Thanks for the shoutout Charlie. This blog post specifically did come up when I was looking on the AthenaHQ platform for our own site. It is now being used as a source by generative engines!
This shift is wild but kind of inevitable. For years SEO was about gaming an index; now it's about being good enough to be summarized. The whole "be the answer, not just the link" angle is spot on!
In other words, LLMs are only going to be useful for product selection for another few years before web content is optimized to defeat it. When search engines came out, it took almost a decade for SEO to get good enough to turn Google, the best of the search engines, into mush. Odds are it will take a lot less time for LLMs to turn into mush, especially once the AI companies realize that they can sell advertisement and product placement. I'm not sure if this justifies the high values of the various AI companies, but it's an obvious play.
Personally, I didn't find those recommendations useful. I'm looking to buy a commuter backpack myself, and that glib AI response didn't really address the stuff I care about. For example, what is the overall capacity. Commuter implies a range, but I tend to pack a lot of stuff. It doesn't specify whether there is one big pocket, a big pocket and some smaller ones, more than one medium sized pocket or the like. I like one big pocket, but tastes and uses vary. It doesn't tell me if there is a cross chest strap, something I find very useful. I suppose I could repeatedly refine my results, but I know that I am unlikely to find just what I am looking for. At some point, I'll still have to go to the web pages to figure out the best compromise.
I agree with most of this, though I think it’s also saying something about how we both research and make certain kinds of purchases. But I also know there are a lot of folks who just want to know “what’s the best one” and buy that - even I do this with certain categories of products where i’m not heavily opinionated. Basically the reason why Wirecutter exists.
This was a good one. I've noticed my own habits changing as well, maybe even more than yours: I pretty rarely "google" anything by just typing into the Chrome browser. Maybe "rarely" is wrong there, now that I stop to consider this - there are loads and loads of trivial little searches in a given day - but for more and more things, I want a conversation.
I will have to share a story about a purchase I made with o4 earlier in the year. IT went sideways, but I am going to try again with a combo of o3 and Deep Research in a month or so. I feel considerably more confident this time, and feel like we are jumping into a new paradigm headfirst.
Good question. I guess the simple answer is that there are 2 types of search I do. One is super fast and I just want a very, very fast answer - like what's the name of that thing I can't remember? Oh yeah, it's (blank).
The other is a thing I don't know as much about, so I'm trying to understand something for the first time. I'm tempted to say I'm doing more of the 2nd type, but I will often just go on autopilot for browser-based search (generally straight into the URL bar) to find something really quickly, and I may not really be tracking all of these.
If I could insanely rapidly get a quick answer and THEN have a deeper convo, that would be ideal for all cases. I think that's where Google is trying to go right now with AI search.
Great article — and thanks for AISG, I built it (with a great Creative team) at HubSpot to help people start to understand and optimize for GEO. There are some *big* differences from SEO. In addition to what you mentioned, there’s much less emphasis on subjectivity and personal experience.
Thanks for the shoutout Charlie. This blog post specifically did come up when I was looking on the AthenaHQ platform for our own site. It is now being used as a source by generative engines!
https://imgur.com/a/KITBG8l
For sure! Also from one YC-alum to another, if you've got case studies or war stories to share, I'd be open to a collab.
This shift is wild but kind of inevitable. For years SEO was about gaming an index; now it's about being good enough to be summarized. The whole "be the answer, not just the link" angle is spot on!
In other words, LLMs are only going to be useful for product selection for another few years before web content is optimized to defeat it. When search engines came out, it took almost a decade for SEO to get good enough to turn Google, the best of the search engines, into mush. Odds are it will take a lot less time for LLMs to turn into mush, especially once the AI companies realize that they can sell advertisement and product placement. I'm not sure if this justifies the high values of the various AI companies, but it's an obvious play.
Personally, I didn't find those recommendations useful. I'm looking to buy a commuter backpack myself, and that glib AI response didn't really address the stuff I care about. For example, what is the overall capacity. Commuter implies a range, but I tend to pack a lot of stuff. It doesn't specify whether there is one big pocket, a big pocket and some smaller ones, more than one medium sized pocket or the like. I like one big pocket, but tastes and uses vary. It doesn't tell me if there is a cross chest strap, something I find very useful. I suppose I could repeatedly refine my results, but I know that I am unlikely to find just what I am looking for. At some point, I'll still have to go to the web pages to figure out the best compromise.
I agree with most of this, though I think it’s also saying something about how we both research and make certain kinds of purchases. But I also know there are a lot of folks who just want to know “what’s the best one” and buy that - even I do this with certain categories of products where i’m not heavily opinionated. Basically the reason why Wirecutter exists.
This was a good one. I've noticed my own habits changing as well, maybe even more than yours: I pretty rarely "google" anything by just typing into the Chrome browser. Maybe "rarely" is wrong there, now that I stop to consider this - there are loads and loads of trivial little searches in a given day - but for more and more things, I want a conversation.
I will have to share a story about a purchase I made with o4 earlier in the year. IT went sideways, but I am going to try again with a combo of o3 and Deep Research in a month or so. I feel considerably more confident this time, and feel like we are jumping into a new paradigm headfirst.
why would you say you want a conversation in lieu of a resource?
Good question. I guess the simple answer is that there are 2 types of search I do. One is super fast and I just want a very, very fast answer - like what's the name of that thing I can't remember? Oh yeah, it's (blank).
The other is a thing I don't know as much about, so I'm trying to understand something for the first time. I'm tempted to say I'm doing more of the 2nd type, but I will often just go on autopilot for browser-based search (generally straight into the URL bar) to find something really quickly, and I may not really be tracking all of these.
If I could insanely rapidly get a quick answer and THEN have a deeper convo, that would be ideal for all cases. I think that's where Google is trying to go right now with AI search.
Great article — and thanks for AISG, I built it (with a great Creative team) at HubSpot to help people start to understand and optimize for GEO. There are some *big* differences from SEO. In addition to what you mentioned, there’s much less emphasis on subjectivity and personal experience.