In this podcast episode, Eric Siu and Neil Patel discuss a range of current marketing and AI trends. They start by exploring AI-driven cold‑calling platforms such as air.ai and bland.ai, which can ingest a sales script, fine‑tune prompts, and make thousands of calls automatically, raising the question of whether traditional salespeople will become obsolete. They compare these tools to conventional auto‑dialers and note that while AI can boost efficiency, true consultative selling still requires human nuance, especially in the short term. The conversation then shifts to how businesses can "rank" on ChatGPT. Patel describes a study where they asked ChatGPT 100 recommendation questions, extracted 82 possible factors, and identified six that strongly correlate with being recommended: relevance (0.91), brand mentions (0.87), reviews (0.61), authority (0.52), age (0.46), and recommendations (0.28). They explain that building these signals over time can improve a brand’s presence in AI‑generated recommendations, much like traditional SEO. Next, they highlight the newly launched OpenAI GPT Store, likening it to the early App Store, and suggest that early‑stage creators can develop niche GPTs (e.g., marketing tutors) to gain visibility. The hosts compare several AI assistants for research and entertainment—Grok, Perplexity.ai, Bard, and ChatGPT—pointing out that Perplexity excels at detailed research while Grok produces more entertaining content. They then discuss Netflix’s ad‑supported tier (23 million users, $6.99/month) as a fresh advertising channel, noting the potential for marketers to reach a sizable, cost‑sensitive audience. Finally, they present an experiment on the impact of listing on Amazon. Analyzing 18 e‑commerce sites over a year, they found that overall sales grew (e.g., 103.9% in months 1‑3, 135.8% in months 7‑9) despite a shift of some revenue to Amazon, indicating that Amazon’s brand lift and ad spend can benefit a seller’s direct‑site sales over time. They also touch on the importance of a strong hook in social media posts, using a low‑performing tweet about the Amazon experiment as an example. The episode wraps up with a call to action for listeners to rate, review, and subscribe.
Chapters
Is this the end of salespeople? So here's the thing. There's crazy new tech coming out right now where you can call 500,000 people at once simultaneously. Have you heard of this before?
No, but I know yesterday you and I were at an event and someone told us how AI can make your sales team perform 5 times better.
Neil and I were at this event and we didn't stay very long. But anyway, so this product is so— by the way, I've talked about this a couple months ago. There's a product called air.ai. And then this one is called bland.ai. So B-L-A-N-D dot A-I. Here's the thing, they got money from Y Combinator, which to me automatically means they're more legit.
Yeah, so it's not easy to get into Y Combinator.
Here's the thing, um, basically with this bland.ai now, you can actually— you know how ChatGPT, you can tune the prompt, right? You can fine-tune it. Yeah. And you can do that with this. You can say, hey, like, here's a script that we use over here. Your name's Jeremy. Oh, sorry, am I boring you?
No.
But anyway, so, you know, I know, just yawn. But basically you can say, okay, here's our script, here's our entire process, blah, blah, blah. And then you're going to have a conversation. You can just keep tuning it, tuning it, tuning it, right? You can speed it up, slow it down, and then it can make calls rapidly for you. I'm watching the video over here. It's like, boom, boom, boom, boom, boom, boom, boom, right? And then it'll have like— it's maybe like 5% awkward, but for the most part, I saw a demo of this where they actually booked a restaurant reservation for like 12 people. And the guy sounded a little annoyed, but like the reservation was booked. So like that's where the future is going, especially like we get a lot of these leads that we just can't serve. But imagine if we just had an AI that can route it to other agencies and give them those leads. That'd be pretty cool.
So it's more like an AI assistant. So like you would say, hey, I want a reservation here for 12 people and they just call and they figure it out.
It can be an assistant or you can just have it on your site and it can be like your call agent, like your inbound call agent. So it can handle inbound, it can be a customer service center, it can do all those things.
Yeah. Dude, you know what's funny is I don't think AI doing calls for you in the short run is going to produce better results. Maybe in the long run. We do use technologies to start dialing a lot of people so that way, you know, our reps are more efficient and that has helped us a lot.
Like what are those called? Auto dialers?
They're auto dialers. And we found that helps a lot. But selling is all about being consultative and helping them. And I know AI can help with that.
At least for us.
Yes. At least for enterprise, for selling alarm systems, a little bit different. And I do think AI can help even in the enterprise end. It's just the technology is not there yet.
Yeah.
And I'll give you a good example of this. So I release a lot of data on Twitter and on my social profiles. A lot of them have charts and I have Solo create them for me. Okay? So once we have the data and our team is done analyzing, someone has to turn them into beautiful visuals. So Solo typically creates them. Sometimes we give ChatGPT, as a fun test, we give them the data and we tell them to create charts. And a lot of times the charts that they create are off. We're like, how the heck are they off when we're giving you the data to create, we're telling you the exact data and where to chart it and it's not right even 80% of the time, it's off quite a bit. That's basic. It shouldn't mess up on those things. Now, I'm not saying that means ChatGPT sucks. I think it's one of the best tools that you can get, especially the paid version for $19 or $20 a month.
$20.
So it's just like, I just don't see AI being there yet. And funny enough, speaking about ChatGPT, have you ever tried ranking on ChatGPT?
Yeah, we've talked, we've talked about this a little bit. Yeah, but go on, go on.
So I know ChatGPT's data is based off of September 2021. But that doesn't mean eventually it won't be updated to real time like Bard too. I think it's 2021. Well, whatever, either one, it doesn't matter. Still old. Bard is typically a few weeks old to a month old at the latest. So it's closer to real time.
Yeah.
And I know you can't quote unquote rank on ChatGPT, but they do recommend services, companies, products. So what we did is we asked ChatGPT over 100 questions to figure out what makes them recommend a company or a product. We asked them questions like, what's the best laptop for college students? What's a great credit card when you have a low credit score? What's the best electric SUV? What headphone brand should I buy if I'm watching movies You know, and I then— we then took the question and the answers, gave them to a statistician and data analyst to see if there was any patterns. So they analyzed 82 factors, but they found that 6 of them had a strong correlation. And here's what they found that matters to ChatGPT.
By the way, guys, here's the hook. This is how you rank highly on ChatGPT.
Yes. So brand mentions, 0.87. Reviews, 0.61. Relevancy, 0.91. Age, 0.46. Recommendations, 0.28. Authority, 0.52. Now, if you build up those factors within your business, it doesn't mean the next day you're going to rank on ChatGPT.
And by the way, this is a score of 0 to 1, right?
Yes.
Okay.
As they continually update their index and you do these things, just like with SEO, it takes time. Eventually ChatGPT can start recommending you more often. If you haven't already done them in the past.
What was the number 1 again?
Number 1 was relevancy, 0.91. Number 2, which was close, is brand mentions. That's how many people mention your company throughout the web or whatever they're crawling.
I think this has evolved quite a bit because what we said, like this is conjecture, you know, earlier in 2023, we're like, yeah, you know, it's going to be about citations, you know, links, number of mentions, things like that. I think still when you search ChatGPT, what's the best marketing podcast? It's still Marketing School. So I think we're good there.
But that's also relevance.
It is citations, relevancy, whatever.
Brand mentions and relevancy. And then authority, we have good authority on our sites. We also have a lot of reviews at the end of every podcast. We ask for reviews and then we've been doing this for a long time. We've been around for how many years now? 7?
Yeah, 7, 8 years, something like that. So yeah, we have thousands of episodes. So citations, relevancy, that stuff's all there at the end of the day. So I think that's interesting. You know what would be interesting later if your people can go and ask and say, hey, how did you decide the ranking for this, like who are the best marketing podcasts? How can one rank higher on ChatGPT for this?
Dude, we've asked that. It doesn't come up with useful information.
What does it say?
It depends what you ask. But like, we've asked that too. Have you ever asked ChatGPT on what makes a website rank really well on Google?
Have you ever asked ChatGPT to make an atomic bomb?
No. Have you?
Well, this could— I'm just reading it here. He's like, one guy was like, I love this tool, but it kept interrupting me. And I asked it for a recipe of an atomic bomb and it told me.
It really did tell you?
No, it's just a joke. Oh, okay.
I'm like, that would be crazy. But no, like, we find that Bard is better when it comes to analyzing a lot of information and figuring out what's causing things to rank or not rank. Yeah, we just found ChatGPT is just not as good.
Yeah. By the way, I mean, speaking of ChatGPT, they just launched the— OpenAI just launched the GPT Store yesterday. And this is a good one. When you think about Apple, when they first launched the Apple— the App Store, Like, ton of people started going into it, right? So now when you go to OpenAI, they have a GPT store. And then I can see in here, like, there's a code tutor for Khan Academy. We should probably make a marketing tutor at some point. Just give it away. And there's a designer GPT helps you create and host beautiful websites. Canva has one effortlessly design anything, presentations, logos, social media posts, and more. Yeah, an AI PDF. There's a lot of plugins in here. And I think there's just going to be more and more that will be released. So that's an opportunity. Like, if you're just starting out right now, maybe you might want to look at the store and get in there while it's hot. Neil and I, we have stuff to do. We might— we'll probably make some GPTs.
But yeah. Dude, you know what's funny? Speaking about GPT, have you ever used ChatGPT, Bard, and Grok to see what creates the most entertaining content?
Grok makes the most entertaining content.
Grok makes the most entertaining content. So I recorded a video about this. This morning. And then, you know, my team asked me after, like, do you still use Grok? And I'm like, no.
Yeah, I don't really use it either. Um, but you know what I do use? Um, perplexity.ai. Have you used that?
I've used it in the past.
It's great for research. I think that one's the best for research. Like, I ask about all this, you know, I take a lot of supplements, right? I just keep forgetting, like, what does this actually do? I'm putting all this stuff in my body. So I'm like, okay, it makes me feel better. Um, they give me more complete answers versus ChatGPT and Bard. It's not as complete.
Yeah. The thing that I found is if you want really in-depth knowledge from like ChatGPT or Bard, you can't just say, here are the 10 supplements I'm taking, tell me what they're for and please analyze them and tell me how they work together. You got to actually start doing one at a time. And then you can say, hey, now, you know, and then when you start combining the data from all of them, then you can get the answer that you're looking for.
Well, we said this last time, you have to let the prompt think just like a human being. You can't just drop it all at once.
Not the prompt think, you have to let the AI think.
Yeah, yeah, yeah. Yep. So look, one more thing on my side. I think this is interesting. So did you know that Netflix now has 23 million users on its ad-supported program? So, and that's—
That means those are people paying a cheaper amount. Now with Netflix, did they automatically, whatever you're paying, just automatically put ads in it and then you had to upgrade for non-ads?
Yeah. So basically look, in the US, the plan costs $6.99 a month, less than half the price of the standard plan. And I think even with standard—
Wait, so they released a new plan?
Yeah, there's a cheaper plan.
But back in the day, there was no $6.99 plan.
No, no.
Oh, okay. Yeah.
So standard plan, I think standard plan still has some ads in there, but you have a lot less ads. I think that's what it is.
I have no ads in my plan. Yeah.
But you know, here's the— yeah, whether you paid for standard or above, same for me.
I don't know what I pay. I just know there's no ads. I've never seen a Netflix ad on my TV.
Me neither. But yeah, so we don't have that. But I think this is an opportunity for marketers because if it's like 30% of their user base or whatever, It's going to be this, this lower, lower plan, then this is more attention for you to go after. Seems like a lot of people are like, oh, CCTV does pretty well. Like, okay, go after Netflix.
Dude, it all adds up from Netflix to Roku. Like, these are all places where you can just get access to more people. You know, funny enough, speaking of ads and, you know, platforms like Netflix, I did an interesting experiment, or my team and I did an interesting experiment. We wanted to see if being on Amazon drove more sales outside of Amazon. So in other words, we want to look at total sales. Do you think being on Amazon hurt or helped sales?
Helped.
All right, so you'll be in a shock. You may be shocked on this one, or you may not, depending on how you look at the data. So we took 18 websites that had minimal seasonality. So these were ecommerce sites that weren't selling on Amazon. And in the first 3 months, guess So if a site had 100% of their revenue come from their website sales, you take the same or similar dollar amount, what do you think happened at the— within months 1 through 3?
103.9%.
Yeah. So you're looking at the charts. So sales did go up, but Amazon took a nice chunk of it and their overall website sales went down. And a lot of the people that would have bought on their website started buying on Amazon. Now, when you look at months 4 to 6, what do you think happened?
Let's look at your chart. It went 114.7%.
Yes. So it went up even more. Amazon sales went up, website sales went up compared to the previous, you know, months 1 through 3, but it still wasn't as high as it originally was. And then when you look at months 7 through 9, it went to 135.8%. So it grew, you know, by 35.8%. And both the website sales grew more than when their experiment originally started and Amazon sales grew as well.
So it takes time.
It just takes time. And then when you look at months 10 to 12, you know, it went down compared to months 7 to 9, but still website sales were still higher than they originally were. And Amazon sales, you know, were gravy. I think a lot of people give Amazon a bad rep for the fees they take, but they forget on how much money Amazon spends on ads and marketing and all the money you save by just being on their platform.
Dude, this tweet, by the way, this is like a complete side note. This is interesting, really interesting data to me at least as a marketer. But this tweet didn't do as well for Neil. It only got 9,000 views, but it's actually a really good tweet. I think the problem with it was probably the hook, right? Does being on Amazon drive more sales outside of Amazon? I think if you had like a really punchy thing, I think it would have done a lot better.
It's too specific. When I do too much B2B or B2C or specifically for people who are selling on Amazon, my content just doesn't do as well. But I still create it because this is what The type of content that we've seen that when we create, it drives the most enterprise leads.
So what I'm going to do, guys, is I'm going to copy and paste exactly this tweet. I'm going to post it on mine. I'm going to change the hook. I'm going to see what happens.
Try it. You'll end up seeing it.
I love it.
I get branding from the charts. Go for it.
Yeah.
And what's funny is we started testing the same thing with other marketplaces, eBay, Walmart, et cetera. Data's inconclusive, but we think we're going to see the same thing, but just on a smaller scale, just because Amazon has so much brand pull.
Yeah, I think it's interesting. It's one, people need to think longer. And then 2, it's like, it's not one or the other. And like the brand of Amazon, there's like a huge brand lift there, right? At the end of the day, it's a massive lift. Yep.
So that's it for this podcast episode. Thank you for listening in. Make sure you rate review this podcast, give us 5 stars. We really appreciate it.
