing towards it and therefore it would it would serve the

best page to the customer based on the references each page gave towards it. This previous shift had some major winners. We've all you've probably all heard of Trip Advisor. They got 200 million monthly organic traffic at their peak. Got to being a 14 billion dollar company. And the way that they did this was by having hund they they hacked this page rank algorithm. They had hundreds of

thousands of relevant landing pages uh you know best restaurants in Sofia, best hotels in London. So when people were searching for specific things, it would they would drive them there. And equally they would give out awards to all of their restaurants and hotels as backlinks. So this would drive traffic back to their website. Therefore, they got lots of traffic to their website and you know 200 million

monthly organic traffic. This previous shift also had losers. Anyone here heard of virtual tourist? Nobody's heard of virtual tourist. And the reason no one's heard of virtual tourists is that they went out of business in 2017. You can't really see it here, but they actually started the month before Trip Adviser and and went defunct. many many reasons why they they lost this game. But if you listen

to the founder of Trip Adviser, their SEO strategy was a key part of their winning and these guys didn't have it. Oh, so large language model search is becoming the new default both on and off marketplaces. Uh just quickly, who here sells on a marketplace? Anyone? One. No one else. who has sell like who which of you guys are selling at all or you a few of

you. Okay. So on marketplaces you have stuff like Amazon Rufus, Walmart, Walabby, other kind of large language model based uh search and then on and then obviously the big one is chat GPT but you have perplexity you have a number of these other search engines. Chat GPT is now getting a billion messages every single day. It is the eighth most visited website in the world and increasingly

it is growing at 50% month and that's from December. So increasingly more and more of your customers are going to these AI answer engines to understand what products to buy to do research and and this kind of 60% of adults in the US have used an AI uh chatbot perplexity chat GPT all of these to do product research in the last 30 days according to A16Z. Okay.

So I'm going to talk about three AI search bots today. I'm going to start with some hacks on chat GPT. I'm then going to get into the science on Amazon Rufus and Cosmo. Are any of you guys selling on Amazon? No. Well, I think it's helpful because they publish a science paper. So, I'll talk through it quickly so you can understand the science behind this. Chat GPT

is a black box. They don't they don't publish anything sadly. Um, so we'll start with Chat GPT. These are some quick hacks you can do now to get your brand mentioned. Um, the first thing you need to do is to audit. So, you should all do this today. Go into chat GPT and put in what is my brand? Ask about your brand. Ask about the topics you

care about. See how you're coming up. And I did this for myself for my company and immediately I saw that we're being referenced by a bunch of links which are old, they're outdated, they don't represent us well and I went and I contacted some of these uh blogs and I updated the content. So now if you ask about us in chat GPT at least is drawing off

relevant stuff which is which is which is better. Number two is to create a custom GPT. So if you don't know a custom GPT you can create in in the GPT store. Um, and we did this, we did one for optimizing for AI search. It gets a thousand uh, conversations. It gets four stars. And we're basically telling Chat GPT what we do as a business. And therefore,

when people ask about these topics, chat GPT has that data. So, if you're selling tables, you could maybe say, "Oh, I'm going to make a custom GPT about how to uh put together tables safely or whatever it is." If you're selling dog harnesses, you could talk about taking your dog for a walk or pet care. Whatever is relevant to your brand that chat GPT should know about

your brand. Put it in a custom GPT. It'll take you 10 minutes and then you'll start to be surfaced. The next one is optimizing for Bing. Again, this will take you 10 minutes. I've done it. I did it myself. You can basically copy everything over from Google to Bing. As you guys probably know, uh, Microsoft invested, uh, hundreds of of billions of dollars into ChatGpt and ChatgPT

is using the Bing index. So, the two things you need is the web master tools and the Bing places for business. You copy that over from Google and you're going to be start to to rank in these algorithms. Uh, and the last one and the most complicated one is AI partnerships. So it it's not easy to do this, but AI um OpenAI is training their data on

Axios on Wall Street Journal. They've established these partnerships with data. So if you can get yourself mentioned by one of these um publications, you're going to be in the uh next training data of GPT5. So you're going to, you know, or six or whatever it will be. So you're more likely to actually be in the training data to understand uh so the chat will recommend your brand

when people ask about it. Okay. So, I'm gonna run through I'm going to run through Rufus. I'm going to do this quickly because none of you are selling on Amazon. But I am I do want to talk about it because it unlike um unlike Chat GPT who don't publish um anything really it's all very secret. the big companies like Amazon, Google and Microsoft uh they all publish

the science papers famously uh you know the the uh the the um science paper all you need is attention which was the one for the LLMs which opens to build chat GPT was originally a Google paper. So the big companies published these science papers and it gives us real insights into how these models work. Um but quickly what I have on this slide here is why why

do people even do this? Uh this is actually my home office and as you can see it's a superior customer experience. You type in your prompt, you get an exact answer. It's much easier than searching through thousands of products which all look the same. And also I've got another example here of of me skiing and and doing the ski gloves. So um this will be bigger than

mobile. Uh so I I I worked at Amazon for six years as I mentioned in the beginning. Uh when I started at Amazon 30% of searches were on mobile. Now that is 80% and this is a significantly bigger uh technology shift than what we've seen before. So I won't go into this again because we're not focused on Amazon but roughly I've done some back of the envelope

maths from uh how much inferentia uh AWS is giving towards Rufus and they give 8,000 chips 80,000 chips and if you work backwards to how much um that would mean in terms of searches it would be about 14% of searches. So people are adopting these AI um search now and and obviously that's relevant if you're selling on you know you're selling anywhere. So as I mentioned um

as I mentioned the great thing about Amazon and uh all of the big um tech companies is that they publish all of their science papers. So Amazon has published the one behind Rufus behind its AI chatbot. It's called a shopping agent for addressing subjective product needs. And in this science paper, they talk about five facets of product needs that you need to put on your product listings

in order to get picked up by these AI engines. And they are subjective property. What is the attribute that you're selling? Is it, you know, is it a sturdy table? Is it at a colorful dress? Secondly, what event are you are you selling the product for? Is it for Christmas? Is it for Easter? Is it for running a marathon? What what is the event? You need to

give the AI the context. Uh the activity. So, are you is it a gaming chair? Is it a are you traveling? What what are you doing with the activity? The goal. So why are people buying your product? And then lastly, the audience. Who's the target audience? So these five things are super important to have in the product listings beyond the keywords. These give you these give context

to what your product does and it helps AI engines and Cosmo. This was a previous paper and I I'll run through this again. A previous paper Amazon published it. The the point I'm making here is that it it tells us exactly the same thing. It tell it it tells us the same thing which is you need in this um in this paper they call it um uh

they call it a slightly different thing but it's the same concept. You need to talk about the function. You need to talk about the event. You need to talk about the audience. You need to talk about seasonality. These are the key things that these AI models are looking for. And I could go through and I could show you 10 science papers all from Amazon um and some

from Shopify, but they all say the same thing. This is all about context and it's no longer about keywords if you want to optimize through an AI algorithm. Um the yeah, search is also multimodal. Again, we see this in the science papers. We know this if you use chat GPT, it can understand the images. So you need to bring the all of these things I just mentioned

event the target audience all of this stuff needs to be in the images as well as the text now because the AIS can can surface both of them. Uh and in the Amazon case they have something called recognition but all of the different AI engines can can and yeah so so this is um you know as as I like to say this is not just theory we're

working with our customers to generate content which is optimized for AI we do infographics lifestyle images text all of this stuff and as you can see on Amazon for this customer they previously didn't have any content through our tool they generate this optimized content it gets picked up by the AI algorithms they increase their bestseller rank um from 60 to 25 uh and they get to number

one on their keyword which is cigar cases. So um yeah what what our technology does is we import the listing data, we look at conversion rates and discoverability uh you know similar top selling listings in the market and we generate optimized r and vision content to get picked up by AI models. Uh so this is what it looks like. We can Oh, didn't work. This is what

it looks like. You can generate the v uh written visual content at speed and scale across um your entire catalog. And um yeah, or you have 15% off there if you're interested. And that's my email as well. I'd love to chat to you further about this stuff. Very passionate about AI search and I think it's the future. So thank you.