Pax (02:05)
All right, Rand, thank you so much for joining us today. I'm excited to talk with you.

Rand Fishkin (02:09)
Yeah, my pleasure Paxton. Thanks for having me.

Pax (02:12)
Um, so I want to dive right in. mean, if you've got so much great stuff to talk about here, so I'm going to cut to the chase. Uh, you had, uh, an episode, um, of office hours a couple weeks ago where you had, uh, Tommy Walker on from content studio. And one of the things that he was sharing was some data that he had gathered, uh, in 2024 about audience research. And, uh, one kind of alarming stat was that 41 % admitted to not doing audience research nearly enough.

combined over half are very infrequently doing it. In our work, I've learned over the years that audience, I grew up in SEO and started doing SEO in 2007, but I've since learned that audience research is the key to taking these big swings, and it's the big swings that make big impact. ⁓ So what I think would be great to dive in today is like,

In your experience, in your POV, what are some of the things that are holding up audience research, both from an investment of both resources and time? What would you say are some of the main culprits in your experience?

Rand Fishkin (03:24)
⁓ I think I would say the biggest culprit is it is no one's job right? so when you if you look at a marketing position of any kind or a product design position engineering position ⁓ You will almost never see market and audience research listed as a must-have skill You won't see any software tools or products that are listed as hey, you should know how to do

whatever, know, use clickstream data or you should know how to use survey data or you should know how to use SparkTor, or whatever it is. That, my opinion, is the biggest culprit. don't blame individual marketers. I blame executives and recruiters and people who design job positions, you know, VPs and CMOs. Those folks...

for whatever reason have decided that, yeah, it's really important that you know how to do email marketing and use MailChimp. It's really important that you know how to use Google Analytics or Omniture, whatever it is. ⁓ But it's somehow not important that you know how to do, how to determine what your audience is doing on the internet, what their preferences are, what their behaviors are, what their demographics are. You and I know that's incredibly valuable.

but it has not penetrated the market.

Pax (04:52)
Yeah. I think one of the things potentially preventing it from penetrating the market that way is what do we mean when we say audience research, right?

Rand Fishkin (05:02)
Yeah, the definition

is fuzzy. It's fuzzy for sure. Look, here's the simplest version from my perspective, which is I want to know who my potential customers are and what they do on the web so that I can target, reach them with the best possible messaging in the best places at the right time. That's like marketing 101, you know? That's fundamentals.

Pax (05:26)
Yeah.

Rand Fishkin (05:29)
Here's the analogy I always go back to. So Paxton, you mentioned you grew up in SEO. I grew up in SEO too, professionally at least. We did keyword research every day. Like it would be absolutely mental to do SEO with no keyword research. Keyword research is one form of audience research. It is, it is the, you know, a primary and key form of audience research for

SEO because you need to know what people are searching for before you go create content and try to rank for it and make your site accessible and all that kind of stuff. It's pretty weird that in no other part of internet marketing, right? Social media marketing, content marketing, PR, email, brand, CRO, you don't also do the equivalent of keyword research. Like that's, that's deeply fricking strange to me.

Pax (06:26)
Yeah. Yeah. I, I, take that even further, ⁓ I used to teach SEO, ⁓ at one of our universities here and it was very easy to teach them the step-by-step of say keyword research. Okay. So you get here, you download this data, you sort of this way, but what I found very challenging, I could only get maybe one to 2 % of people to wrap their minds around this is to say, what, tell me about the person who is searching this word.

Rand Fishkin (06:33)
Mm-hmm.

Pax (06:56)
So when people search for queen mattress sizes, what do they want? Everyone always goes, they always say like, well, they want to know the size of queen mattresses. And I say, yeah, yeah, But like, why? What's happening in their life that they want to know that, right? And it was so hard to get people to think outside of just that one word. And this is, it's like a problem I haven't solved. Like, I don't know how to help people think broader than these like channels that they develop over time, right?

And so I'm curious, do you feel like is that just a, it's an innate attribute? Is it something that we can learn?

Rand Fishkin (07:33)
It is, it

is something that we can learn. This might be the educational process of marketing, ⁓ taking a step back, right? If you go, if you go back to the 20th century, so sort of pre internet era and you imagine an advertising agency and a client has come to the ad agency and they've said, Hey, we're selling queen mattresses. You know, we want to develop a campaign to, reach folks. The natural question, right? The ad agency is going to say.

Who's our customer here? Are we talking super high-end luxury buyers? You you're charging three, four, $5,000 per mattress. Are we talking students at college who are getting their first, you know, queen-size mattress? Are we talking about, you know, middle-income families in suburbia? Are we talking about rural area buys? Tell me who we're buying. maybe we're selling to commercial and industrial use cases where they're buying like tons of beds for...

The military is buying a crap ton of Queen beds. So it's bulk orders. Very different. Like every one of those is different. And of course the company is going to say, okay, well we make this relative quality, this style, ⁓ these features it's for this audience, those kinds of things. And then the ad agency would say, great, with those target customers in mind, we are going to say, let's say it's students, right?

So individual students and their parents almost certainly. So therefore we are going to run local advertising in college towns and areas because we know that people aren't going to want to buy those mattresses before they come to college, right? Cause then they'll have to transport them a long distance. So all those things are being thought of beforehand, but in the digital marketing era, we go, ⁓ queen mattress. Okay. How do I rank for that? And,

Pax (09:28)
Mm-hmm.

Rand Fishkin (09:30)
All of those other questions, those intelligent, of thoughtful product and marketing and audience research related questions are not there. But it can be, and they'll make you more effective. One of the things I like to, I don't know if I like this, but Paxton, when I look back on my career and I say, yeah, SEO, that was such a wonderful field for me. And it really helped my company ride this rising tide. ⁓

my personal brand too, I squint and I can see maybe a few dozen companies out of the millions that got built during the internet era who essentially built dominant market positions because they were great at SEO. Incredibly few. Almost all the companies that had massive success over the last quarter century

Pax (10:17)
Hmm.

Rand Fishkin (10:27)
built it on brand first. They might've also done SEO, but that is not how they built their competitive advantage and their marketplace and their audiences. Other channels were often more useful than brand being the primary one, but certainly there's a bunch of companies you can point to where you're like, yeah, they want on email, they want on sales and relationships, they want on...

Social like there's more that one on social I think then then pure search search is something every company does but it is not a competitive advantage, let's say

Pax (11:05)
Yeah. Yeah. I agree with that. Yeah. I was just thinking recently about like an analogy there where like the search tactic, even, you paid, ⁓ it's kind of like that's the, you're laying the tracks, you're paving the road, but the success of the tracks is, is both in the tracks and in the cargo that's going down the tracks and you can't build an entire company on just

tracks without sending great cargo down, right? And ⁓ so it's, and I think some people interpret that to say, okay, so now SEO or paid search or whatever, these are subordinate to brand or, but in my opinion, I think they're actually kind of like, you need to clear the way still for the brand. Like you need to pave that way. You need to build the track so that you can deliver that brand message. But

It's not the entire company, you know, the paid in search.

Rand Fishkin (12:06)
I think that's a good analogy, yeah.

Pax (12:10)
So who do you think should own it? Audience research then if you were, you know, waved your magic wand and say here's this fortune 500 company who with this company is going to own or maybe let's say even mid tier company who owns that.

Rand Fishkin (12:24)
Yeah. mean, yeah. The good news or bad news, if you're a smaller tier company is that fortune 500s, they all do huge amounts of audience research, like absolutely tons of it. They, they tend not to use something like spark Torah. Like they'll go and buy the data, the raw data themselves, and then do the analysis and have, you know, teams of analysts working on it. They'll buy market research data. They'll commission reports.

All that kind of stuff, right? They might go direct to, use Datos, which are clickstream data provider, but they'll go to like three or four different clickstream providers and buy millions of dollars worth of data and then analyze that. You go one tier down from that, maybe two tiers down from that, and you get into companies that are making between, you know, 20 to a hundred million dollars of revenue annually. Oftentimes they are doing little to no audience research. ⁓ And worse, they...

they often assume that they're in good place with it because they think that their agency is doing it for them, right? Or consultants or marketing team, right? So if you ask the CEO, hey, tell me about your audience research. They're like, well, marketing is taking care of that. Like they'll figure out who our customers are and where to reach them and what channels and tactics and which specific accounts to reach out to and where to target our PR efforts, all that kind of stuff. And then you go dig into it and you're like, ⁓ they're not doing that at all.

Pax (13:36)
Mm-hmm.

Rand Fishkin (13:49)
Like nobody's nobody's responsible. So I think it's a ⁓

Pax (13:51)
Okay.

Rand Fishkin (13:57)
Gosh, what would I call it? It's when you know you should be doing something because it's hygiene and good, ⁓ a good breast practice and you know that it will improve your results, but it's on your important, but not time sensitive list. And so you just push it to the background, right? Like I don't need it right now. What I do need right now is

Pax (14:15)
Yeah.

Rand Fishkin (14:21)
My boss is breathing down my neck to get this email campaign out. My boss is breathing down my neck to get our cost per click down. My boss is breathing down our neck to get our whatever number of new SQLs, know, sales qualified leads up. But my boss is not breathing down my neck saying, Hey, are you making sure that the channels and tactics we're investing in are the best ones that match up to our audience's real behavior? What's, what's crazy is you can, you can get your

lunch eaten by your competitors if they do that and you don't.

Pax (14:54)
You're right. Right. I mean, it's often overlooked, I think, because it is so basic. It's what we learn in school. When we first take our first marketing classes, you know, you have to understand this audience. think ⁓ a big transfer I would on the, you know, Fortune 500 companies that you're saying, you know, they invest a lot. Yes, absolutely. However, I feel like a lot of people would be surprised to learn the number of times like we meet with

Rand Fishkin (15:03)
Yeah? Yeah.

Pax (15:22)
a marketing team at these giant organizations and they, don't have that data or they have it in a deck that they never look at. And, you know, and so it's like, you, you're still competing with barriers around like data silos. And now what do I do with this? Or it's so much information that I don't know how to process it. So I think that's another area where people get stuck. ⁓ you know, I'm looking at all this information that has been provided to me. These are like,

Rand Fishkin (15:27)
wow.

Pax (15:52)
dots and then I have to figure out how do I connect them. Do you have any tips for people as they're staring at a sheet of data? How do you connect them in creative ways to better reach an audience?

Rand Fishkin (16:04)
Ooh, you know what I actually recommend now rather than, you know, rather than trying to like fight the tide and Hey, you know, you should, you should get this data before you need. What I recommend now is do you have the problem that this data will solve? So for example, ⁓ maybe your boss comes to you and says, Hey, ⁓ we've been thinking we need to do some more, you know, YouTube investment. Like we think that some of our customers, can you like, go figure that out? Okay. Now your job is what percent of our audience is using YouTube?

Pax (16:19)
Mmm.

Rand Fishkin (16:34)
Which channels are popular with them? What videos? ⁓ Should we, what are the options for tactics? Like can we get on people's shows? Can we, should we do comment marketing in there, right? Like old school type of stuff. Should we ⁓ try to create our own YouTube channel? Should we buy like specific, you know, one-to-one advertising where we can sponsor a show and be mentioned by the host and talked about kind of like podcast advertising. Great. So.

Now that's your job. Now you should go do audience research. Like now that you have the problem, go, go solve it with data rather than, well, boss, I spent a few hours watching some YouTube videos and like, here's what I learned. No, please, please don't do that. Please, please get data at scale. Right. That is informative. The other thing, unfortunately, people are starting to do Paxton. They're starting to ask chat GPT like, Hey, what percent of

Pax (17:18)
Thank you.

Rand Fishkin (17:32)
student mattress buyers, you know in whatever the Salt Lake City area are using ⁓ YouTube and which channels they subscribe to and If you know anything about how large language models work, you know that what it will do for you is give you Stereotype dancers, you know spicy autocomplete The these words most frequent. Yeah, very confidently it will say

Pax (17:54)
very confidently.

Rand Fishkin (17:59)
Students in the greater Salt Lake City area ⁓ use YouTube 84 % of the time for their mattress research. Like, well, where did that come from? It came from, here's a bunch of documents on the internet. The word, you know, the number 84 % followed, you know, some research report that had the words YouTube and like students and mattress somewhere on the document. And so ChatGPP presents that.

And then you're like, can you, you know, give me a source for that. And then you start clicking links and you're like, yep, none of this is, that's not true or accurate. Very few people, unfortunately do the followup to figure out like, where does that come from? It's like asking chat GPT for keyword research in SEO. You know, it's a terrible idea. It's easily provable. You can say how many times do people in Salt Lake City search for a mattress variance and what are the most popular variants? And then you compare that actual keyword research data. You're like, ⁓ yeah.

Pax (18:39)
Mm.

Rand Fishkin (18:55)
You don't know what the hell you're talking about because chat GPT doesn't have the data, right? It just, it's just giving you words that frequently come after other words on the internet. So you should go gather data from sources that collect it, you know, ⁓ in a real way. And then you can upload that to your LLM of choice and get some interesting stuff. But I just, I just worry that, ⁓ yeah, man, this, this research, if you need to do it can go.

way off the rails if you don't know what you're doing.

Pax (19:27)
anecdotally, I had a, I was in a CMO roundtable just last week and the question went around, what's, what's your favorite AI tool? How are you using it? And I was blown away at how many times people just said chat, GBT, that's just my only tool. And it does everything for me. It writes all my ads, it writes all my content. And, and, ⁓ on the one hand, I was like, ⁓ man, I'm kinda, I'm dubious about the output of that.

But on the other hand too, so many of them are so strapped for resources and time. That's like, get it. I get, you know, why you ended up there. But I think, and you know, this is not intended to be a commercial.

Rand Fishkin (20:05)
Well, it makes it easy to compete with them.

Pax (20:08)
Yeah, yes, it does. It does. But you know, but to your point, it's like whoever has the best data wins. And we are in a time now where there's so much data, great data, but spending that time like,

Rand Fishkin (20:09)
You

Pax (20:22)
Again, I don't intend for this to be commercial, but Spark Toro, we use Spark Toro at 97th floor. And the amount of data that you get from that is so great. And it's also so fast that that I think is if you can take an extra two to five minutes and then work with AI, you you're going to be light years ahead if you're providing it with better data.

And I was interested to see how you, cause you've made it clear how you're dubious of AI and I think rightfully so. ⁓ So I was interested when you guys released the AI component into Spark Toro, but you did it in a way that I think was very counterintuitive or how like it wasn't what I assumed it would be when I first read the headline. Do you want to describe like your thought process into like having it be more about the search than it is about the data?

Rand Fishkin (21:19)
Yeah, yeah, exactly. what, you know, when we've been talking here, right, we basically said like, AI produces words that frequently come after other words. And if you need that, it's awesome for that. If you need to know like, Hey, what are the concepts and topics that are most associated with ⁓ student housing in my, at this college? AI is going to do the absolute best job of like any product for that. It's really, really good.

Where, where things fall apart is where you say, Hey, give me information that, ⁓ that this, that it doesn't have. Right. So like privately collected, you know, clickstream data or my analytics data or, ⁓ keyword research data or, know, any of those kinds of things in the marketing world, you need to send it that stuff first. Like you've got to tell it, Hey, here's all these things. So.

When we were designing this, we're talking about a feature called conversational queries that the Paxton is referring to very kindly. It's something we just launched at SparkToro. Basically the problem was lots of people have an audience that is hard to describe in, you know, as just people who visit a particular website or people who search for a particular keyword or people who have a particular term in their bio or profile job description, whatever. so Casey and I were always struggling with this. Casey's my co-founder at SparkToro.

We're always struggling with like, how do we make it so that people can describe exactly the audience that they want to go after? Like if you know, Hey, I don't want everyone who visits my mattress website. Or for example, like we're Ikea. Like people visit us for bajillion reasons. analyzing the Ikea audience is useless. What we really want are students in Utah going to this university. They're in the greater Salt Lake City area.

They're looking for their first mattress. want a queen Tell me about that group of people like I want information about them. I want their demographics. I want their behaviors I want to know what YouTube channels they subscribe to you. I want to see which subreddits they're on I want to know what podcasts they're listening to I need to know what websites they're visiting I want to see what search keywords they're using all that all that stuff so I can target my marketing In case you never like okay actually AI is a really good tool for you write a long description it can be

Rambling it can be short. It can be in bullet point forms. It could be sentences. It could be a few phrases You describe it. However, like your brain works the AI can then turn that into ⁓ We essentially ask it for hey from whatever the user entered give us ⁓ Behaviors specifically like like websites search keyword behavior, you know that kind of stuff

Pax (24:02)
Thank you.

Rand Fishkin (24:10)
And then we will take the set of outputs that the sort of stereotyped outputs that it's given based on this. And we'll turn that into a group of audience behaviors. And then we'll analyze the audience that does those overlapping things. That produces very accurate results because AI is really good at stereotyping. It's really good at saying commonalities of a, a, you know, set of words and turning that into this stuff. So.

Pax (24:28)
Mm.

Rand Fishkin (24:40)
What's nice now is you can go to SparkToro and you can say, Hey, my audience is software engineers in the UK who ⁓ are heavily using, I don't know, Pearl, you know, like from 15 years ago or whatever. And tell me about that group. Or you can say, Hey, my audience are ⁓ early content creators who are buying their first.

Pax (24:53)
No.

Rand Fishkin (25:06)
high resolution digital camera, they need to use it on the go, they have these sorts of requirements. Great, you just describe the whole thing to us and we'll give you actual audience data about that group of people based on this translation layer that LLMs can do. We don't promote it as being very AI-centric. Yeah, sure, it uses a large language model.

So that's not, you know, we're not an AI powered tool or anything. We're just using AI to make this one process that it's good at easier, which in my opinion is how all software companies should be using AI. Do your, do your customers have a real problem? Is that problem best solved by a large language model? Great. Do that. If the answers to those are no, you don't need to adopt AI.

Pax (25:45)
Mm.

Thanks.

Yeah. I'm not that there's some parody there. think between, ⁓ that view and what you had stated earlier about how to, and when to employee audience research where it's like, I'm not going to just do it randomly and all the time. I'm going to say, I have this specific use case in this thing that needs to be solved. Is this the right answer? Like way to solve it. Right.

Rand Fishkin (26:21)
Yeah, yeah, I think that there's, look, is there value in doing, you know, sort of some research ahead of time where you're like, hey, I want to understand how AI works. I want to not be behind the curve. Great. Yeah, do that. I'm not saying you should stop that. Same thing with audience research. Do you want to like every year do a few reports to like get a sense of your various different audiences? Okay, like what's changing with them? gosh, you know what? They're using more Reddit than I thought. Or, hey, these people are...

starting to adopt AI tools, my audience starting to adopt AI tools at much faster than average. Great. But like, yeah, it does not, it does not need to be, ⁓

you can have the problem and then get the solution.

Pax (27:10)
Yeah, yeah, yeah. So one thing that I, as I reflect on like our work as an agency over the years, ⁓ they're like our biggest, best campaigns, all of them have some connection with an insight into this audience. And that allows for innovation or twisting something on its head or doing something they haven't seen before, all in the name of like converting. So,

I'm interested in, you know, as we like finding that insight faster and finding more unique insight ends up then being the name of the game. If that's the Genesis of every great marketing campaign. Right. And so I'm curious, ⁓ one thing we've been stuck on in this, I mean, maybe you've, you've, we've already touched on this, but I, I do want to hit on a little bit more. There's a degree of serendipity both in the research and

the execution.

so years ago you talked about like investing in serendipity, which is like, you know, I'm going to invest in some things. I'm not sure if I'm going to get ROI on that. We'll see. ⁓ I also find that like with research, a lot of our best ideas come from serendipitous insight where we weren't necessarily looking for it, but we were looking for learnings, but we learned, Hey, there's this weird thing about this audience that they do this. You know, we just did this with, ⁓

Rand Fishkin (28:27)
Hmm.

Pax (28:36)
⁓ a pickleball paddle company and we learned, you know what is a really big deal in the pickleball world is who your partner is. There's like a partner appreciation day and like it's actually really, it's almost like another relationship who your partner is. And so can we do some campaign on that? But we didn't go looking for that insight. was just kind of, we were just looking around and it's hard to tell a marketer, just keep looking until you find something. And I can't tell you if that's going to come in five minutes or five days, right?

So I'm just curious, the serendipitous nature of research and those big ideas, what would you tell a marketer today? Like, should they be investing their time in that or with the, how it being unpredictable, is there more methodical way to look at it?

Rand Fishkin (29:21)
Yeah, I think and there's there's so many threads to pull on this. Okay, so what one thread to pull for sure is that? serendipitous marketing has never been more valuable because almost every channel and tactic where you can You know, hey, I'm gonna invest a dollar here. I know I'll get a dollar and a penny back Almost every channel like that

is completely saturated, incredibly competitive. Think Google AdWords or, you know, Meta ads or LinkedIn ads, whatever. Like those spaces where you just pay to play. Those are filled with every single company you're competing against. Every one, the fortune 500s, the startups with tons of VC cash to burn the AI companies. Like everybody is pouring dollars into those channels. ⁓

And as a result, it's very, very difficult to have any kind of competitive advantage. But the inverse is also true, which is any channel where you can't quickly prove ROI, right? And you don't have great metrics around it. CMOs hate it. CFOs hate it. So nobody invests money there, which is why people are like, my God, you know, we did some billboard advertising and we saw a crazy lift in branded search, for example, in our area.

Or we did some guerrilla campaigns with like, you know, posters and like, you know, just tacking them up on, on telephone poles around our city. my God, like we saw a 15 % lift in month over month sales from this weird thing that costs us like 500 bucks to have some neighborhood scamps like run around and, you know, tack these things up like, my God. Or, and the thing is you can't prove that the 15 % lift came from that. You can't prove that anybody saw.

Pax (31:07)
and

Rand Fishkin (31:15)
You know your ad on a telephone pole same thing with we can't prove that because You know a bunch of reddit threads in this subreddit started talking about our brand and company We had more sales We can't prove it because there's no direct link the referral strings missing on on mobile reddit doesn't even pass referral link even if there was a link Redditors hate links they hate sales like all that kind of stuff you were mentioned on a youtube video great And then what happens somebody goes to google and searches for you?

And the CFO is like, yeah, put more money into Google. You know, like just everything comes right comes back to this. So those channels and tactics, the serendipitous ones, the hard to prove, hard to measure, huge opportunity. Your competitors are not putting dollars into it. And that's dumb. It's just, it's just dumb. Like anybody who's there's no one who has this conversation with me who's like, well, I don't believe that everybody agrees. And yet nobody puts their money where their mouth is.

Pax (31:47)
This is so cool.

Yeah. Yeah.

Rand Fishkin (32:14)
So you can be the exception and win that way. The other thing I'd say about serendipity,

you know, in terms of like ideas is that this is, it's frustrating, almost nobody does this, but yes, there is a huge amount of value in getting, gaining a deep understanding of your audience and customers and then, and being open-minded to opportunities that come from that understanding. I'll give you an example. We,

So at SparkToro ⁓ after, know, sort of, so Elon bought Twitter, what was that? Like November of 22, from like maybe March of 23, ⁓ SparkToro was initially built like our key data source, our primary data source was Twitter. We had to get off of that, rebuild our whole system. We built it on top of clickstream data instead, which has lots of benefits. And I like it better for million reasons.

Pax (32:53)
Mm-hmm.

Rand Fishkin (33:13)
but it's also extremely expensive and ⁓ hugely time consuming, which is why we didn't start with it. fast forward 18 months and Spark Toro had basically had like, had declined about 15 % and we were like ⁓ kind of cruising, plateauing, flatlining in terms of growth. Casey and I were looking at each other like,

This sucks, man. We were on this great growth trajectory. Everything was going gangbusters and then this big data change. We commissioned ⁓ Asia Arangio from Demand Maven to do some research for us, basically customer research. Her process is she goes and interviews 20 of your best customers for your SaaS product, interviews 20 people who look like they have all the ⁓

Pax (33:53)
Mm-hmm.

Rand Fishkin (34:07)
Attributes of those best customers but are not yet using or tried the product but didn't adopt it and she asked them a bunch of questions about their usage and their behavior and what are they doing she came back to us with this like a bunch of research a bunch of videos like she has video calls with everybody so we can watch the videos and stuff and she's like here's the reality you think you being Casey and I you think that people are using spark toro to say hey I need to do some YouTube marketing

Which channel, you know, which YouTube channels do I need to do that on and where, and let me get the email addresses of the people running those channels so I can do outreach to them or, plug them into my Google ads account or whatever it is. She was like, that is, that's like 20, 30 % of spark toro use. Most of the spark toro use is people go to the report, they download the CSV and then they make a pretty chart or graph, which takes them forever that they can put in a spreadsheet or in a presentation.

and show their boss or team or client to try and get budget to do the investment. That's the, like SparkToro is way more strategic, high level, prove that a channel is worth investment, way less tactical, find the channel. And we were like, ⁓ crap, we've been building all these like things to help the 20 or 30 % and nothing to help these people. You know what we should do? We should make customizable, pretty graphs right in the product for every single section so that when you get the report, it's super visual.

We launched that in May and then we had like our best three months in a row. You know, tons of people using it and like it was great. Like was very serendipity, right? It's like, it's like, go do the research, find this opportunity and then like build the thing that people want. my God, look at that bunch of growth. ⁓ even in a really rough time for the SaaS market. So, geez, I mean, to your point, I, ⁓ I cannot.

Pax (35:40)
You

Rand Fishkin (36:05)
I can't recommend that enough, but so few people are willing to do it.

Pax (36:10)
Yeah, yeah, so true. ⁓ You cut up at the very end of that. it's so frustrating. Yeah.

Rand Fishkin (36:18)
I was just saying that it's,

yeah, it totally works, but so few people are willing to do it.

Pax (36:23)
Yeah. Yeah. 100%. ⁓ I want to get into, ⁓ just very quick to, ⁓ towards the end here, your thoughts. mean, you've been quite prolific on your thoughts around Google and AI and the future of search as a channel. And, ⁓ I know you just held a webinar with data sharing some of your, ⁓ findings there. I'd love to, would you mind sharing some of the highlights from that?

Rand Fishkin (36:52)
sure. so there's lots of speculation Paxton. I'm sure you've seen it that AI is replacing search or reducing the amount of search activity. If you squint really hard at all the numbers. ⁓ so we use desktop numbers, which is way friendlier to AI, right? Like, like AI tools are far more used on desktop than they are on mobile. ⁓ if you squint really hard, you cannot find.

the narrative in the numbers that AI is absolutely growing. And during the last two years where AI has gone from, you know, 18 % of Americans using it to almost 40%, like 38 % of Americans, you know, visiting an AI tool once a month or more, search has not declined. you know, visits, number of searches per searcher, number of total searches, Google, Yahoo, Bing, DuckDuckGo, that there's no

There's no slope to those graphs. Like they are, you know, they bounce around a little bit. Maybe you could look at like one month and be like, but look, it looks a little down here. No, you can't find it. ⁓ and when people say AI is taking my traffic away, I agree with them a hundred percent, but it is not in the way you think it is not. People are switching from Google to chat GPT or perplexity or deep seek or Claude.

It is people using Google and Google instantly answering results using AI or AI overviews or whatever. And zero click searches rising. That's, that's the reality and rising far more on mobile than they are on desktop. Like, ⁓ I think we looked in it's maybe, you know, zero click searches are 10 % up five, five to 10 % up on desktop, but on mobile, you know, it's off the charts. Right. And so look, the.

⁓ the media narrative is really powerful. Like if you want to get clicks as a journalist, you should write a story about how chat GPT is decimating Google. People will click on that shit, right? They will read your story and you will get some money. And so, and that is why that media narrative is so prevalent. Same thing with AI taking people's jobs. Like if you look at the big Stanford study,

that was done a little while ago, you can be like, okay, for entry level jobs and like these five roles, you can see a 13 % decline, you know, since the AI era. Chat GPT is taking like, you know, less jobs than tariffs are. Like it's just, you know, it's so, it's just so obvious in the data, but journalists aren't gonna write about that because that doesn't get clicks. Your aunt Sally is like, my God, Paxton, you're not gonna have a job next month. I just read in the New York Times, you know?

It's fear that sells and so that that fear that newness that like you should be scared of the new thing That narrative is all over the place and and the data does not support that so that was one of the biggest findings I think one of the other interesting findings going down in the details is There is a little bit of a decline of YouTube's dominance in ⁓ Desktop visitation so like YouTube I think

Pax (40:09)
Mm.

Rand Fishkin (40:12)
A lot more YouTube activity is switching to mobile devices and the mobile app. People are watching less YouTube on their desktops. This is true in both the United States and the EU and UK. ⁓ that's an interesting behavior change from years past where, know, more than 75 % of all devices were watching lots of YouTube videos every month. Reddit on the other hand, the reverse we are seeing, you know, it's, it's now, ⁓ over a quarter of us devices are active on Reddit every month.

Pax (40:23)
Mm.

Rand Fishkin (40:42)
That platform looks like it's on the rise on desktop. Their mobile app installs have been huge. They've been in the top 10 on the Apple ⁓ Store and the Google Play Store. I think, yeah, there's some interesting changes that are happening around that stuff. ⁓ Another big one that ⁓ I think is really interesting is

Pax (40:57)
Hmm, interesting.

Rand Fishkin (41:10)
About 20 % of Americans now are heavy AI users. So meaning they visit one or more of the AI tools 10 or more times each month. 20 % heavy users. And yet those same people perform more searches on average than the average Google searcher.

Pax (41:32)
Mm.

Rand Fishkin (41:33)
So you know what that tells me? It tells me that like this 20 % of heavy AI tool users are also heavy Google search users. These are not cannibalistic. That group of people are just heavy internet, like researchy type people. I bet it's people like you and me, right? Like I do a lot of chat, chat between, I do a lot of Googling. I do a lot of redditing. do a lot of YouTube. I'm, I'm probably a terminally online person, right? And like a fifth of American devices are, are that.

Pax (42:02)
I think we've potentially made a mistake in conflating the two to say like chat to BT is the new Google. when it's like, I mean, someone else could say, Hey, Facebook came along. This is going to eat up Google. It's like, not necessarily. Those are two different tools, two different things. Yeah. Yeah. and, ⁓ I'm curious the flight from YouTube. Do you have a theory as to why, like, do you think advertising and like increase in ad?

Rand Fishkin (42:18)
Yes, thank you for remembering that.

Thank you.

Pax (42:32)
like add interruptions versus the quick answers of Reddit or do you have any theories as to the flight from YouTube?

Rand Fishkin (42:39)
If I had to guess especially over the last two years, I would say it's tick-tock I think I think tick-tock and short form video, you know Instagram reels I think that's taking some of it YouTube shorts for sure is doing well also ⁓ But yeah, I don't think this is a death knell for YouTube at all like I you know, we're talking about it maybe five six percent drop on desktop, but my guess is a lot of that is just switching to mobile so people are like, I

Pax (42:44)
Hmm.

Rand Fishkin (43:09)
I like watching YouTube on my mobile device more than I like watching it on desktop. That's my theory.

Pax (43:16)
Hmm.

Yeah. Yeah. I like that. ⁓ well, I, I am so grateful to you for your time, on the show. This has been a really great conversation. I just want to, ⁓ wrap up with one call to action for, ⁓ actually probably a couple one from a, like marketing philosophy perspective, ⁓ would be invest in audience research. ⁓ too many marketers have channel blinders on.

and they're fighting for incremental improvement. But the way you're gonna put your company on the map is gonna be through big swings. And the big swings become less reckless and less risky when you have solid audience insight and you know it's gonna resonate. So you gotta take those risks, gotta take those swings. And to be honest, it's more fun. It's way more fun than fighting for an extra 3 % here, an extra 4 % there. So like have fun. So that's my call to action for people listening today.

Rand Fishkin (44:06)
You

Pax (44:14)
Rand, you've got so much stuff going on. have ⁓ Spark Together coming up. What would you like to call attention to and what would you like people to engage with?

Rand Fishkin (44:20)
Yeah.

Gosh, so we talked today about this ⁓ Dados report. You can find a bunch of those charts and graphs if you wanna like prove this to your boss, your team, your client on the SparkToro blog or Dados has their first half of 2025 state of search research from all their clickstream data that I worked on with them. That is also published. You can grab that report, big old PDF. The thing I would say is...

These macro trends are interesting. They're fascinating. I care about them. They're fun to discuss. But what you care about is your audience. You care about whether your users are adopting AI tools at greater than average rates. You care about whether your users are using more AI, more YouTube or more Reddit or more Pinterest or TikTok or whatever it is. And to figure that out, you need data at scale that is passively collected.

You know, and monitoring this stuff, cannot ask people in surveys and interviews to give you this information. It's like asking people, Hey, what did you buy at the grocery store last year? What? No, just look at the receipts. Like, go look at the receipts and clickstream data is looking at the receipts. You can use SparkTorrent for this. There's other products out there you can use. Certainly you can do custom buys with the clickstream providers. But I would, I would urge you to find out what your audience pays attention to where those opportunities are.

Not just the macro level just because ai tools are on the rise doesn't mean they're on the rise with your audience just because youtube is declining slightly on desktop doesn't mean they're declining with your audience just because reddit's getting adopted doesn't mean your audience is on there and if they are Which subreddits are they on and what are they talking about and like what do they care about? You should know those things. That's that's part of your job as a marketer. Even if your cmo didn't tell you

Pax (46:11)
I love that. love that. Well, thank you so much for being on the show. Really appreciate it. ⁓ and, thankful for everything that you do in the industry to share your knowledge.

Rand Fishkin (46:22)
Yeah, my pleasure Paxton. Thanks for having me man. Take care.

Pax (46:24)
Okay. Take care.