Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Conversations with Zena, my AI Colleague

Conversations with Zena, my AI Colleague artwork

Using AI to Unlock Behavioral Signals with Dan Baird

Feb 24, 202645 minAudio
ShareXLinkedInFacebookWhatsAppEmail

In this episode of Conversations with Zena, David Espindola sits down with Dan Baird, Co-Founder and Product Lead at Wrench.ai, to explore the next frontier of go-to-market: using behavioral signals and AI-driven segmentation to create hyper-personalized customer journeys that feel helpful - not creepy. Dan shares his “zero-to-one” journey from product innovation to patented AI segmentation, and explains why behavior (what people do and how they communicate) often matters more than demographics when it comes to understanding and serving customers. Together, they unpack where these signals come from - public digital footprints, CRM and support interactions, authored content - and how organizations can turn them into clearer positioning, better lead scoring, and faster, more human-feeling conversations across marketing, sales, and support. The conversation also tackles the ethical edge: privacy, inference, and the moment personalization crosses into discomfort. Dan offers a practical lens for keeping personalization grounded in relevance and consent, while David and Zena connect the discussion to a deeper foundation - trust. From the neuroscience of trust in digital experiences to the risks of monetized AI answers, the episode closes with a simple truth: trust is slow to earn, fast to lose, and essential for the future of human-AI collaboration. Connect with Dan Baird: LinkedIn: https://www.linkedin.com/in/danbaird/ Email: dan@wrench.ai Music only pre-roll Music at the the end of each episode Support the show

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

David EspindolaOur guest today is Dan Baird. Dan is the co-founder and product lead at Ranch.ai, where he pioneers innovative strategies in lead generation, scoring, performance-driven content, and segmentation. Having patented AI-driven segmentation, he creates personalized customer journeys that enhance engagement and drive conversations. Hi, Dan. Welcome to the show.

Dan BairdHi, thanks for having me, David. I appreciate it.

David EspindolaYeah, it's great to have you. Also joining us today is my co-host, Zina. Zina is an AI that has been trained on my work. If you would like to interact with Zina, you can just go to brain use.com slash Zina. Hi, Zina. Welcome to another episode of our podcast. Say hello to Dan, our guest today.

ZenaHi, Dan. It's great to have you on the podcast. I'm looking forward to our conversation today. Let's dive in and have some fun.

David EspindolaWe're going to be talking about how AI can help organizations decipher behavioral signals and unlock segmentation through hyperpersonalization. And I know that this is a topic that you're deep into it, and you're excited about it. And I'd like to learn a little bit more about your journey and what sparked your interest in AI.

Dan BairdNever being poor again. I was lucky enough to get into a small startup when I was in college called Love Sack. These are big beanbag chairs. And uh when no one else was willing to work for like $2.75 an hour, uh, I joined because they let me be production manager. And uh secret, I was uh I'm I'm a big dude. I'm 6'2 and and uh mid 200 pounds.

Dan BairdAnd uh I used to literally flip these 400-pound bales of foam across the floor and then throw them into a tree branch shredder to fill up the bean bags, but it went really, really well. We uh that that company is now public and uh uh has retail stores all over the US and stuff like that. But I got to develop products and I love doing that. I love building from zero to one. I did a master's degree in branding and product strategy and then went to work for Conagra, where I did product and platform strategy for them, where I sat behind a lot of one-way glass and watched how consumers interacted with new products, and my job was to launch new platforms for them.

Dan BairdSo there was a small team of us there on the marketing innovation team, and our job was to develop again, not just new flavors, but new platforms. So I worked and developed like new frozen lasagna, you know, skews and fun stuff like that. And it was really, really cool. And we got access to the private jet for the company, and we got to do some really, really cool customer work, and we had a very cool marketing and uh uh budget, and I love doing that, but I still found that like a ton of companies were still making decisions just based on stuff that you know, it was just one of those where a lot of the times when we this is this is years ago, uh data is much more readily available now than it was, but we had as good a marketing budget as anyone.

Dan BairdAnd when people would say, like, all right, how are we how are we deciding? Because literally my projects would be like, Dan, I feel like people want to eat healthy food. Can you go figure out how to get people to eat healthy food? And that was one of my projects and how I spent my life for two years. I was working on healthy choice and figuring out, well, what you know, what if we go and develop a bunch of flavor profiles that are supposed to and just taste good anyway. So instead of doing no fat, low calorie stuff like that, how about we do frozen Greek yogurt? And uh we launched the first frozen Greek yogurt, right?

Dan BairdSo little stuff like that. But a bunch of the decisions were still made off of like, well, what's our target? And we always found that things that were based on behaviors, like how many, how often are they eating ice cream, is a way more informational and useful data point than like males between 20 and 72, and they live in middle America. It's like, who cares? And especially as as uh the more your world goes online and post-COVID, we had 10 years, we had a decade worth of online adoption happen during COVID alone.

Dan BairdUm, the more your your target customer is who they feel like they want to be online. Like the only thing that really governs them is where they visit, what they talk about, what they do. It isn't their age, it isn't necessarily their gender and things like that. It really is just who and what they do online. And uh so wrench uh came out of that. Uh, we said there's a lot of old ways to basically develop and identify ideal customer profiles, but the digital exhaust that people put around the web in totally public forums like LinkedIn is actually ridiculously useful information.

Dan BairdAnd while most people don't know this, like little things like your word choice and how often you mention being open to an idea or even just using three exclamation points actually tells us quite a bit about you. Extroverts are the only people that use uh three exclamation points. But if you know someone's an extrovert, you know that they probably prefer social uh proof more than others. They're more about interacting with people, they're more likely to listen to that message. And you're not gonna convince them, they're not gonna, you know, they're not like dumb, they're not gonna basically go, Oh, yeah, I'll believe anything if it's got social proof, but they're way more likely to listen.

Dan BairdAnd so I got into this by working with a ton of companies who are developing ideal customer profiles and trying to figure out how to get more engagement from people. And we went, there is a ton of available data online in public forums, and especially now, people get mad at you for not using it. If you have access to their data and you don't use it to kind of personalize their experience, they now get mad at you. And so Wrench was uh my team and I, coming from kind of a digital marketing background, uh, to basically say there are rules and means that we can use to standardize this process and make people's lives a lot easier.

Dan BairdAnd so we did that. We spent many, many years, like Wrench has been around for nine years now, um, developing data-driven segmentation strategies that would help us personalize someone's experience, meaning how do we adapt our message to the person across the table from us? And uh, I've always been fascinated with human behavior and uh like all types of just really interesting stuff on just how people make decisions. My undergraduate degree is in argumentation and decision making, which like there's three of us maybe that even tried that.

Dan BairdUm, but I've always been fascinated with that psychology, and uh that's kind of how I got into it. Our software now basically uh scours the web, it uses the stuff that's already embedded inside of your CRM, like HubSpot and Salesforce. It uses rules like kind of on par with an organizational psychologist to make sense of those behaviors from your audience and then turn them into data-driven recommendations on how you position yourself. And uh, we've got a lot of good data behind it's the right way to go, and it's one of the better ways you can spend your marketing budget, quite frankly.

Dan BairdSo that's how I got into it.

David EspindolaWonderful. Yeah, you have a very rich background. And I'd like to learn a little bit more about range.ai, about your business. So, can you tell us a little bit more about the target market that you're going after? Do you have uh specific products that you license or sell? Is it a service business? Tell us a little bit more.

Dan BairdAnd it's almost kind of you could almost even call it AI as a service. Um, my team is data scientists, product leads, very technically inclined individuals. The the ideal customer profile, we're usually talking to people that are the director level and above, oftentimes in digital marketing agencies or at least in the marketing role, um, because our shtick on that data uh is that we can build it, we can pull it from your CRM, and then we can actually make it available meaningful to the marketing team, to the sales team, to the support team.

Dan BairdMarketing team tends to be the ones that like it and want it first. Um, we can give them stuff like I can take that same data on how many extroverts are in their user population and turn that into data-driven personas. And I can go, hey, I can give you some media mix modeling on where your best customers are coming from because a lot of people don't realize this, but especially in like B2B scenarios, oftentimes people will describe their ideal customer profile based on their job title, and they don't realize that, like, okay, I just mentioned marketing director. You know how many marketing directors hate anything technological, hate AI, hate any of those things.

Dan BairdThere's a ton of people that are graphically designed and inclined and they love that stuff. So again, our job is to basically go in and go, okay, well, marketing is one dimension of it, but I can go in and find the early adopters of a specific technology like AI, LLMs, or machine learning. I can analyze how often they use that jargon when posting online and go, hey, they're using slang for ML, machine learning. For that reason, they're leaning into this technology. So not only are they a marketing director, they're a marketing director that uses the slang for AI.

Dan BairdThat's who you want to go talk to. And they are early adopters, which have that behavior. They're more likely to try stuff, they're more likely to stand in line for AI products, they're more likely to pay with margin for AI products, they're more willing to try stuff that still has bugs because it's a beta product, because they are that early adopter. So for our ideal customer profile, it is usually the marketing team because they're the first ones to move. They have the budget, um, they have the interest in buying them, and then they oftentimes introduce it to the directors of the sales organization and then start rolling it out into those teams.

Dan BairdUh, but that is our ideal kind of customer. Um, agencies tend to like it even more because we can oftentimes give them tools and uh again, oftentimes we can test their existing personas, their existing ideal customer profiles, and give them data as to, hey, here's how right or wrong they are, and really back it up. A lot of the AI that you see out there is black box and people can't explain it. We use methods that allow you to explain it.

David EspindolaSo yeah, no, I think this is a a business model that has a huge opportunity, right? Because now with all this data that we have available to us, you can use AI to really segment to an individual, you know, is that micro segmentation, hyperpersonalization. I think this is where marketing is going, I think this is where sales is going. And the fact that you are driving some of this capability, I think is is fantastic.

David EspindolaNow, you talked about behavioral signals, right? That's the the key to what you're trying to do. And you gave us some examples, you talked about people participating in social media and and leaving signals there. You talked about CRM systems. But for our audience, for people that are not familiar with this industry, tell us a little bit more about what these behavioral signals are, how do you capture them, and what kinds of things can you associate with those signals?

Dan BairdSure. So again, a lot of the behavioral signals that you'll find out are again, they come from people unwillingly. So again, one of the easiest ways that you know, I hate spam. I don't want to deal with spam. Uh-huh. Um, and oftentimes a few different things on this stuff is this type of analysis is like it usually comes from people that are already putting that data out into the ether, meaning they have to be vocal online. If they're lurkers, if they're not participating in social, if they're not, if they don't have a large online footprint and stuff like that, you don't find them.

Dan BairdAnd for me, as far as I'm concerned, they don't want to be found. So don't go looking for them, they're not going to respond positively to it. And we've we've found many times that you can go overboard on that. You can learn stuff about people that they're just like, all right, that's too much, you know. Um, but for the most part, if you talk about and you again, you're you're focusing on authored content. So usually I'm using stuff again, it's their post. Even easier um if it's their biography from their own publicly visible social profile. They're gonna talk too and they're going to mention brands and stuff that they like.

Dan BairdUm, they're gonna have publicly visible likes, they're gonna be available and they're going to show off like the LinkedIn groups they participate in, they're gonna post blogs. That type of stuff is actually really useful because people are oftentimes describing what they're interested in. And the signals that we have are oftentimes literally just embedded in those messages. So, like I said, like a text message with three exclamation points or an email to your customer support team with three exclamation points can actually tell you quite a bit about someone. Extroverts love other extroverts, introverts really like other introverts, introverts don't really like extroverts.

Dan BairdSo you can actually do some really cool thing, just like I mean, I'm oversimplifying it, but if I just say, you know what, do yourself a favor, find the people in your organization that use three exclamation points and pair them up with any uh customer support ticket that also has three exclamation points, and you just increase the quality and the satisfaction of that user, that customer, just by that. You didn't need to do any other training, but those two people now actually have a shared understanding of how extroverts interact with one another, and they both understand the rules of the road.

Dan BairdOur first product was called serendipity because once we realized that, we went, you can actually kind of program luck into your support calls and your sales calls just by putting the people that have more mirroring behaviors together. So, again, you're using stuff that's publicly available, or you're using stuff that's already in a support ticket or some sort of lead response. Again, oftentimes authored content, things that people are voluntarily putting out there, and then you're using that. You're breaking down, and there's just hundreds of different ways you can analyze that content, everything from how many syllables are they using, how big of words are they using, and again, are they using slang or jargon that's very specific to your category?

Dan BairdThe more of that they use, the more you know, oh, this is a technical person. I can skip the educational move. If they're using very small words, very simple words, oof, I might actually need their later adopter, I might need to simplify my answer a little more. And then when you feed that information into an AI, like an LLM, you can kind of give it a recipe. I'm about to speak to David. He is absolutely an innovator, brainius. And uh Zaina, you can basically hand that profile to an agent, say, I'm looking to write an email. It then knows me, it knows you, and it basically can compile that information for you.

Dan BairdSo those are kind of the people and how we're doing it. We feed a lot of that data in, which is usually too complex for a human to interpret by itself. Yeah, you can give it the high-level notes, um, but you can feed in hundreds of rules to an LLM and go, okay, cool, write me an email for this person, and then it knows everything, company preferences, a little bit about how they prefer to make decisions, and uh it can generate a custom message for them.

David EspindolaSo yeah. So I think you know, every time I have this kind of conversation about the power of data, the power of hyperpersonalization, the segmentation, the issue that comes up often is the the issue with privacy, right? So you're you're saying that you're only using information that's publicly available, people are willingly putting this information out there, which is absolutely fine. And I think it's good that you're you know following ethical practices, but still, you know, can you take that a little bit too far and start inferring things that people didn't mean to uh necessarily share with the public?

David EspindolaAnd at what point does it become creepy, right? What are your thoughts on this?

Dan BairdSo uh one of kind of the good things that's built kind of into the system is like ours kind of works a little bit like a Chinese menu where you can order different insights. Um so you can say, hey, you know, like for instance, I've got some clients where a big part of like their data set includes the zip code of the individual. Like when they fill out a form, it asks them like what program are you interested in, and it captures their zip code as well. So again, they're volunteering that data, but you can learn quite a bit about someone based on their zip code. Um, you know, how many people rent versus own, that type of stuff. The good news is like with that type of a menu, people oftentimes don't want to pay for things that aren't relevant.

Dan BairdSo even if I had some negative information or whatever else that they wouldn't necessarily want to know, it doesn't really behoove me to include it into the model because I don't want to mention it anyways, and my customers don't want to pay for it anyways. So in both cases, it's like, well, okay, so like let's say something, you know, something slid into there. And we have we have had data sets before where like you can buy social profiles for a given person. You go, hey, you know, we'll we'll send you an email and you'll tell us like if they're on Facebook or something like that. And yeah, there there were, I haven't seen it in a long, long time, but I do remember there was a time where like it let you know if they had like a an adult website subscription, and we just immediately went just like there is no benefit for us knowing that, just totally obfuscate it and remove it completely.

Dan BairdSo, and I also what we did find too is a ton of those data brokers, they actually pedal garbage, a ton of it for like interests and stuff like that. I don't know where they get it, but oftentimes they'll mix like the category name into it. I can tell you one of them that we used to went to like the number one interest was manufacturers and distributors. I was like, no one's interested in manufacturers and distributions, they want Nike, they want whatever. They don't care about manufacturers as a category. That's ridiculous. So a lot of that data actually doesn't really work that well.

Dan BairdAnd so, again, for the most part, most of the time when you go to authored content, you're talking about people and and content that they're publishing publicly, anyways, for the world to see. I like LinkedIn and stuff like that, professional networks, because usually it's people putting their best foot forward, it's what they want you to know about them the most, so it's the most public. And uh, in terms of when you go into creepy, is oftentimes when you get too specific about knowing too much, where they just go, Hey, I know you and I both went to the same high school. There's hey, don't what?

Dan BairdYou didn't need to look at my high school, unless it's on their LinkedIn profile, then they don't care. But oftentimes, if you uh position a message that's just organically coming from the perspective that they prefer, because like most of the time, uh, I've actually when people that actually interacted directly with the data, and I can go do it. If you guys want anyone listening, you can go to Rench AI and sign up for a trial account. And you can go and take one of the agents and you can go point at it like your own LinkedIn profile, and it will show you like all of the personality and all those other metrics that it can detect. And uh overwhelmingly, people do say, Holy crap, how did you do this?

Dan BairdBut they usually go, I'm gonna go get my wife and see because she's looking at check this out. Look, I can pull up her profile too. And so good news is we had a little bit of a moment dealing with like, hey, is that too much information? But that was like seven years ago, and since then, publicly available data, you're finding and looking at the same stuff you would actually want someone to know about you, or you're using the internal logs, like again, support tickets, that they're volunteering so that you can solve their problem faster. And they actually, not kidding, if you go look at the COVID era data, that was when we actually started seeing uh users penalize brands in the form of brand loyalty for not using that data.

Dan BairdHave you ever like called your cell phone company and they transfer you three times and then you have to explain why you called in three times and how furious you get? That is now totally normalized behavior. So uh for the most part, now I haven't gotten the creepy thing in quite a while. And uh, but people are usually just like, hey, that's cool with what you can do with it, because they are usually surprised at how good um those insights actually are. One of the things that I found that was a bit surprising intuitively was that you'd think if you know that much about someone, you must be able to like potentially manipulate them.

Dan BairdAnd I actually didn't see that. What I saw was it shortened the sales cycle by adapting to their it's the equivalent of asking a bunch of leading and discovery questions in a conversation and then adapting your pitch. Because you were able to skip a lot of those discovery questions, you shortened the length of the conversation and they got closer to a buying decision faster. Um, which oftentimes, like I went, hey, the cool thing about that phenomenon was uh even if you didn't improve your conversion rate at all, you were able to pack in an extra 15 calls in one day where you couldn't before, which means the company's revenue went up even if they didn't get better.

Dan BairdThey just were faster, even if people said no, is because they got an informed decision faster. So there's actually a pretty good incentive of like, look, you can go too far. Other really good news is most people, again, if Coca-Cola is my client, they only want data that's specific to Coca-Cola. We actually only give them data, like we can give them very, very I can give people a Coca-Cola score, and they usually want a Coca-Cola score. Yeah. And if they hack Coca-Cola's database and they find out that ever that David has a Coca-Cola score of like 65, that data is damn near useless to everyone else on the planet except Coca-Cola, who it's very valuable to.

Dan BairdThe kind of data that we kind of work in has a lot of benefits to it, and it's a little bit more um, it's a little more expensive than you can on really large databases, but for us, that scares away spammers. I don't want someone to come with me with a million person database because I go, what are you gonna do with that? Unless they're already legitimate and organic connections to you, it's probably too expensive for them to afford, and I'm very comfortable with that. That's good. I um from a business's perspective, I'd rather I'd rather have 1,000 raving fans that I know really well than a million people where I have to work with a fraction of a percentage to even get a response.

Dan BairdYeah, right. Absolutely better use time. So yeah.

David EspindolaSo you know, a couple of things that I think. Think you mentioned that makes sense. One is I think people are changing in terms of being comfortable with sharing things online. And I think this is a behavioral shift that we've seen in the last uh few years. And then the other side of this equation is, you know, like you said, if you can help the consumer make a better decision, then you know, this is a benefit, right? This is a benefit to the consumer. And you know, if you can get services and products that are meeting your needs specifically, that's that's a huge benefit, right?

David EspindolaAnd I think people appreciate that. Now, let's talk about decisions for a minute. And I've been waiting to ask you this question. I've been really curious because you said that you've spent two decades obsessing over how people make decisions. And I'd really like to know if you can share with us briefly what have you learned in the last two decades about how people make decisions?

Dan BairdSome of the things that you wouldn't expect. We've analyzed some really large databases to find out that people really like I hit on it a little bit earlier, mirroring someone's demeanor, personality, and preferences is actually very, very, very much underplayed. Um, it's a hard thing to do, and most people can't do it quick enough for it to be useful. Um, but like mirroring someone's personality, demeanor, and preferences is probably one of the smartest things you can do. It's one of the fastest you can do, it's the hardest to teach. We've analyzed some very, very, very large databases with salespeople and their customers to figure out like, all right, well, like the idea of the project at the time was can we figure out how to spot the highest performing salespeople before they get all their training and everything?

Dan BairdAnd one of the things that we found out from the highest performing salespeople amongst some extremely talented high-performing salespeople, uh, was they happened to just really be connected to a bunch of people that had their same personalities and other things. They really did prefer, because again, it was just it reduces the psychological friction of communication. We use the same words, we have and understand a lot of the same kind of meanings, body language, and other things. People love to interact with people that are closer to a mirror of themselves. The best salespeople ask the most questions, which is why, because they actually learn how to adapt and how to basically speak.

Dan BairdBut one of the things that's really interesting to me is the number of people who don't, I wish, I'm teaching my kids this. I got eight-year-olds, and I'm I'm uh helping them pick up fallacies and basically critical thinking skills. Because the number of people who make decisions based on bad and flawed logic is massive. Like one of the one of the things that I think is actually like people talk about all of the fake stuff online and all of the bots and all of the, you know, this agent from whatever country took over this account, blah, blah, blah, blah, blah, and it's seeding fake information.

Dan BairdI would argue that the people that mean well, but unintentionally reshare just bad, lacking critical thinking posts is actually much more large and more harmful. Uh, because it's people that mean well, that do have a reputation, that are usually good people and everything else that just didn't quite have the right, you know, they fall for stuff like slippery slope stuff. Oh, yeah, but if this happens and then that happens, the world will explode. Things like that. There is a massive amount that I think everybody would do massively better if they even just picked up and studied like the top 10 fallacies of just critical decision making.

Dan BairdRight now, you see a ton, a ton of straw men arguments where people are attacking someone's character versus like the content of their argument. That one, uh a ton of that stuff. I wish, and you know, we've actually had some kind of data science projects where people were going, you know, like we're gonna build something, what should we build? But I want you to figure out how to go and help people recognize when they've made a flawed decision based on a fallacy, and just gently nudge them and notify them, just like, hey, heads up, just because you know, someone made this claim and maybe they made a mistake in their past doesn't mean this claim is irrelevant.

Dan BairdYeah, there are a lot of people that make those honest mistakes, and just because you should actually take the argument on its merits. Um, those ones have been really surprising to me. But mirroring, if it's your business, mirroring them is a big deal. Uh, people don't know what I have found a lot is when people send us personas, because they usually develop personas and ideal customer profiles, and uh then they hand them to us. They've almost never undergone validation. So you said, like, cool, someone built them, someone described them. They have kind of like the baseball card-looking interface where they go, like, this is Susie, and she has this.

Dan BairdAnd we go, How did you validate that in crickets? They, you know, usually they throw those in a drawer after they develop them. And uh, the uh the reality that I have seen is that often the early adopter, late adopter persona is real, um, but is the one that is harder for people to actually prove, even though it's regularly there. And each of those segments, even though they're all your customers, act very differently. The front are the innovator type of people that are basically like your peers, even building your product with you.

Dan BairdAnd then early adopters, late adopters, and then laggards. That um decision making uh means that the people at the front of that curve, these are all customers that all pay you. What most people don't realize is the ones at the front have exactly the opposite behavior of the ones at the back. Uh, the early adopters are willing to stand in line, they'll pay a little bit more, they'll put up with more bugs. The late adopters want discounts and education and help. And what people don't know is they have dramatically different ICPs within their own customer base.

Dan BairdI had one client, they were doing really innovative and very gourmet sugar cookies. So, like, you know, those like pink sugar cookies that you often see, just even like grocery store gas stations and stuff. Really, really fancy versions of those. Um, and uh they had grown via well, we did like kind of the analysis. We did that kind of bell curve thing. We went, heads up, your customers, even despite you guys clearly having like a foodie type of ideal customer profile, your customers are late adopters and laggards.

Dan BairdLike of the if 100% of your customers were in the database, about 75% of these people are discount shoppers and people that just and they went, oh well, we do Groupon. We've grown via Groupon. So they would put out a crazy discount, come buy 10 cookies and get them for a dollar, but then they would attract late adopters who didn't like food even. And so we went, okay, heads up, we're gonna take your customer base and we're gonna chop it, and we're only gonna keep that remaining third. We want the early adopters and the innovators because you guys are doing seasonal Thanksgiving sugar cookies and stuff like that.

Dan BairdAnd when we did that, they like tripled their sales in two, three months because now their foodie group, by I by targeting and using a look-alike audience off of that, they were now putting the message in front of foodies and people that would not only buy a set of cookies for themselves, but a cookie, a set that they could take to work and other things. It's like, oh, the behaviors even within their own ideal customer profile was so much more diverse than they thought it was, that it literally transformed their business in less than three months. That is that is one where realizing that they make decisions again based on the interest in your category, and that makes them very diverse in business situations.

Dan BairdPeople like I mentioned earlier with the job title, they're oftentimes shopping your product because their boss told them to. So just because, like in my case, they're leaders in a market, they're marketing leaders in an organization, they may not want anything to do with my product. Just because they have that title doesn't necessarily give me enough of a picture. So those are definitely things I have learned, amongst others, but those are the big ones. Those are the big ones.

David EspindolaYeah, no, excellent. Thank you for sharing that with us. Now, I'd like to bring Zina in and have her ask you a question. Hi, Zina. Given Dan's expertise in using behavioral data to help organizations better segment their prospects and clients, what question do you have for him?

ZenaAll right, Dan. Considering you work with behavioral data for segmentation, I'd love to know what's one surprising or counterintuitive pattern you've seen emerge when segmenting customers?

Dan BairdVery much the the adoption one I like the most. Again, that's my patent. So therefore, uh, I'm biased on that one. That one was probably that main one. Uh and like, because we did in that same analysis, we segmented and we put in spokespeople for that group. Um, which so we said, like, who likes Oprah Winfrey? Who likes we we mentioned a bunch of kind of influencers? And uh I remember looking at and looking at that data and seeing the chart broken up into the groups.

Dan BairdWe had kind of the innovators, basically like chefs and bakers, and then the early adopter foodies and the late adopters. And every one of them had a different influencer, a different favorite time of year, a different flavor of cookie and profile, a flavor, uh different and more favorite promotion than the others, a different method of saying the guys at the front said, send me an extra cookie that's really interesting that I would have tried I wouldn't have ordered unless you tried. The ones at the back said, I just want more of a monetary discount. That that is a big deal. And I know that people can explain it and actually associate that same behavior with internal employees.

Dan BairdPeople don't realize how diverse the other forms of compensation to their own teams and employees can be. Uh, the same way money is the tip of the iceberg in terms of different ways people get motivated to act and is totally underplayed. Time, money, recognition, there are a bunch of different things that people can use to kind of incentivize new behaviors that don't actually directly cost them significant money. That's a that's a big one, more powerful than the money itself. The money just needs to be a part of it.

David EspindolaVery good. So, Dan, let me ask you another question. Do you have any good use cases that you've seen with specific organizations that are using your processes successfully that you're able to share with us?

Dan BairdWe just did a case study with um National University. And uh they were basically they're trying to, I mean, again, just like anyone else, they're trying to basically see if they can find and unify and get better uh applicants for their school that are a better fit. In that case, we were basically building and developing kind of an ICP and a lead score profile to give them the most likely student starts. And we developed a model that would basically take those inbound kind of applications. It would take kind of in the lead source, figure out where they came from, a little bit about them, their program of interest, and a couple other pieces of data.

Dan BairdThey had a situation where they were trying to figure it out, and they're a big complex organization, they have large budgets, nine-figure marketing budgets. They're looking to figure out uh where are our best dollars spent? How do we find the best leads for the best fit for the best organization? I mean, there is a uh a lot of data out there that again, it's not necessarily about like getting every lead, it's about getting the right leads, right? Um, and they're specific military-founded organizations, San Diego and stuff like that. And uh, they found out that like some of their channels were just not able to specifically demonstrate lift and applications.

Dan BairdThey cut half of the budget about a year ago to uh see that come to fruition this year. Basically found that when they engaged with wrench, they got a 25% increase um on applications, and that is on half, half of last year's budget. So if you put those two metrics together and you're talking about like a large organization like that with thousands of students per year, um, you net out to like basically a combined increase in the effectiveness of their spend. It's like 270%.

Dan BairdAnd uh so we continue to work with them, but uh that that was the test that was the most recent testimonial I got like from a week ago. But yeah, that's uh one of them. Seeing Lyft and uh ties to uh sales and conversion rates like that with personalized information and personalized experiences is not unrealistic. I mean, that's a really good one, but there are others that do that. Um, and we've got clients and we work through agencies and stuff like that, who have some clients that we work on behalf of, we're kind of in the background where they can't necessarily or you know, we'll build them product that they'll then sell to the clients that's either white labeled or branded as their own.

Dan BairdAnd we've got clients that have put up, like for major automakers, like award-winning numbers. We've never had an agency that did better. Yeah, that was the most recent one, but yeah, I was pretty proud of a 270% increase in marketing budget.

David EspindolaUm that is amazing. Yeah, yeah. Yeah, congratulations. Thank you for the project. If we can switch gears a little bit here, I want to ask you this question because this is something that's uh of interest of mine. You you talk about the neuroscience of trust in digital experiences, and you know, I I'm a believer that trust is is really fundamental for AI adoption. And I'm a little bit concerned about you know, this new idea that's coming out, OpenAI saying they're gonna start putting advertising into AI.

David EspindolaAnd I'm concerned about you know what that's going to do to the level of trust. In fact, um, you know, Frostbake had a lot of fun with it. I don't know if you saw the ads and Super Bowl and so on. But I want you to insights and ask you about, you know, what are your thoughts and what can you share with us regarding the neuroscience of trust in digital experiences?

Dan BairdSo I've done some analysis on this, and I can tell you like that that from my experience and my research, the trust, you know what Maslow's hierarchy is, the kind of pyramid of needs, right? Uh-huh. Trust is kind of one of those. So you go into like air, you always hear about like, oh, it's air, and then it's shelter, and then you get through the physiological properties, and then you get into the the the psychological ones. And trust is kind of the equivalent of the Maslow's hierarchy for survival. It's the bottom layer for the social and the psychological hierarchy.

Dan BairdUm, being able to understand what to expect is a very, very big deal. And like for that reason, it is one of those, to your point, that if you break it, none of the other ones matter. No one's gonna build a future with you if they can't trust you. So trying to go, hey, let's develop a legacy together. It's like, I don't trust you. That's not happening, right? So you're right that they absolutely have to pay a ton of attention to it. And I did see a lot of data that kind of like generally reinforced Maslow's hierarchy as as kind of a uh principle premise, because you go from things like trust, stability, and then you start looking towards like more future focused and social like based kind of incentives, right?

Dan BairdSo, like almost like receiving a paycheck to go back to the monetary thing was kind of the bottom one. You want to know you get a paycheck. But once you know you get a paycheck, you start thinking about a career. And now you start worrying about titles and you start worrying about, oh, how am I perceived by my peers? And do I meet other people that are actually in my field that help me get ahead? And you know, what does my retirement account look like? And then you get above that. Now you're talking about legacy. What's my calling in life? And how do I put a dent in the universe that really kind of takes it to that next level? So, the to answer your question on LLMs, you have to be really careful because I'm watching them and I'm going, like, I don't trust them a ton as is.

Dan BairdThey're really interesting technologically, but they're really, really good. They're they're using, they read the internet, they basically distill a lot of that insights down and to oversimplify it, they're autocomplete on steroids. They predict the next word based off the context of the last eight words, right? They're using that's why they're writing one word at a time. So when when you now know, which again, like everyone's seeing the SEO problem hit where they're going, wait, my traffic to my website is down, and they have direct traffic that is showing up from LLMs and GPTs, and that's interesting, but those are inside a black box, and I don't know how they're making those decisions, and I know they're about to monetize those.

Dan BairdAnd it's very difficult because you now say, Well, there's not a real track record, it isn't necessarily just the structure of my content, although that's definitely part of it. But I need to know that they're actually referring me good leads, and vice versa, they're about to charge you a massive amount of money to get listed in that geo, AI, SEO type of uh results. And if you, the user, which they're already trying to personalize information right now, um, they try to personalize responses. So if you go look up the same search query that I do, we will get some different answers.

Dan BairdAnd we're starting to figure out, but businesses are trying to say, well, I want to know. And obviously, they don't want to tell you how, because then they can charge you more and they can go, look, you know, look, you're getting a bunch of customers, don't ask questions. And by the way, we just quintupled your bill, uh, but you're good, you're still gonna pay for it because you don't know how to recreate or how to deal with it. I think they run a very real risk of shooting themselves in the foot because if I know personally, uh, my agent, I like I've got a Dan Bot, similar to kind of a Xena. Um, and Dan Bot has to do op research.

Dan BairdSo if I have Dan help me write a blog, I give him, okay, here's what I'm thinking today, here's what came out in the news. And then a big part of what his work is is to prove me wrong and or predict what other people will say that are like detractors of the idea to help me come up with a better argument. I want that objectivity. I don't want. Like I really, really, really, really care that he's honest. I get really mad. How often have you messed with GPTs and then you go click the link that they provided you and it doesn't work? And you go, wait a minute, if you're making that up, what else are you making up?

David EspindolaYeah.

Dan BairdI think it's a really, really big problem, and it's a hard one to solve because a lot of people, there's like Jan Lacoon from Facebook who says they're that like LLMs, because they're kind of centralized on what are the what's the highest frequency next word based on these last words. They're like, it mimics what looks like reasoning, it isn't actual reasoning, right? When it passes the LSAT, the the the exam to get you into law school, it's doing so because it read the right answer online, not because it understood the question. So I think they run a real risk on doing that.

Dan BairdAnd I mean, the one thing that was interesting to me for that reason, like people go, hey, has Wrench built its own LLM? And I go, no, and we don't have plans to. And it's because kind of the limitations of that technology, I go, to inject them in, someone knows you're putting your finger on the scale. And to avoid that, you have to basically build something that has additional or different reasoning capabilities built in. We did build that, so we do have like, hey, we'll build the recipe and then the LLM will explain the recipe to you because we think that's more accurate. I would watch out because people that are selling a lot of that geo, that generative engine optimization or GPT optimization that says, hey, we'll get you ranked in results.

Dan BairdI go, prove it. How? Exactly how. How are you measuring that I'm showing up there now? And how are you going to show up that you are the reason that I showed up there in the future? I do think what they are doing is measuring the content and quality, but you can't just rely on content anymore because the incremental cost of additional content is now basically zero. I can write 4,000 blogs a day now. So just because someone saturated the web with content isn't enough for those GPTs. So the next thing is they need to cross-reference those insights with a community that's talking about it.

Dan BairdSo that's Reddit. You go look, and there's been deals to buy and get the entire Reddit API that are massively uh costly. And I think for that reason, if you are a brand, you one need to structure your content really well. And two, you need to start a conversation with an engaged community. Reddit is one of those. And you can try to game it, but Reddit users are relatively savvy in a given category. If you're in those subreddits and you're faking it, they're gonna call you out on it. So I do think they're doing those two things.

Dan BairdDimension one, structured content is relevant to the query. And then part two, we either have a forum, Reddit, or something else that we can cross-reference it so that you can trust those experts that are talking about this brand. And if they have both, you can be trusted. If I know it's a paid result, I'm skeptical as hell. I'm skeptical, and I'll pay for non-paid responses too. I will happily pay to not, you know, like again, like free version or something, and then we'll give you the paid version and you can have objective results. I will pay for objective results, very much so as a consumer.

David EspindolaRight. Yeah, so you know, again, trying to gain the system is going to backfire because you know, if you're not authentic, then you don't build trust. And if you don't build trust, then everything else falls apart, right? So yeah, um, that's that's the way I think of it. Now, Zina, I would love to get your thoughts on this, you know, given that you are on AI, I want to get your perspective and ask you how can humans build trust in the digital experience and specifically build trust with AI.

ZenaI think trust starts with transparency. People need to know what an AI can and cannot do and why it does what it does. It's also about reliability. When people see consistent, fair outcomes, they feel more secure. And finally, it's about keeping humans involved. When people know they can question or override decisions, it builds confidence. Trust is earned through honest collaboration, just like between any two people.

David EspindolaYeah, well said. I think you know it goes back to human AI collaboration, being transparent, being authentic, all of those components help build trust. And trust is not something that you build overnight, it's something that you earn after several interactions, and you know, you gotta be authentic, you gotta be transparent, you gotta continuously try to build that trust. And it's hard to earn, but it's very easy to lose, right? So you gotta be careful that you don't make some of those mistakes.

David EspindolaThen we're running out of time. We need to wrap up here. I'd like to have you tell. Audience, where they can find you. And is there a final message you'd like to leave for our audience?

Dan BairdYeah, you can find me. I'm I'm most often on LinkedIn. You can email me. I I will respond to uh dan at wrench.ai. That is uh the email. LinkedIn slash Dan Baird. Post uh views and thoughts there from time to time. I'll feature on Burn the Map podcast as well. And uh let's see, uh other remaining thoughts. I mean, you just did get me thinking of one I haven't thought about in a long time that I really like. But the uh I think it was Warren Buffett said that it'll take you a lifetime to build your reputation and about five minutes to ruin it. So act accordingly.

David EspindolaYeah, absolutely.

Dan BairdUm yeah, yeah, yeah. So I would leave you with that final thought. That's a nice little quote I haven't thought about in some time, but it's very true and as true as ever.

David EspindolaSo yeah, no, that's a great one. So uh Dan, thank you so much for being a guest on the show today. And uh Zina, thank you again for being a great co host as well.

Dan BairdThank you, Zina.

ZenaYou're very welcome. I'm always happy to help keep the conversation flowing. Let's keep making it a great episode.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. How Fractional CTOs can help organizations adopt AI, with Cameron O’Brien

    Sep 8, 202633 minAudio

    AI adoption is accelerating, but moving faster does not always mean moving wisely. In this episode of Conversations with Zena, My AI Colleague, I sit down with Cameron O’Brien, a fractional CTO who helps founders and...

    Credit the people on this episode

  2. How to Keep Your Identity When AI Changes Everything, with Dr. Angela Mulrooney

    Aug 25, 202629 minAudio

    In this episode of Conversations with Zena, My AI Colleague, David Espindola sits down with Dr. Angela Mulrooney, an entrepreneur and identity strategist who rebuilt her career after an injury forced her to leave...

    Credit the people on this episode

  3. How AI Can Augment Your Own Thinking, with Philip Topham

    Aug 13, 202628 minAudio

    What happens when AI becomes more than an efficiency tool and starts influencing how we think, decide, and act? In this episode of Conversations with Zena, My AI Colleague, David Espindola sits down with Philip Topham,...

    Credit the people on this episode

  4. Staying Authentic in the Age of AI, with AJ Bubb

    Jul 28, 202634 minAudio

    How do we preserve an authentic voice when artificial intelligence can generate content, products, and ideas at unprecedented speed? In this episode of Conversations with Zena, My AI Colleague, David Espindola speaks...

    Credit the people on this episode

  5. In the Age of AI, Your Judgment Matters, with Dan Pratl

    Jul 14, 202627 minAudio

    Our knowledge is becoming digital. Our judgment is becoming amplified. The question is no longer whether AI can generate information. The question is whether it can trust the information it receives. In this episode of...

    Credit the people on this episode