5-Minute Whiteboard: Is Generative AI in Search Results the SEO Apocalypse?

In the not-too-distant future, we’ll look back at the days when a full 1/3rd of Google searches resulted in a free click to the open web and marvel at how good we had it. For the last few months, Google’s been telegraphing that generative AI answers will soon appear in search results. These pieces of content will be drawn from training data (i.e. all of the web content you’ve worked hard to create) and amalgamated to produce user-satisfying answers (i.e. prevent searchers from clicking on anything that takes them out of Alphabet’s properties).

Will this spell the end of SEO? Or the end of the good times in search marketing?… Maybe. But, I think it will also open up an entirely new way to do marketing in the search results.

Transcript:

Howdy SparkToro fans, and welcome to another edition of 5-Minute Whiteboard.

Because of my history in the in the worlds of SEO and search marketing, I get this question a lot: Is generative AI gonna be the death of SEO?

I get this from, folks in the press. I get this from folks in the SEO community, from some folks who maybe are seeking SEO services and look, obviously, I have not been in the field as deeply as I once was in my previous company, but… No. I don’t think it’s the apocalypse.

I do think, however, that we in the marketing world (whether you’re in search marketing or not) need to be thinking about SEO very differently than we have in the past. I’ll show you what I mean here in these quick five minutes.

So from approximately (I’m gonna use approximate dates) let’s say 1998 to ~2014. Right? We had this really good run where Google’s whole goal was to get people off Google. Remember Larry Brin, and (sorry!) Larry Page and Sergey Brin saying exactly that. Right?

That essentially, you’d come and do a search for “What is the best cleaner for Wood slab tables?”, and Google’s goal was to get you off Google to the open web. They wanted to do that because they wanna encourage clicks in these places. Why? So that you would also be trained and learn to click on their paid results.

Because that’s how Google made money when people would buy ads above here. That’s how they made their revenue. Now, fast forward a little bit: 2014 to 2022 in this era of Google’s development. We started to see essentially the paid results always at the top.

Right? Always above as opposed to on the right hand side or in other places. But, they would also include lots of instant answers and those instant answers were drawn from all over the web. In this case, right? There’s a a link to the site that provided the answer.

These are called featured snippets. Or rich snippets or instant answers. Right? And here are these tips for keeping your wood slab table protected.

This content was stolen from here. I say stolen. Obviously, it was crawled with permission. This website’s permission is provided (via Robots.txt or Google Search Console), but in sort of a way that perhaps this website didn’t intend.

They hoped that Google would provide the link, but maybe hoped they wouldn’t scrape all this data and put it all right in the search results. But, Google learned that users really appreciated this. That they loved getting those answers super fast.

From Google’s Supercharging Search with Generative AI post

Fast forward to 2023 and Google is now showing off some of their generative AI search experiences. I think in May, they had an event where they they showed some examples of what generative AI in search results might look like. This almost certainly was because of pressure from Wall Street after Open AI, right, and ChatGPT came out. So paid results still at the top, not that dissimilar. And the organic or SEO results are essentially taking amalgamated answers from many people like this who’ve produced content that answers the query effectively and smush them together.

I say smush because generative AI is essentially just taking all of these big data training sources, right, from all over the web and they are putting that content together in ways that they believe will will predictably answer nearly any query. That’s essentially what ChatGPT or Bard or or any of these others are. And these amalgamated answers come from a variety of sources, but those sources no longer need to be cited nor will Google probably cite them, right? They might take an image. They could even generate this image from generative AI, right? Like things like see with with Midjourney or Bing’s generative AI, right, image creators. And so you’re gonna get these kinds of answers, more AI answers with no citation.

What’s really changed here is the missing link to a single source. What hasn’t changed is that this answer and these answers could still contain your brand. They could still contain, hey, the way to treat wood slab tables was is with Method cleaner or or, you know, with, Simple Green cleaner or whatever it is. It’s probably not simple green.

But I’m just saying, right? It’s not Windex either but those brands could get into these results. They could get into these results and they can get into *these* results. It’s just a different process.

You have to think differently about how you get into the results… and the way you get in is how?

That’s right. These are drawn from pieces of content across the web. So if all the content across the web that talks about cleaners for wood slab tables mentions your brand, mentions Method or Windex or Simple Green. That’s what’s gonna show up these answers. If it mentions baking soda and vinegar, that’s what’s gonna show up in these answers.

That’s how you win in an AI generated world. And that’s a very different way than marketers have previously thought about winning in search and SEO, but it absolutely will be the future if Google truly moves to this type of methodology. It’s my belief that practices like PR, practices like content creation, practices like finding the sources that are feeding into the generative AI training data, those are going to be important parts of what web marketers have to do in future regardless of their diffs as blend if they wanna be in SEO results.

In the future, I expect a lot of marketers will be tasked with creating co-occurrences between their brands/products/services and certain topics and key phrases so that AI answers will be more likely to surface them in those replies.

As Wil Reynolds and I discussed in the comments:

The way we think of link-building and guest posting today may be very different in a future where Google’s answers come without a link to the source… Generative AI marketing through seeding of training data sounds like a pretty interesting future to me!

See you next week for another edition of five minute whiteboard.