In this episode, Jimmy, Chase and Brian uncover what most brands get wrong about personalization, the biggest challenges in building a recommendation engine, and explore the importance of human oversight in leveraging AI for marketing, to ensure it serves as a tool to enhance human creativity rather than replace it. They also discuss the future of eCommerce personalization and what’s next for Nacelle.
Welcome to Send It where we discuss anything and everything retention marketing specifically in email and SMS marketing. This is Episode 21 with special guest Brian Anderson, Founder & CEO of Nacelle.
Jimmy Kim: What is up everyone! Welcome to Send It where we discuss anything and everything retention marketing specifically in email and SMS marketing. My name is Jimmy Kim and I'm here with my co-host Chase Diamond. Chase, we're back! Episode 21: AI product recommendations that drive revenue, and we have a very special guest with us today. Chase, you want to do the honors introducing our guest today?
Chase Diamond: I'd love to. As you guys can see, we've got Brian Anderson on the pod live. He is the founder and CEO of a company called Nacelle, and essentially they're focused on the future of commerce. They really believe, as do we, that it lies in creating better online shopping experiences. Their mission is building the latest technological innovations around AI and other types of technology so that marketers, founders, and brands can ensure they're delivering the best kind of personalized, most intuitive, and engaging journey for all consumers. This way subscribers and customers can get the most dynamic experiences, the most personalized content, and that way brands are going to make more money. So Brian, welcome to the show!
Brian Anderson: Yeah, thanks for having me Chase. Thanks for having me Jimmy. I'm excited about this one.
Jimmy: Awesome! Well Brian, before we jump into anything, we still want to hear your background. We want to learn about your journey into ecommerce. Just a short blur, so the listeners want to know: who is Brian? Tell us about yourself.
Brian: It's actually really funny because people ask why I'm an engineer. I went to business school. I went to a college called Babson College, which is in Boston, and Boston's really cold in the winter. My sophomore year they put me in this dorm that was literally as far as you can get from the cafeteria.
It turns out Babson has a sister school called Olin, which is strictly engineering, very small school, very smart people go there. It turned out—I don't know how we figured this out—but I could swipe my cafeteria card into Olin's cafeteria, which was much closer than the Babson cafeteria. So I started to go there, and you know when you first walk in it's like, "Okay, what's this alien doing here? It's a Babson kid, why is he being here?"
But I started to make some friends, and despite going to business school, I ended up taking some credit for engineering classes. The kids I was hanging out with during lunch convinced me, and that kind of started my journey to writing code. So when I graduated, I basically made my early career writing code and working at startups, trying a few that didn't work out as well, and then essentially finding myself landing right in the grips of the Shopify ecosystem just because it was up and coming and I think I saw a ton of value in it early on.
I started an ecommerce agency for several years, and that was all the buildup to Nacelle, which brought us to where we are today. So that's my brief background and basically been doing ecommerce for maybe not as long as you Jimmy and Chase, but pretty darn close—over a decade.
Jimmy: So do you consider yourself an engineer, a marketer, a business guy? What do you think about yourself?
Brian: Well, I think in today's world you have to have all three, that's the truth. My specialties are certainly business, finance, and subsequently engineering, just given that that's what I've been doing for so long. But you can't get away from marketing when you're talking about ecommerce.
Reading the Wall Street Journal about a year ago, they put out this special about what's the most likely path to CEO at a Fortune 500 company, and actually the least likely way was through marketing for every industry except one, which was retail. So your brand, how you speak to your customers, your strategy for acquisition, your strategy for retention—this is what makes or breaks ecommerce. This is the number one thing. So there's no way around it: when you're talking business and you're talking ecommerce, we're talking about marketing.
Chase: Totally agree. I know before this you gave us a pretty cool kind of behind-the-curtains look at what's going on. I know you were actually talking about building some cool stuff that's acquisition focused. I know we're a retention podcast so we're going to focus on retention, but you have some really cool stuff coming. Let's start off with this big question: Why is retention marketing in your opinion more valuable today than ever, and how do you see AI kind of playing a role within it?
Brian: Great question, good way to start off. I think we have to rewind just a little bit to 2017, because 2017 was the start of Apple's change in privacy policy. I know most marketers know this, but Apple actually doesn't have as much market share from a device percentage as Android. The problem is Apple has three to four times on average more buying power. So essentially folks who use Apple devices have significantly more discretionary income, and that makes their policy the de facto standard for all marketers in ecommerce.
In general, Apple has aggressively changed the landscape of what's possible with marketing online and the privacy policy. If you look at who it affects the most, in my opinion, it affects direct-to-consumer brands. It affects SMBs, mid-market, and even the enterprise brands. The people who it doesn't affect are your Amazons, even Netflix of the world. Why is that? Well, with Amazon you have a signed-in experience. More often than not very rarely are you not signed in. When it comes to a direct-to-consumer brand, maybe 2% of your traffic is logged in and identified at any given time, but about 98% is anonymous.
With the advent of the privacy policy changes, especially the big one in 2023, not only can you not identify who a customer is, but if you can somehow get some information from them, even with a first-party cookie—people don't realize this—it wasn't just the third-party cookie that was affected. The first-party cookie window is only seven days. So if they don't come back within seven days, they're anonymous just like everyone else. For contrast, in 2022 and 2021, you could set a policy on one of these cookies for over a year.
This beautiful idea of personalization and marketing has always been one-to-one: I can send a direct message, I know who this person is, I can really hone into their preferences and relate to them in a way that I wouldn't be able to otherwise. Segmentation has been pushed down as something that's not as glamorous. But now with the new privacy policy changes, you can't do one-to-one in the world of acquisition.
In very unique circumstances you know who your customer is at this point—or your shopper rather, they might not be a customer yet. And so that is true for everything on the acquisition side, which means that your retention strategies and tactics become incredibly important and where you can leverage modern tools like AI and one-to-one personalization. Unfortunately, it's just not going to be on the user acquisition side; it's going to be when we talk about retention. So I think a lot of the glory and glamour will be on the retention side of the house, and a lot of the brute force work will be on the acquisition side.
Now I want to be clear, you really need both. I know Send It is obviously about retention, but I'll say something obvious: if you don't have the top of the funnel bringing in customers, you're not going to have retention. So they do play hand in hand. But when it comes to the big changes that happened recently, I'm not sure just how much people realize this has changed the landscape radically and how much it really puts retention on the pedestal.
Jimmy: I mean, it's the yin-yang, right? Like we talk about the acquisition and the retention. One has to be there for the other to occur, and it kind of works together. I know how you like dance around it Brian, but the magic word is iOS 14.5. You throw that terminology out and marketers, anyone who's been around this space, all quiver a little bit. That was the marker.
I always laugh because even today while we're at iOS—what, 18 or 19, whatever we're at now—when people talk about their talks, especially in the acquisition side, they're like, "iOS 14.5, remember that?" And then retention people are like, "Well, iOS 15, remember that when the open rates became less valuable?" Like every change that is occurring—and you're 100% right—is not only driving a different result and a change in behavior, but to me, one of the things I've noticed over the years is it's just weeding out the weak ultimately.
The people who don't have strong businesses weren't able to survive. People who didn't have the best products or services, or they didn't have enough focus on either retention or acquisition, started to struggle. So I do agree, obviously we love retention here. We know the importance of acquisition, but we talk about retention because, well, it's the money maker. It's the true thing. Acquisition drives that first customer, but retention drives the second and third and fourth purchases. And that's where the dollars really come into place.
Brian: I'll add to that too. That second customer that makes that second purchase, that's where you really can crack it open. I think that's a really important point too—if a customer just buys once, maybe that's a slightly different category than the customer that buys twice. I think it's really important to recognize that now more than ever with the changes in privacy policy, everything you just said Jimmy is just something marketers need to be hyper aware of. Because their acquisition might not be as efficient as it was in the past, they're going to have to make it up with really great retention strategies and tactics.
Jimmy: All right, I want to jump into the next part, which you know we were talking about before this pod started. We were talking about your recommendation engine that you're working on right now—your product recommendation engine. And we were joking behind the scenes like product recommendation engines in the past have not been very meaningful. They haven't been very worthwhile.
I'm not here to pick on any products or anything like that, but even at the ESP level, as someone who's been in that level, we weren't doing anything magical. You would take a bestseller or the most frequently purchased thing because we know there's only three, four, maybe five products on a store that make up like 80% of the sale. If you just throw one of those on the top of the recommendation engine, if they haven't bought one of those, they generally do okay.
But the reality is, well, there's more data than ever and more things that are happening. So talk to me about product recommendations being truly meaningful here today.
Brian: Well, one nice thing is you have some data if someone purchased from you, which is okay. So you can start to say, "Well, people who bought this also bought this." I don't think that's particularly novel.
But with the idea that the privacy policies have changed so much, we kind of went into the lab and thought what else can we do and what other kind of data can we get. Engineers call it clickstream, but this is kind of the idea of what happens when a customer is navigating through your brand's store. Is there any data that we can collect there in an acceptable, ethical, and also privacy policy-approved way that actually can add to strategies that help increase AOV, like product recommendations?
Specifically, if your customer has a product in their cart, and very regularly they have this pattern where they're viewing a PDP or adding something else to their cart, or they go to check out but they don't complete—all these things actually have intent. If you're missing that side of the house, that side of the data, you're not really bringing a product to market when it comes to something like product recommendations that's taking advantage of all the surface area that you have.
Now this is not to say that you need to tie identifiable information to that data. I want to make this really clear. All you're doing is saying, "Okay, the people who click on this that have this in their cart often like this product." And that is just such important data. Why wouldn't you want to collect that and then aggregate it and then change your product recommendations based on that data? That was one of the things that we really wanted to design as a core component into our product recommendation system, because we think it's essential. We think it makes much better recommendations at the end of the day.
Chase: Dude, that's awesome. One of the things that we talk a lot about was moving past just purchases. So you kind of answered and talked about what we think about and what we believe in as well. That's awesome.
Chase: I guess the next thing I want to talk about is there's a lot of discussion about the balance between personalization and privacy. It's kind of this fine line that you have to cross, or I guess kind of walk delicately so you don't cross it. Where do you think and how do you think about this balance between the personalization and the privacy components when you're developing features at Nacelle, and how are you focused on making it quote-unquote "not creepy"?
Brian: This is a great question. You put it into the context of the privacy policy maybe for Apple or something like this, which is good and relevant for ecommerce. But if you zoom out and look at the big picture, there's also a bigger consumer trend around this.
One example I'd love to point to is this case study of Target, where Target was able to figure out that a teenager was buying some products in their store that sort of indicated that she was pregnant. What ended up happening is—let's just say it went beyond the creepy line—Target started to send discount coupons to these sorts of products to this gal's home. Dad opens up the Target envelope and sees, "Oh, you're pregnant. Here are some products that align with that." Well, this was pretty bad, right? The dad didn't know that the daughter was pregnant.
But what this case study is really showing is that your job as a marketer is not just to maximize AOV and conversion rate. It's also to long-term establish trust with your most loyal customers or who can potentially be your most loyal customers. So there's sort of this natural balance in the market where, even if you had all the data in the world on someone who just visited your site for the first time—which is an unrealistic scenario—you might not want to use all that because it might create distrust between you and your shoppers and customers.
So I think, just to be clear, it is the privacy policy issues that have changed the landscape. But long term, I know marketers like to hate on Apple and have my own opinions about Apple especially when it comes to their AI strategy or lack thereof, but customers don't want to feel creeped out. That's the bottom line. I think we could all agree with that. So you do have to strike that right balance.
Now I also happen to think that different products that are marketing to different segments probably have different tolerances. So hey, if you're selling to a group of engineers, maybe they might be a little more open to some more aggressive policy. Certainly if you're selling to marketers, they're like, "Bring it on!" But different brands and different brand voices will have to find their own balance for what's too aggressive and what isn't.
That said, there are some things that are fair assumptions that you can make. If you ask a customer for certain data and they give it to you, you've actually created trust. For example, I love the idea of beauty companies asking about quizzes for your skin type or something like that. When you submit your different colors, different tones, and it says, "Okay, you're combination skin or dry skin," that's a great relationship getting established between the customer and the brand because you're helping the customer buy in the process. Maybe they learn something when they take a quiz or something like this. But also you're not doing it in a way that's creepy. I didn't surmise that you have oily skin because I looked at all the other websites you've been to before you got to my store—I just asked you.
And I think that is essentially the way we can do this in the future. I do think segmentation should not be discounted. I think in the past marketers have found segmentation to not be the glorious side of personalization, and I think any modern personalization strategy that doesn't logically think about segmentation is a huge loss.
One problem, and I think certainly the email service providers are the first folks to recognize this, is I don't know if there's been great organization strategies for segments and how you manage your segments and how you should think about your segments in the past. So to me, some low-hanging fruit for this new world with the changes in privacy policy is to really think through who your segments are and what tactics and strategies you want to use to approach those different segments.
It's nice to have some segments for acquisition where you don't necessarily know details of the customer. It's also equally as important in the retention world to understand who your MVP folks are and going after them and make sure that you align the right tactics to the right segments, but really take the time to map this stuff out. If you open up your email service provider and you have like a thousand segments, it's just not practical—you're not going to be able to use that in any way.
Almost the way great marketing organizations have organized UTM parameters and SOPs around how you make the right links and stuff, you need to take your segments just as seriously as things like that that have to be very organized.
Jimmy: You know what drives me crazy, Brian? As many segments as you see these marketers create, they don't use any of them. That's the worst part of it all. They go off and create all these segments and they go, "Okay I've got 58 customers here, 48 there, 4,000 here." And then they go off and they just batch together 90-day customers and they just blast them as they mail. They do all that work and they do nothing special.
I'll ask them why they do this, and oftentimes it's, "Oh well, it's for reporting." And I'm like, "Well, what's the point of building these things if you're using it for reporting and not actually going to use the data and segment?" Or they go the other way and they go gung-ho crazy and they got 20 emails going out in a day overlapping everything, and then they wonder why customers are upset that they're getting seven emails because things are overlapping.
I'm with you on that with the segmentation side of things. With great power comes great responsibility, and sometimes they got a lot of power and not really sure what to do with it, and they ultimately end up causing more headaches than anything else.
Jimmy: In fact, let's roll right into that. I want to talk about this because when you say that, we know that retailers are also doing the wrong thing when it comes to personalization outside of just being creepy. Dude, what have you seen in this market?
Brian: The funny thing about personalization is that it's so wide. Everyone also has a very different definition of it. Even for some of our marketers that we are fortunate enough to be working with—some great brands and some great logos—when we talk to them about what personalization is, they're like, "Well, we're releasing a new women's line so we have like men and women, or we have like buyer and customer." And I'm like, "Okay, cool starting point." But there's so much more to get into.
It does relate to the point you just made too, Jimmy, about organization. I think some people, especially some marketers that are trying to do the right thing, think it's the right thing to create all these different segments or all these different strategies and tactics. Just like anything else in marketing, you need to have the big picture plan and then you need to follow that plan.
The same way you shouldn't really buy a SaaS offering or an app because your biggest competitor buys it. That's kind of a well-known thing, but it actually should fit your whole marketing plan, and you really need to be thinking about it from the top down.
So I would say the number one mistake people make when it comes to personalization is they're just following what they think the right thing to do is because they heard through the grapevine it was a best practice. But whenever you're doing something because you think it's the right thing to do, just ask yourself: how does this fit into the bigger picture of what we're trying to do for our retention marketing? And if you can't answer that question clearly, just take a step back. It doesn't mean you have to stop what you're doing. It just means slow down, take a step back, grab a coffee, and think through how this fits into the bigger strategy first. And that's the number one thing—even if you don't use technology that's fancy or anything like that, if you just do that, I think you're going to get some great wins.
So think about it and make sure it fits your wider strategy. That's number one.
Number two is understand the difference between segmentation and one-to-one marketing. I think a lot of vendors out there have done a terrific job conflating these two, and unfortunately the folks that suffer are the DTOC brands that are SMB or mid-market. Because it's very easy to confuse a group of people by using one-to-one strategies where you don't have data to use the one-to-one strategy.
I think educating yourself on when it's appropriate to use one-to-one personalization versus segmentation strategies, and just having a boundary between those two things in your mind's eye—if you do that and you pair it with the wider marketing picture, that right there is going to be some low-hanging fruit. You're going to get some wins right off the bat.
Jimmy: Chase and I, we talk about this all the time, just about this exact thing when we talk about segmentation. It's too many people. There's the dream of course. Like I would love if I could send every customer the perfect email each and every time, but the reality is that's just literally not possible. You've got to use the data that you have, the segmentation you can, to generalize the batch of people, and getting closer to that right answer is all you can do. So I'm with you on that on personalization.
Brian: I'll also add, back to the idea of what's creepy and what isn't: the more they buy from you, the more trust you're establishing, maybe the deeper you can go. If you think about it, top of the funnel all the way down—the lower down the funnel you get, even post-purchase, the more you can lean in. I would say a general rule, this is a rule that can be broken, but a general rule is the further you go, the more one-to-one you can get without acting creepy.
Chase: Love that. Yeah, totally agree. One other thing that Jim and I were talking about that we think is really interesting about what you guys are doing at Nacelle is your guys' focus on metadata and kind of product context. Can you walk us through why this matters so much for effective recommendations?
Brian: Look, I think context is always king, especially in the world of AI. Your system, no matter if it's AI-driven or traditional—one-zero driven, if you will—you're going to have to pair the right product and the right content to the context that's being displayed. So it's not just the customer, but it's the context and the way that content and the product is displayed.
For example, you might not want to show a product if someone just purchased that product on a landing page. It's a total waste of real estate. But in the world of retention, we get this incredible insight into, "Oh, we know in the past 60-90 days that this customer purchased said product." So not only should we use that data and insight to our advantage, but we should also create content that pairs with it.
So you really can't execute a strategy like that where you're pairing both. You're putting together the magic three: the context, the content, and the product. You really can't pair those three together well if the data isn't set up correctly.
Jimmy: That makes a lot of sense actually. When you think about it, people always want to do all three, but it's hard. It's really hard because the data is not there. And historically it's not something that you can pull out if that makes sense. Like it's been kind of trapped and locked behind different places, and it seems like you guys are really thinking about all sides of this whole thing when it comes to recommendations.
Jimmy: In fact, one of the areas that I'd love to discuss is in the beginning of this show, we talked about the manual control side of things. Because well, marketers like to have recommendations, but they all want to do their own thing. Chase and I will talk about things, and it'll be like, "Well, the data says this," and it'll be like, "But we think this is better idea." So what led to your decision when you were going through this? Is this just from interviews or from historical information?
Brian: I think there's two areas in our product where we thought about this question quite a bit.
The first is when you first come into the Nacelle product, you're greeted by our AI technology called Paige, and that's a persona that we created. That's essentially our LLM. One of the first things she needs to do is have a conversation with you to understand your brand voice. So she'll go through and she'll start to look at your product catalog, look at orders, look at your store, but through conversation she's also able to learn your preferences.
Why do we do that? Well, first of all, I think with our headless product, one thing we learned is if it were just about numbers and data, I think I would have already been retired and then some. But it's not that. The brand is an alive being and it needs to be represented properly, and it needs the human touch. And this is coming from someone who's a huge AI advocate.
It was very important for us, if the AI was going to work correctly, that it had to understand the nuances of what the human was thinking and what their opinion was.
Now, the other area where this really comes into the mix is when it comes to product recommendations. I remember years ago when I ran a Shopify agency, a lot of our clients at the time were installing product recommendation applications, which were relatively new category back then. To my surprise, they were certainly spending a lot of money on these product recommendation algorithms, but to my surprise, they all would like to override what the recommendations were. And you know you are going quote-unquote "against the data."
When I ask, "Well, why are you doing that?" They're like, "Well, maybe it affects the brand experience" or "Maybe there's some distrust there." That's actually valid. I think someone who is just a data scientist would take the other side of that argument. But to me, I think the AI should be following or a tool for the human to use. And the minute the human is a tool for the AI to use, we have bigger problems in the world.
I think it's really important, and what this comes down to is understanding the brands and understanding how to interpret the data. Just because a report says one thing, going one step deeper and understanding why that report is saying that thing, especially if there's an outlier or something interesting there, it actually is important. And so this is sort of the nuance of being a marketer in 2025. You have to balance these two things in a way that probably is still uniquely human.
As an engineer and a data guy myself, I probably was pretty flabbergasted by this—I was—but I think as time went on and I learned about what makes a brand a great brand, I realized that it's the human that makes the decisions behind the brand, and the data is used as an input but not the only input.
Jimmy: That makes total sense. With AI, one of the things Chase and I always talk about is like, yeah, there's a lot of things that it can do and a lot of data it can ingest, but the reality is the big power of AI today is that it's 10x-ing the marketer ultimately. It's supercharging us where we can do more than we could ever do, get more output than we could do, and you can build and create things that you may have not been able to do in the past without a lot of time.
There is that part of it, but then on the other side of it, it's not always right either. I don't know if people know this, but AI will make up stuff if it doesn't know, just kind of like a human would. When people understand that, they're like, "Wait a minute, AI could just make up answers and think that this is right?" And the truth is, it can. So it's good for a human to kind of sit in the middle and still control it and still have some narrative behind it.
So I totally get that, and the human intuition is still one of the strongest things. The gut feelings of a human are very important in a marketing world still today. So I agree with you.
Brian: I agree with that. If I can add a comment to that point, Jimmy, in our engineering team we use a ton of AI. Actually, one of the reasons why we're able to offer our products at a competitive rate is because our cost of doing business is lower than most of our competitors because of the way we're able to leverage AI. There's some secret sauce there that none of your listeners would ever care about.
But I think what's really important is it's not that AI was replacing the engineers. It's that the engineers became the architects or the directors and had the AI do the busy work. By the way, writing code, the actual process of writing code, is not particularly sexy. Coming up with the architecture and the strategy of how this thing is going to work, that's the fun part for engineers.
And when I talk to marketers, that's the fun part for marketers. The only difference is they're designing or architecting the strategy and the go-to-market. The idea of actually making all the content for like 20 different segments or 30 different segments, this is miserable—who wants to do this? No one wants to do this. That's a great job for the AI.
So I think the human should act as the director. The famous Steve Jobs quote is, "A computer should be a bicycle for your mind" to multiply the thoughts that you have. I think that is the right dynamic and relationship in today's world between human and AI. Now, invite me back in five years, we'll start to have a different conversation, but I think for now this is a really important factor.
In general, the world still confuses AI and automation, and they're two different things. It's not to say these advancements of agentic models and stuff can't get us closer to automation, but at the end of the day, it is still not just human in the loop but human as the director. And it's essential for marketers to really understand that in 2025.
Jimmy: It's just like the biggest buzzword right now, Brian—AI agent. That's the biggest buzzword that's hitting right now.
Brian: I don't know what that means.
Jimmy: Do you know what that means? I do! It's just a very advanced workflow. That's it. It's like a very advanced Zapier workflow connecting multiple apps together. And when I explain that to people, they're like, "Wait a minute, it's not like things that do it by itself." And I'm like, "No, man, all you're doing is telling one workflow to talk to another app, get the data out, pull it to another app." And they're like, "But that's what Zapier does." It's like, "That's why Zapier and Make and these guys are the AI agent leading platforms right now today."
I think what people wish for is what is called AGI, which is the change that's supposed to happen this year. So I'm deep in it with you, Brian. I don't know what I'm doing half the time. I don't know how to write a lick of code, but you can ask Chase, I code all day long trying to understand how to write for AI because I'm so interested in it. I think it is the future.
Brian: I mean, no, there's no doubt, and you actually bring up a good point there. I think this is particularly important for your listeners. There is a coming arms race in the world of marketing. The people that figure out the right way to use AI, they're not just using AI for the sake of using AI, but rather using AI in a way that 10x-es their output. Not to sound too cheesy, but this is true.
If you can figure out what those strategies are, it's not just going to ChatGPT and writing in some prompt that someone gave you, and then all of a sudden this is done. It's understanding how you can compete to get a leg up in the marketplace. And I think the people that embrace that and spend the time to understand it are going to have a huge leg up in the market.
I'll just remind everyone that content is king. The context that you're feeding the AI, the content that you're feeding the AI, the information that it understands about your brand, about your product line, about your competitors—this is not just a copy-paste "put a prompt in" that you read on LinkedIn or something. This is quite literally a whole mechanism you have to build to get the right content into these AI systems to get the right output.
And I think as marketers start to learn and understand that more and more, it's going to be a huge boon to their efforts and their work and their leverage in the marketplace.
Jimmy: Yep, it makes sense for sure.
Chase: By the way, we're big fans of Nacelle. We're pumped to be working with you guys. We're singing all your praises. When you think about the successful folks that are implementing Nacelle, what patterns are you seeing and how these merchants and these agencies and these types of folks are actually using the platform?
Brian: Just full disclosure, this is a new product line that we are releasing. We have other products where we work with great brands, and so I just don't want any confusion out there. This personalization suite we're talking about is quite novel.
That said, I think what's been very interesting is different brands and different agencies have different approaches to how they solve problems. I think this is okay, not the most novel statement, but like this is true and this is actually really important. Because if you can inject that thought process or your unique way of doing things into the tools that you're using, then you will have better what I call alignment between the humans and the AI, and you're going to get much better results.
I like to think of when I work with AI personally, I like to think of it as like an incredibly smart but somewhat naive intern that is just way too smart for their years, but also tends to make weird mistakes. Nine out of ten times the mistakes are because there's lack of alignment. That could be because the instructions you gave that intern were a little vague, and you didn't realize it, but they were actually a little vague. It could be because the AI just doesn't have the same context that you have.
For example, one of my favorite things to put at the end of a conversation or a prompt with an AI is, "Hey, is anything in what I just said vague or ambiguous?" Before you do the work, let me know what's confusing here, and I'll answer these questions and let's clear the air before we get into the work. Because I want to make sure the context I just fed you is logical and makes sense.
You'll be very surprised at how often it comes back with, "Okay, well, can you explain this, this, and this to me," and then you do that and the outcomes become significantly better.
Jimmy: I actually think of it like humans, right? Like when you're at an office or you're working together, and one of the problems that happens is communication's not great. There's not enough detail. There's not an explanation. This is literally the same thing that is AI. It's like if you don't do a good job explaining and communicating to your coworker or your AI, and you don't give it all the details, it's going to do what it heard ultimately.
It's the same thing that's happening. I use that as an analogy often to people when I tell them how to prompt. I said, well, prompting, it's the same thing as like talking to a human. If you've been told that you suck at communication, well, there's a chance you're not going to be very good with working with AI because you're not good at communicating as a person to another person. No difference when you go to AI. In fact, AI is just taking things so literal to the T. So you've got to even be more careful than ever.
Brian: One of the best ways to improve your prompts with AI is to send it to a colleague and ask them if they know what you're talking about.
Jimmy: That's a good one. That's a good one.
Brian: You'd be surprised because as humans, especially when we're working with machines, we just make all these assumptions because we've been thinking about a problem for so long. As an engineer, I can't tell you how often I've had conversations with people outside the engineering department, and they're like, "Dude, what are you talking about?" It's like, "Oh right, I've been thinking about this problem for so long, I'm just leaving them in the dust."
Unfortunately, AI can't read your brain quite yet, and I think we're still far away from that. So you really have to do a good job of taking the time to explain what's happening and what the context is. I think that makes all the difference in terms of the output of these tools.
Chase: That's cool. Well, Brian, what excites you most about the future of ecommerce personalization? Obviously Nacelle is very exciting, but what's the future of ecommerce personalization to you and how are you thinking about this?
Brian: I think personalization and data go hand in hand. I just want to make sure that's super well established. I think we're hitting the nail on the head on that one. It's not just the more data you have, but the right data you have, and the way that you use data in a way that establishes trust between you and your customer is essential. So that's principle number one.
And then I would say if you can do that well, then the idea of actually building full shopping experiences that are built for the data that you have given the customer you have, that's the goal here at Nacelle. That's exactly what we want to do.
I think with privacy concerns, that's certainly a hit that most marketers are taking now. You're gonna have to offset that with some really clever ways to do personalization that feels natural, normal, and comfortable to your shoppers and your customers.
I want to be clear, it's not just product. I think content is king. If you can pair the right content with the right product and the right data, that's the trifecta, and that's what every marketer should be striving to do. So I think the future of personalization online is going to change the shopping experience.
Now, I feel like every couple years there's "Oh, TikTok shops or Facebook shops is going to destroy the ecommerce website." I think that's unlikely. I think if anything, it's another channel. It's supplemental, and that's okay.
But what does personalization look like when you're shopping on an ecommerce website of the future? First of all, if you're coming from an email or maybe an SMS, there's already a lot of information we know about you. If I come from an email and I land on a page that's the same page that everyone else lands on, I'm telling you right now you're leaving money on the table there. That just should not be acceptable.
Your conversion rate for when your customers land on that first page that you're linking to from your email or your SMS—if that's not performing ultra well, then that is such a great spot to start to think about where can we add some personalization tactics and techniques to really start to drive some of our core metrics. So I think that's going to be the starting point. And then I think from there, building the site as the customer shops is going to be what comes next.
Jimmy: Dude, this has been awesome. I've got one last question to kind of round up this podcast. What's next for Nacelle? We talked a lot about what you guys are doing today. What does the future look like? And again, without revealing too much and giving away any trade secrets, what's coming down the pipeline?
Brian: I love natural conversation. I love podcasts. I listen to podcasts. I listen to audio books. I think audio and communication, especially since we went through the whole COVID crisis and workers still continue to work remote, that's been very normalized. There's a blend with that and what we're talking about right now at Nacelle.
Today we have our product recommendation system going out, and very soon we have a landing page builder going out that you folks will see first, which I'm very excited about. One of the cool features that we have is if you want to hook up your Zoom or your Google Meet to Nacelle, to Paige, which is the AI persona we have, she will learn what's important for your objectives, what's important for a given campaign that you're working on, and start to craft both campaigns and landing pages and different product recommendations based on the conversations that you're having with your team today.
So again, you could think of it almost like another very smart intern that's sitting with you, sort of taking notes but also preemptively starting to create ideas for you, using best practices for what is the thing that we should go and build. In other words, the shift is not just analytics and looking at a report. It's not just having a meeting and then sending out to-dos. It's actually the direct action that can come from conversations.
Jimmy: Makes total sense. Very cool, man.
Jimmy: Well, Brian, thank you for being on here. We've gone through a lot of stuff here, and well, there's a lot of exciting things going on. And of course, thank you for the sponsorship too. For those that haven't heard the commercials or seen the logo or anything else, just to be very clear, Nacelle is also a sponsor of this podcast, which has been awesome and amazing too from your side.
With that guys, that's a wrap, Episode 21: AI product recommendations that drive revenue. And this episode was all about how Nacelle's Brian Anderson is transforming ecommerce personalization.
If you enjoyed this podcast today, let's make sure you subscribe, follow, leave us a rating on your favorite podcast listening platform. You can follow the show on X, LinkedIn, YouTube, IG, TikTok, etc. And there's a few ways to connect with us.
First things first, if you don't know, we have a newsletter. Obviously we drop those newsletters three times a week right now, and you can find that at ecommarketer.com.
Secondly, you can follow us on social. Hey, Brian, what's your social? Do you have social media? I think you're on LinkedIn, right?
Brian: Yeah, LinkedIn is my primary channel, and you could find me on there. It's Brian V Anderson.
Jimmy: There you go. If you want to talk more to Brian about Nacelle or personalization, reach out to him. Go to his website. There's a lot of ways to connect with him, but we definitely recommend it.
And then, of course, Chase, where do we find you on social?
Chase: Most platforms, Chase Diamond.
Jimmy: There you go. And then, of course, myself, I'm found mostly on X and LinkedIn here today.
All right, next week, Chase, we're back! Next week we're kind of pivoting a little bit away from this stuff, and we're jumping to data hygiene. We're going to be talking about maintaining a clean, high-converting list. As you know, we just did a big migration, and we figured we'd talk about it as well.
So with that being said, see you next week! Thanks for tuning in. Cheers!