In this extended episode of the Near Memo, Greg Sterling , Joy Hawkins, Darren Shaw & Mike Blumenthal explore the role of generative AI and how it has evolved one year after CHat GPT was released. We cover:
How Local SEO agencies use AI to improve client content:
How Local SEO agencies explain AI to their clients:
Local SEO Agencies & their AI future:
Subscribe to our 3x per week newsletter at https://www.nearmedia.co/subscribe/
In this extended episode of the Near Memo, Greg Sterling , Joy Hawkins, Darren Shaw & Mike Blumenthal explore the role of generative AI and how it has evolved one year after CHat GPT was released. We cover:
How Local SEO agencies use AI to improve client content:
How Local SEO agencies explain AI to their clients:
Local SEO Agencies & their AI future:
Subscribe to our 3x per week newsletter at https://www.nearmedia.co/subscribe/
Greg (00:01.383)
Hello, everybody. Welcome back to the near memo. And today we have a special edition talking about AI chat GPT SEO and how these technologies are going to impact the future of local SEO and joining us today in addition to Mike Blumenthal, who's always here or here 99.999% of the time are Joy Hawkins owner of Sterling Sky and Darren Shaw, who's the founder of white spark and these guys are
leading local SEOs and leading thinkers in the space. And so we're very pleased to have you with us. And we wanna sort of jump right into the conversation unless you guys wanna say anything more about who you are, your perspective or what you're wearing or what life is like.
Darren Shaw (00:45.13)
I'm wearing a Beyonce shirt that I got from the concert.
Greg (00:50.751)
All right. And how is the weather in Canada? I know you're in different places today. You're both in Canada.
Darren Shaw (00:57.186)
This is the first November in history, on record, that Edmonton has not had snow. It is weird out there. It is like, I think this is the end of the world, actually, that we don't have snow. So it's not good. There's something bad out there. It feels bad.
Joy Hawkins (01:16.6)
Oh, like a couple hours south of you had snow. I was in Alberta last week and there was snow everywhere and it was freezing. So it's true.
Darren Shaw (01:23.726)
Well, you were in the mountains though. So yeah, in the mountains. So Lake Louise, so beautiful there.
Joy Hawkins (01:28.544)
Yes, so beautiful.
Greg (01:32.191)
All right, and with that nature reference, we can jump right into the conversation. So we've got a lot of stuff to talk about, very interesting discussion. But let me start out by asking both of you, how have you integrated any AI tools, chat GPT, into your agency work on a regular basis? Tell us about that and in as much detail as you can. So Joey, why don't you start.
Joy Hawkins (01:59.92)
All right. Well, I think it varies quite a bit depending on what department or job people do. So on the creative side, it's been implemented quite a bit. So my husband that does all our video editing, he uses AI features like crazy, thinks they're wonderful, they're really cool. They save him an insane amount of time. So on that end, I'd say it's kind of meeting up to the hype. Where I feel like it's not meeting up to the hype is more on the content side.
I think when AI first came out, everybody thought it was going to replace content writers. And you literally had people tweeting that, you know, content writers were afraid to do their job and things like that. And what we've kind of found over the years is that I do not think AI is near or anywhere close to where it would need to be where you could literally click a button and then have an article that you're ready to publish, you know, within five minutes. I know people are doing it, but definitely not my agency.
Mike B (02:50.91)
I have a quick question, Joy. That article of ours about Darren's research that you shared in your newsletter, did you read it yourself or not? About, oh, okay. The way we created that article is sort of interesting and I think it addresses the question of content to some extent, not completely. We recorded an interview, David and I and Greg recorded an interview, took the transcript.
Joy Hawkins (03:01.574)
I did not.
Mike B (03:15.71)
dropped the transcript into ChetGBT, told ChetGBT to write an article based on the transcript. And then David and Greg both edited the article for accuracy and flow and quality, coherence. And so by limiting it to a subset of information, i.e. David's conversation about Darren's research, we were able to, in 20, 25 minutes, publish an article that at least your
Greg (03:29.492)
Coherence.
Mike B (03:45.562)
editorial staff thought was up to snuff. And we thought it was up to snuff too. Well, it's not a...
Greg (03:49.779)
Ha ha, the joke's on you. No, I'm kidding.
Joy Hawkins (03:51.928)
No, no. So I need to clarify because I do agree, Mike, I think that it's great at that. Like you gave a really good use case for it. If you have content that has already been input by humans and then you need to repurpose that somehow, like take a list and make it a blog, take a blog and make it a script, it does a good job of that and requires minimal editing. But like coming up from scratch being like, you know, I want a 500 word article on this criminal law in Texas, it's really inaccurate.
and the amount of time it takes to fact check it is about the same amount of time that it would take just to write the article from scratch.
Mike B (04:24.493)
No, here you go. Absolutely.
Greg (04:27.711)
Darren raised his hand. Or are you just echoing?
Darren Shaw (04:30.018)
Well, just that's the whole thing the amount of time that it takes to fact-check it It's like I do wonder like that exact example you gave Mike How much time did you actually save like would if you had just written the article would you have saved that much? I'm you just take the transcript and then wrote the paragraphs yourself. Would you have like I know would take me maybe five minutes longer I don't know
Mike B (04:53.undefined)
It's hard. Go ahead.
Greg (04:53.235)
I wrote, Darren, I wrote an article the preceding week about the findings themselves. And then we talked about them and then that became the second article. And it took me, you know, it took me, I don't know, half an hour maybe to write the article, I would say. So it's equivalent.
Darren Shaw (05:11.278)
Mm-hmm.
Mike B (05:13.41)
And it isn't quite though. I mean, I spent five minutes on the technical side, gave the article to David who is very, he doesn't like to write anymore and he's very resistant to writing, but he agreed to look at and edit it. He said he spent five minutes. I don't know how long you spent editing it, but it was eight minutes was a sunk cost on it. We recorded it. We already had it. So it was repurposed. So I think all in Greg, it probably was.
Darren Shaw (05:33.784)
Mm-hmm.
Mike B (05:42.462)
half the 30 minutes or of extra time, do you know what I mean? Compared to what you spend.
Greg (05:47.621)
But if you consider the recording itself, it's comparable, I think, you know.
Mike B (05:51.31)
If you consider the recording itself, it is comparable, but I think it does set a model for how agencies can interact with their clients. You do a five minute interview about a topic and then it's in their voice with their information, much less fact checking required. And so it is a model for agencies to do content.
Darren Shaw (06:12.946)
And that segues nicely into some of the ways we use it at Whitespark that I, you know, to answer Greg's original question. That is one of the things that we do. So we use it actually to seed questions. So for content for our clients, one of the great use cases is you ask a series of questions like, what are the common concerns that people have with a drain cleaning service? And so if you ask ChachiBee to come up with that list of
of concerns and then you say, okay, now phrase them into questions. Then you give those questions to your clients. So, hey, we want you to answer all these questions. There's a couple ways you can do it. You can either send them a form where they fill out their answers by typing, or you get on a Zoom call with them and record it. And then you just take their words, turn it into the transcript from the Zoom recording. And then you can actually see the...
the content the way that you did Mike, into getting the first draft and then you have to finesse it and edit it because in fact check it. And so we do actually do that for our clients and you do use it for a number of other things like coming up with Q&A questions because we like to pre-populate the Google Q&A section. We use it to come up with Google post ideas. So we'll take, like we know that this, let's say dentist has
pages on all of these different types of services they provide, we feed that to ChatGPT and say, create a Google post based on this content, or create like five Google posts based on this page of content. And so it creates the Google post content, and then our team still has to finesse it and edit it. But it does, it's like it takes some mental load off of the job. Like if you just give an employee, the like one of our account managers, the job of taking this content.
Greg (07:54.404)
Accelerate.
Mm-hmm.
Darren Shaw (08:05.61)
and turning it into Google posts, there's just a lot of thinking. And so it does a lot of that sort of upfront thinking for you. And then it's just the editing. So it does speed up the process. I read, I watched a really great podcast episode with Aaron Wyke on his new conversion podcast with Andy Crestedina. And Andy talked about his process for using ChatGPT for content generation. And I think he said that.
you're saving about 30% of the time and that feels about right. I think that you actually said it right there, Mike. You said you save about, you saved maybe 30%. It was a little 30% faster than if you had just done it from scratch. And I feel like those are some of the benefits we're getting.
Mike B (08:43.986)
Right, but as you said, it relieved the burden and allowed us to get an article out the door, which is a huge, huge thing, right? Because it really, some.
Darren Shaw (08:50.326)
Yeah. That's the thing. Yep.
Joy Hawkins (08:55.684)
Yeah, we use it for posts as well. I think we measured it cut down the time it takes to create a post by about half. So it was substantial for posts, but it was because we had a really good prompt that was like including all the things that we wanted in a Google post based on the study that we did. So like make sure it has an emoji, make sure it has a sense of urgency. Like we have a list and I think the prompt we put on our blog. But yeah, using that, I love that it just places the emojis for you. Like that's a huge time saver.
like going and finding the right emoji and putting it in, that takes time and it can automate that really well.
Mike B (09:28.19)
That speaks, that's an interesting point though, just that you did previous research that you brought to bear on the prompt so that it could fulfill your level of expectation. Well.
Greg (09:28.617)
So,
Joy Hawkins (09:40.356)
Yeah, you got to be specific. Like I found if you're really generic and you're like optimize this title for YouTube, it's bad. So you got to specifically tell it what to do and like what things to include based on, you know, studies that have been done on what types of things to include in titles or.
Greg (09:55.091)
So just to clarify, this is ChatGPT4 that you guys are using? Yeah. And you subscribe, obviously, because I think it's required that you... Right. Sam Altman is listening now that he's back at the helm. So are you using anything else? I mean, there are a bunch of other tools out in the market. There's...
Darren Shaw (09:59.106)
Yes.
Darren Shaw (10:02.954)
Yes. Easiest 20 bucks a month I spent, probably.
Joy Hawkins (10:06.965)
I was like, I'd pay more, but don't tell them.
Darren Shaw (10:15.843)
I'm sure he's listening.
Greg (10:24.824)
and many of the sort of SaaS tools that people have been using are adding AI components. Is there anything else that you guys are using beyond ChatGPT in your workflows?
Joy Hawkins (10:34.36)
For images, we are using Mid Journey and Adobe's. I'm a little worried about images though, because I feel like it's very, very new. And I'm kind of watching to see what happens with some of the legal stuff and Mid Journey being trained on copyrighted material, stuff like that. I just have no idea what's going to come from that. So I feel a little better. My husband was telling me Adobe's AI was trained on their own stock images. So that feels more comfortable to me as an agency using something.
like that where you know you're not going to run into issues down the road.
Greg (11:08.403)
I think they're all indemnifying their customers against litigation. I think that most of these companies now have agreed to pay whatever fees or legal fees and settlement fees are involved. But that's a different topic. What about you, Darren? Are you using any other tools other than Chage EBT?
Darren Shaw (11:26.898)
Yes, so there's a couple. One is Bard, actually. So I didn't know this until I asked my team if they had any other ways that they were using AI tools in preparation for this. And so we were using ChaggDBT to help with review responses, re-responsive reviews for all of our clients. And so it's nice to kind of create a first draft of a review that references back to the review content. It is helpful for that. And apparently, they are preferring
review responses over chat GPT. So they actually use Bard for review responses. And so that's interesting. I didn't realize that until this call. And again, mid journey, which I find is so much better than the new dolly one in terms of generating images. So I really we use that for some image generation, but not a lot.
Mike B (12:17.482)
So I have.
Greg (12:17.923)
I actually think the new Dolly 3 that you can get through Chet GPT is actually so much better than the previous version. It's almost as good as Mid Journey, I think. Almost.
Darren Shaw (12:29.931)
Yeah, almost as good.
Mike B (12:31.426)
Here's an interesting thing I did. You know, I love the fact that as an author, I can generate a graphic art that often is a, for me, a visual joke. I don't explicitly tell people what the joke is. They have to figure it out, but it gives me the ability to do that. And the other day, David, in one of our podcasts, referred to one of Google's new search upgrades as whipped cream on SHL.
Darren Shaw (12:45.29)
Right. Yeah.
Mike B (13:00.346)
And I thought, oh, what a fascinating visual metaphor. I'm going to go to mid journey and try to do it. Well, mid journey obviously wouldn't accept poop, dung, excrement, or any of those others, nor would Dolly. But I was able to come up with a phrase that got through, which was a brown pile of overcooked mush with whipped cream on top. I gave it to mid journey, Dolly, and Dolly threw chat GPT.
Darren Shaw (13:00.593)
Right.
Darren Shaw (13:12.188)
Oh.
Darren Shaw (13:20.871)
Hahaha
Mike B (13:28.886)
And interestingly, Mid Journey had this very optimistic, glowing sort of cupcake look to it. Dolly had a somewhat deserty look. And Dolly through Chet GPT looked exactly the way I wanted it, like a little bit of whipped cream on top of a big pile of poo. And it was just fascinating because Chet GPT provides an interface layer that somehow understood
Darren Shaw (13:35.605)
you
Mike B (13:58.65)
what I was looking for was closer. That it interpreted it for Dolly the command and gave Dolly a different command than I gave it. And it gave it an image.
Greg (14:01.617)
Well
Greg (14:07.36)
That's the great thing. That's the great thing about Dali versus Mid Journey is that Dali is taking your simplistic request and it's translating it into something much more specific that the machine can then deliver. So I mean, I think that's pretty interesting. And in Mid Journey, I think you have to do the verbatim, you have to keep refining the prompt until you get what you want.
Mike B (14:30.258)
And it's, yeah, I find mid journey also with a lot of command lines. I find the interesting thing about check GPT with, with Dolly is that it's the, it does the command line work for you, which is fascinating. So anyways, if you want to see that image, it's a, in one of our most recent podcasts, you can.
Greg (14:42.183)
Right. So.
Greg (14:47.735)
Yes, yes, we can link to it in the text on the page hosting the video. So have you talked to your clients about your use of AI in any way? I mean, Darren, you're writing review responses, and you're writing content, presumably, that shows up in multiple forms on their GBP or maybe even on their websites.
Joy Hawkins (14:48.209)
I'm going to see this image.
Greg (15:16.143)
What have you said to your clients, both of you, about like how AI tools are being used?
Darren Shaw (15:22.194)
Yeah, so the clients are asking us about it all the time. They're interested to know, hey, are you guys using AI? How are you using AI? I think one of the big concerns they have is whether or not it might have a negative impact. So they're interested. They want to know, oh, is this going to cause us problems down?
down the road, but we are upfront about it. We do definitely tell them. It comes up in our sales calls too. They often ask about it or we let them know what our processes are. And so we do tell them and most of our clients are kind of thrilled about it. They're excited about it. They wanna talk to us about it and find out what we're doing and how we're using it. And we haven't had any complaints, just that like surface level concern, like, okay, are you sure this is gonna be okay? And I...
we think it is going to be okay. We don't think we're doing anything nefarious or potentially damaging to their future rankings, which is what they care about. And so yeah.
Joy Hawkins (16:17.34)
Yeah, I agree. I think the biggest concern people have is like being overused in content and then that content not performing well in Google. Um, so we kind of phrase it like we use it like an assistant, right? So yes, our writers use it, but it's to get ideas, not to do the writing. And everything that we post is human edited and, um, isn't going to just, you know, be something a bot spit out and then paste it on your website.
Darren Shaw (16:25.387)
Yeah.
Greg (16:43.504)
And are your clients at all saying, hey, this is a time-saving thing for you guys and how's that gonna impact our fees or our costs? Have you gotten any questions about that where people sort of, assuming there's greater efficiency now and you don't have to invest as much time, would it, has that come up in any way?
Joy Hawkins (17:06.396)
It hasn't on our end. I mean, it could though, right? But I would say we think it's more of a way to keep costs down, right? Because at the end of the day, it actually could work to their favor if it means that we don't have to raise prices as often because now we're more efficient. So if it does come up, that would kind of be my response, but it hasn't really come up as of yet.
Darren Shaw (17:26.506)
Yeah, and we would have an easy defense for that. We'd be like dude We're already the most inexpensive SEO company you've ever met. So It's like we were we're already Serving you great service at the lowest possible price. We want to make it accessible to everybody so we could easily defend that
Greg (17:47.307)
You just said something very interesting, Joy. You talked about your clients having enough sort of sophistication and understanding of Google ranking and search quality to be concerned that if you pushed out AI content without sufficient human intervention, that might cause them to suffer. I mean, can you elaborate on that? Because that seems to me to be a pretty sophisticated question from a client at this point.
Joy Hawkins (18:15.876)
Yeah, I mean, our clients, to be honest, are usually somewhat aware of SEO. Like they're not the average small business owner that like doesn't know how to use a computer or something. Right. Like so I would say that like maybe our clientele is a little above the average when it comes to that. But I do think that it's common knowledge that Google doesn't like crappy, low quality duplicate content. And so far, every conversation I've had with any business owner, that is what they actually expect from AI. It's low quality, crappy duplicate content.
Greg (18:45.819)
It's interesting that that's sort of already kind of pervasive in the in the in the zeitgeist. So Mike, why don't you go ahead. Sorry, Joey.
Darren Shaw (18:48.631)
Mm-hmm.
Joy Hawkins (18:51.86)
I mean, yeah. I was just gonna say, I kind of agree with their assessment. Like that's basically been my experience too. Like some of the answers I've gotten from content when I tell it to write stuff, it's like, it's terrible.
Greg (19:05.191)
Well, so just a quick anecdote for me. And then, Mike, I want to ask you about your review use case that you talked about in the Monday newsletter, where you asked a bunch of questions about review volume and star ratings. But so my younger daughter is now a fresh freshman, fresh person in college. And when she was writing her book,
Mike B (19:16.138)
Sure.
Greg (19:33.691)
essays to apply to different schools, she gave them to me to look at and give her feedback. And some of her essays were better than others in terms of polish and coherence. And so I ran them through ChatGPT at the time, and it made them all better. And I thought, this is really scary. I didn't use that, those...
chat GPT versions, you know, I worked with her to improve them in the in the old fashion, sort of human way, but that was a sort of scary revelatory moment for me. But Mike, talk about that, the review use case, because I think that was pretty interesting.
Mike B (20:16.906)
Sure. So I was working with a large multi-location client who had 10 years worth of reviews. I was just shy of 2000 total reviews and having worked with gather up and worked on the reporting for years, you know, and always frustrated by the fact that I, in the end to get some of the questions I would often have answered, I would need to get, you know, take the programmer away from whatever they're doing and get some SQL query going and whatever. So I
I took this client's 2000 reviews, put them in a spreadsheet, dropped them into JetGPT-4 and asked it to calculate the average review score and totals for each of the locations and for locations that have lower ratings in the past 12 months than the previous years, show the count of the reviews for previous years and rating averages for the previous years and the rating averages for the past 12 months. And then asked it, and then in the same query it said,
you know, if the rating is less than 4.5, how many reviews would that location need to get to 4.5? And then to provide any other insights that might help these locations perform or which ones are doing well, which ones are doing poorly. And some, like just the ability, for example, to extract this year versus all previous year's averages for 25 locations was significant. But then being able to go in and pull out.
just the ones that had less than 4.5, how many reviews they would, five star reviews they'd need to get over 4.5, and being able to compare last year's performance to all previous years was incredibly insightful. It's the kind of reporting that you would want, but it's very difficult because everybody wants something different from the data. So I see in this that chat GPT becomes a reporting extension for SaaS tools.
or whatever other data structure you have at a higher level. But it was just, it allowed me as a consultant to go back to the client and say, you know, here are the locations, here's how many reviews they need. What precipitated it was a conversation in Slack where all these local SEOs were saying, what's the formula to figure out how many reviews I need to get to 4.5 stars? And it's a basic algebra, but I thought, oh, well, doing that on 25 locations.
Mike B (22:37.546)
So that's what got me going on this. And the output was just astounding to both the client and to me. And it was a lot of fun doing it.
Greg (22:47.435)
I mean, that, go ahead, Jerry.
Joy Hawkins (22:50.865)
If that's the article you're referencing, I did read that article and I thought it was awesome, the table and whatnot. And I agree, we've had that question come up so many times and it's super annoying to calculate. And all the tools I've seen out there too, there's errors with some of them. They don't give you the exact number because they round or something like that. So I think that's a really good use case for it.
Mike B (23:10.322)
Yeah, also just the last year's, you know, last 12 months versus the previous period, right? So you could see which locations along amongst the many were going up, which were going down, which is hard for, you know, multi-location environment to parse, you know. So it was just in a little bit of time, got a lot of very interesting data.
Joy Hawkins (23:21.502)
Mm-hmm.
Greg (23:25.696)
So the.
Greg (23:29.371)
That's a super specific use case and one that I would just not have expected and a really, really complex set of prompts in there. And so are there anything comparable that you guys, Joy and Darren, that you've outlined some high level applications? Is there anything like really obscure or specific?
Mike B (23:39.591)
Yes.
Greg (23:54.419)
that you found you can do that kind of was surprising to you with ChatGPT or Bard or anything else like that, something really sort of weird and unexpected.
Darren Shaw (24:07.07)
Well, I have a great one. Um, and I actually think that Mike's example that to me is the most. Powerful thing you can use chat GPT for so much better than content. So writing Excel formulas, writing little Python scripts, writing SQL, analyzing data that for me is where, uh, I really think that that's a revolution. That's where chat GPT becomes such a powerful tool. I can like, it's like.
You could move forward in your life, never using a calculator, doing long division for the rest of your life, or you could be like, oh, we have this thing called a calculator now, so I'm gonna use it, I'm gonna make my work much more efficient, I'm gonna do things much better. I really feel like Chad GPT gives us that power now. And my example was in preparation for my local you presentation where I analyzed a million businesses, right? I had so many problems with that and Chad GPT was my best friend through the whole thing because
First, I had to import the data into a database, but I had all these problems with Newline. So ChatGPT helped me write a Python script that I could run on the data. And I was like, how do I install Python on my machine? ChatPT gave me the step-by-step. So I installed Python. I didn't look at the code. I just pasted the code in and pressed Run, and it worked. And I was like, ah, thanks, ChatGPT. And then once I had all the data in the database, I had to analyze it. So I'd be like, very similar to Mike, it's like, I want to see.
give me this data from the database, write me a query to do this. And it's like, gives me the query, I just paste it in, boom, I get the data. I was like, okay, now I want it, so I just write it in my natural language. I just tell ChatGBT what I want. And then ChatGBT gives me the SQL query for it. I paste it into BigQuery, which is where I ended up putting my data. And I got all the answers. It really powers data analysis in a way that we couldn't do before. And as Mike said, used to have to pull a developer off.
of whatever they're working on in order to get that. But you've got a developer right now, and it's $20 a month. It's ChatGPT. It's pretty awesome.
Mike B (26:09.874)
Yeah, no more lookup tables, no more pivot tables. Oh my God.
Darren Shaw (26:12.538)
Oh my god. It's so great.
Greg (26:17.252)
What about you, Joy? Anything really specific or obscure that you've used these tools for that kind of was a surprise?
Joy Hawkins (26:24.18)
No, I wouldn't say any of it's really obscure. One of our biggest use cases for it is like dealing with service area pages or optimizing for near me. Like these are really annoying things where I feel like the bot can do it better than a human because it's just like, oh, you know, put near me in the sentence and make it sound natural. It's harder to do. Like you have to kind of think and you're like, oh, you know, it's a weird kind of thing, but it works really well. So we use it a lot for stuff like that and service area pages because the content is so similar. Let's be honest, it's really the same stuff over and over again.
There's no reason to have like a human re-write all the time. So that's something like if you have a single page that's already been written once, you can kind of repurpose over and over. And you know, chatGBT can literally replace cities. You can be like replace Edmonton with Calgary everywhere it appears and things like that. It's great at stuff like that.
Greg (27:12.851)
What, so Darren, you just referenced, you've got a developer for 20 bucks a month. So how, how in chat GPT, how, how are you thinking today about staffing or use of freelancers or future hiring? I mean, has this changed your thinking about those issues? Like we can do more with our existing staff or we don't have to use, you know, as many freelancers or whatever. How has that sort of reshaped if at all you're, you're thinking about, about staffing and personnel?
Darren Shaw (27:40.37)
Yeah, I think that's a good question because at the moment I have a really stupid, not very good developer for 20 bucks a month. If you tell it exactly what you want and how you want it, you can get it, but usually it gets it wrong. You got to do a lot of hand holding with this developer. So there is no risk of my development team being replaced by ChatGPT. The developers that I have building our software at White Spark, ChatGPT,
can't touch them. And maybe in five years, that'll change. I don't know. But right now, it's like, no, Chad GBT is good for these like little one off tidbit II things that I wouldn't want to bother my developers with. But it can't build me. I can't just say, build me a brand new rank tracking application that does grid ranking. And Chad GBT will be like, I don't know. And so it really requires
Mike B (28:33.15)
Have you tried that?
Darren Shaw (28:35.118)
I will try it right after this call. We'll see how it goes. But I have low expectations. It doesn't, it's good for like very, very specific tasks. The types of tasks that both Mike and I referenced, so good at that kind of stuff, which I think is a huge benefit to my development team. It allows them to stay focused. So they actually, it does improve our productivity from a development perspective at White Spark because I am not.
bugging the team with all my stupid ideas all the time. So it's good from that perspective, but no one has any risks of getting fired, that's for sure.
Joy Hawkins (29:11.976)
Yeah, definitely no one is at risk of getting laid off at my agency at all. I would say though, like in all honesty, it probably is gonna decrease slightly, slightly not a huge amount, the amount that we hire in the next few years. Because whatever we're shaving off that's more efficient would have theoretically been people, right? Couple of years ago.
Mike B (29:32.946)
I mean, for me, it's kind of like a superpower, right? It just extends my, I know I need a lookup table. I know it's gonna be really problematic. And I think, oh, I can just use Chet GPT instead of struggling through a lookup table and having to reference instructions every 30 seconds. So it just.
Greg (29:33.32)
Yeah, I mean they're...
Greg (29:48.363)
Well, do you, sorry Mike, go ahead. Well, so in the way that smartphones have kind of gutted people's memories of telephone numbers, right? You just tap the button and you call whomever and you can't remember the number except for the sort of immediate few numbers of your closest friends and family. I used to have just a ton of numbers memorized and now I don't know anybody's number except my wife's pretty much, not even my kids.
Mike B (29:51.724)
No, no, I was done.
Greg (30:16.991)
So do you think there's any risk of that de-skilling your workforce? Do you think there's any, by analogy, like the fact that some of these tasks are being turned over to chat GPT, do you think that will impact people's skills?
Mike B (30:27.346)
When I was a boy, I had to remember this kind of story. Is that what we're talking about? I was like.
Greg (30:32.463)
Exactly, exactly. Well, I mean, there's an argument that technology creates a sort of dependency and it descales people in a certain way. You can disagree with that, but I mean, is there any risk of that with your staffs becoming too dependent on these tools, not knowing how to do certain things that they might have once been able to do at the risk of, I don't know, manually at the risk of oversimplifying?
Joy Hawkins (31:02.596)
It's a concern that I've had. I'm not sure if it's true or not, right? Like I haven't really found a way to tell if it's making people dumber or more dependent on them. But I think that's a risk, regardless of if you're using AI or not, because we rely heavily on processes. And I've also heard it said, if you have people check boxes and go through a list instead of figuring out how to optimize a page on their own, you're causing them to think less. And I'm like, yeah, that's true, but you can only do so much to keep consistency, right? When you're trying to keep the quality of work high.
you have to have checks and balances in place. I don't really see this as any different than that.
Mike B (31:36.618)
That's a great point. And the best employees are going to ask, why is this checkbox here? It's just like they should continue to ask, why did you have GPT do it that way?
Joy Hawkins (31:42.844)
Yes, 100%.
Darren Shaw (31:50.214)
Yeah, I look at it similar to the calculator analogy. And actually I love that I don't have to store those phone numbers in my brain anymore because I have very limited brain capacity. And so that allows me to have other things in my brain. And I don't have to remember those phone numbers. And I think the same thing happens. And yes, so different skills, we don't have to really focus on those. But that allows us like Mike said, it's a superpower allows us to focus on things that are
more valuable. I don't want to have to do long division on paper anymore. I got a calculator for that and that frees up my time and my mental capacity to do more impactful work.
Mike B (32:29.071)
I went to school right at the end of slide rules. And then we were just transitioning from slide rules to calculators.
Darren Shaw (32:33.139)
Yep.
Greg (32:40.426)
I thought you went to school when the wheel was invented, Mike.
Mike B (32:43.418)
No, slightly after that, but slightly just before calculators.
Greg (32:48.518)
Um.
Joy Hawkins (32:48.805)
know what that is. So, you know.
Mike B (32:52.204)
Class of 69, I'm proud of it.
Greg (32:52.363)
So I want to... Okay, so enough with the age. Yes, the age-related humor. And I started, I'm sorry. So let's zoom out a little bit. We're sort of coming, I think, to the end of the discussion. And there's been a lot of discussion. So we've had to have GPT for approximately a year. And...
Darren Shaw (32:57.175)
Good year.
Mike B (33:02.015)
I left Greg speechless.
Greg (33:20.667)
Over the course of that time, Google sort of changed its public position on AI-generated content. It started out where the perception was that Google was rejecting it, rejecting automated content, and then the position is publicly shifted from, we don't want AI content or automated content to, we just care about quality, and it doesn't matter how you generate it, right? Human versus AI or some mix of both. So what do you really think about that? What do you, you know...
Google is really, I think, struggling to get out in front of a lot of the criticism that has been leveled against it and a lot of the sort of competitive pressure that it feels. And so, you know, we have had like a bazillion algorithmic updates and changes this year, a lot of volatility. What do you think Google's, how Google's going to react to all this AI content? Certainly, sort of low quality SEO types are going to pump out a lot of volume.
using AI cheaply, and there's going to be a lot more spam or quasi-spam online. But what is your opinion about going forward, how this is all going to impact search quality and spam versus good content? And Google's reaction? Well, that's it.
Joy Hawkins (34:35.208)
So I listened to, there's a podcast in the affiliate marketing world I listened to called Authority Hacker. And I love it because the affiliate marketing world is like super spammy compared to local, right? And even they're admitting that like AI content is like, like they're telling people don't do it. Like if you run an affiliate site, like don't just publish 5,000 pages with click of a button. There's lots of people out there that'll tell you to do it. I think that Google had to change their stance on it though, to not be against AI when they are also producing AI.
And they're starting to use AI. We see it in like Google business profiles. People have tweeted that, you know, they're seeing the description now being generated by AI. So, you know, they can't contradict themselves there. But I think it's more of like, when is it appropriate and when is it not? And I would say for the majority of content that we come up with, it's not super appropriate to use AI to generate the content. And we're seeing more that Google is favoring real life data and real life stats. Like you talk about case studies, things like that.
They work phenomenally because you're putting in your own data, you're publishing something that doesn't match 500 other pages online. So we're seeing that kind of stuff win really well in local. Like if you have your own information on your website, that's not just like regurgitated stuff about a specific topic.
Greg (35:52.359)
Well, and that's kind of goes to their first person. I'm sorry, Darren. It goes to their sort of first person authority, kind of social media emphasis in a certain way. Go ahead, Darren, sorry.
Darren Shaw (35:52.458)
Yeah, I was going to say something very similar.
Darren Shaw (36:06.518)
Yeah, I was gonna say something very similar. I think that Google had recognized, well, there's no way to perfectly identify what's AI-generated and what's not. It's not like they can make some hard stance about it. And I think, like, you can really tell the difference between regurgitated content, whether it was regurgitated by a machine or a human, you can still see that it's garbage, it's nothing original. And this whole move towards EEAT stuff, so...
you're speaking from a place of expertise and authority and trust when you are doing that with your content and experience, which is the new E that they added, it makes so much sense. And that's what I'm actually really trying to squeeze out of our clients. When we talk to them, that's why we wanna do the interview style. We don't wanna just have chat GPT write the service page on Invisalign or drain cleaning or whatever it is. We wanna have the client write that page. And we know they don't have time to write.
So you get ChatGPT to write the questions. You get it from their voice. Why are you special? What do you do that's different? And then that's a new piece of content that Google hasn't seen before. And I think that is the type of content that's going to work now and into the future.
Greg (37:17.375)
So are you guys in a certain way kind of ironically saying that content is going to get better in a certain weird way because of AI, because the tendency to want to spam is there, and so Google is going to reward authoritative, real, human-created content, and that will force everybody into that sort of lane? Is that sort of an argument that you guys are impliedly making?
Joy Hawkins (37:46.728)
think it's the ideal, but we see all the time that Google's ideal, what they want is not reality, right? So then they say they want EAAT. Yes, they want that. Do they actually serve that in their search results? A lot of times? No. So I think traditional SEO plays are still super valid. Like to give a clear example, we have a lawyer client who I got to outrank the government of Texas. Is he more authoritative than the government of Texas? Hell no. But he outranks them because we do a lot better SEO than...
gov sites generally do. So I think that again, it's like a balance, but I think that in an ideal world, yes, Google wants to reward EAT, but I do not think they are there all the time.
Darren Shaw (38:29.674)
It's an interesting question though. Uh, you know, does the, you know, influx of so much AI generated content push the industry to create better content. And I think maybe, yeah, I guess I hadn't thought of that before. And maybe that is happening. It's happening in our agency. We're thinking about that and trying to like level up our content, add more expertise, not, not just sort of.
droning out service pages that are the same as everyone else's service page. I think our process before might've been more like, let's look at the top performing service pages in the big cities like New York and Seattle and Los Angeles, get some inspiration and then write our version of it. And now, uh, that isn't as good cause you end up with more of the same. Whereas when you can extract the unique voice of your specific client, that feels like better content. And so, yeah, maybe, maybe Greg, this is better.
Greg (39:21.515)
And you're using ChatGPT to enable that process. You're using ChatGPT to enable that process. Again, it's sort of an irony. Mike, did you want to add anything to this about Google? And it's kind of OK.
Darren Shaw (39:26.883)
That's right.
Mike B (39:33.914)
No, I don't. I mean, it's to be determined. And obviously Google has not just this issue, but many other issues on their plate. And it'll be interesting to see. We live in interesting times, as they say. And I'm glad I'm along for the ride.
Greg (39:48.38)
Yeah, we sure do.
OK, so as a kind of final question to wrap up, I mean, we could go on and go down a bunch of different kind of areas and explore more. I mean, this has been a really interesting, I think, and useful conversation. But as a final question, what do you guys think the impact of AI is going to be on SEO, the industry, and local SEO in particular in the near term, let's say within the next three years? What do you think?
what specific changes or...
Greg (40:25.912)
What specific changes or innovations do you think are going to happen?
Joy Hawkins (40:32.212)
I'm definitely not pessimistic about it. I don't think my job's in danger or Darren's job's in danger. I think we're fine. I think local is one of the most genuine things out there as far as industry go, when it comes to reviews and photos and so many things that you just can't fake. So I'm not really worried in any way shape or form. I think the changes will probably divide what type of local SEO agency you are. The division between the people that are relying heavily on AI versus the people that are
More like what Darren and I are talking about. Every time I think everybody's in line with us, just visit some Facebook groups and let me tell you, you'll get the polar opposite approach to how to do this kind of stuff. So I do think that division probably will grow, but overall I'm really optimistic about where the industry is headed. I don't think anyone should be afraid for the jobs.
Darren Shaw (41:21.046)
Yeah, I always think that for one word answers, like what is the square root of that? How old is this person? Like you're doing those kinds of things. Our AI assistants will answer those questions. And that I think is, if that's the SEO's world you live in, then your SEO is in danger because we're gonna get those answers from, when Siri gets up to speed, we're gonna just ask our assistant for these one word answers. But with local, you need to explore the businesses. You're not just gonna be like,
Joy Hawkins (41:25.872)
Mm.
Darren Shaw (41:51.098)
I need a plumber to come and fix my drains. Can you just call one and have them show up? That's not going to happen because you're like, I need a plumber that's not going to drag mud all through my house and break my pipes and do this. And so you want to do the research. You want to look at the reviews. You want to look at the photos. You want to get a sense of whether or not this is a real business. And so because there's that research component is not a one word answer. I think that local is one area of SEO that will continue to thrive.
You know outside of like recipes and one word answers But I do think that the way that people are getting data will change and I actually do think that Google's at risk if they don't If they don't pick up the pace like if you look at the new chat GPT voice thing it's really good and they even ask it all kinds of questions in just a Natural way. It's almost like we're achieving that movie her where he just
speaks to his assistant and gets everything he wants and gets everything done. Like I see us heading that route. And so right now we think about the map pack and ranking businesses in these three results or into the Google Maps area. It will the interface be different in the future? And will our jobs as local SEOs be to optimize our clients to be returned by chat GPT's voice thing or however Siri is pulling results. So it'd be our job to figure out how does Siri determine
which five businesses to return and how do I make sure my business is listed first? Like instead of focusing on the interface of the map pack, we're focusing on ranking our businesses in the voice results that people are getting back. I could see a shift like that happening in the next three years. I don't think it's going to happen the next year, but you know, I think SEOs will try to stay ahead of it and they'll ship, they'll pivot along with the interface as things change.
Greg (43:45.395)
Well, interestingly, just to add a comment onto that, you know, people were very excited and worked up about voice search, you know, a couple of years ago and talked about voice optimization. And it never really materialized, right? It never really happened. I mean, people use speech to text and input queries with voice, but the transformation that people were talking about never happened, but this may ultimately be that for certain...
people or certain use cases or certain segments of the population. We talked before the session about younger users really gravitating potentially to that kind of a UI as opposed to the traditional Google SERP. So it'll be really interesting to see how that plays out over
Mike B (44:33.522)
And I, two comments there. One is that the incumbents that are in your pocket, Android and Apple, have a huge advantage in that because if it takes a year or two for this to roll out, they have the resources and money to do that versus chat GPT building a piece of hardware. Um, but, uh, and then on the other side, which is the local SEO side, I think that what Joy pointed out, which is great photographs and good content.
entity issues are going to provide the foundation for any future search. And so I think that what we've been doing is we'll stay largely the same in broad parameters of making sure that the entities are accurate, detailed, down to the product level, up to the corporate level and everything in between, including reviews and workers and all that stuff. So to some extent, it's going to be the same because
They need the data to rank this stuff, however, however you ask for it and however it delivers, and it's going to be knowledge graphs driving all that. Which also gives Google a huge advantage because their local knowledge graph is still, despite its many flaws, still the best.
Greg (45:46.915)
All right. On that note, thank you, Darren Shaw and Joy Hawkins for a fascinating conversation. We'll pick it up again in another six months or a year and see where we are, see how much of this stuff has come true. Thanks for being with us. Thanks for listening. If you're in the United States, have a great Thanksgiving or hope you had a great Thanksgiving, depending on when you're listening to this. If you're in Canada, go skiing or enjoy the sun. You're in Edmonton, Darren, right?
Darren Shaw (46:15.498)
Yeah.
Greg (46:16.791)
Yeah, okay. All right, so we'll see you all next time.
Darren Shaw (46:22.882)
Thanks for having me.