Marketing AI Institute | Blog

[Marketing AI Show: Ep 7] How AI Can Accelerate Revenue

Written by Sandie Young | Mar 10, 2021 4:36:00 AM

The Marketing AI Show—the podcast that helps businesses grow smarter by making artificial intelligence approachable and actionable—is BACK with another episode.

You can listen now on your favorite podcast app, or keep reading for more on what to expect in this episode.

Episode 7: Maggie Crowley and Jordan Esten, Drift 


In this week's episode, show host Paul Roetzer dove into conversational marketing and conversational sales with Maggie Crowley, Director of Product Management at Drift, and Jordan Esten, GM at Drift. If you’re not familiar, Drift is a revenue acceleration platform that combines chat, email, video, and automation to remove the friction from business buying. 

During the conversation, Maggie and Jordan discussed how they use AI to increase quality pipeline that closes faster. They also shared how Drift thinks about AI, plus how the company is integrating AI into its platform. Listen to the full episode for more on:

  • Virtual Selling Assistants and how they differ from other chatbots.
  • What an AI-powered industry means for marketers.
  • Tips to get started with conversational marketing or advance the process you already have. 

Watch the Video 

Read the Full Interview Transcription 

Disclaimer: This transcription was written by AI, thanks to Descript

Paul Roetzer: Welcome to The Marketing AI Show. I'm joined today by Maggie Crowley, Director of Product at Drift and Jordan Esten, GM of Drift AI. Welcome Maggie and Jordan. Great to see you. So in this episode, we're going to explore how Drift thinks about AI, how the company is integrating AI into its platform, and what an AI powered industry means for marketers.

[00:00:30] But first I want to know how a former Olympic speed skater and the former leader of a voting technology company ended up working on AI at one of the fastest growing SAAS companies in the world. So let's talk origin story, Maggie. How did you get where you are today? Leading product address?

[00:00:47] Maggie Crowley: Yeah, that's a good question.I mean, I think that the age old answer to this is when you get a seat on a rocket ship, you should take it. Uh, so I joined Drift well before we had, uh, the AI product that we have now and have just been working [00:01:00] on how do we help. Uh, change the B2B buying process to be more buyer centric, looking at, you know, new technology trends and trends and how humans communicate.

[00:01:09] And we sort of saw the writing on the wall as a business, that there were better ways to, um, help buyers get what they need and that we could take advantage of some of the new technology. Um, out there, especially with AI. And so we acquired a company and I was sort of right place right time and got the opportunity to work, uh, on that acquisition.

[00:01:27] Um, and bringing that technology into Drift. And since then, it's really been really interesting since I've worked on our, our other bot technology to see what's possible with AI and how we can improve the way conversations, work with that. So been a really interesting learning experience, excited to share more about that today.

[00:01:43] Paul Roetzer: Awesome. And Jordan what about you?

[00:01:46] Jordan Esten:  I mean, I, you know, I'm a few years later than Maggie and I used to follow her on Twitter and really valid draft. I mean, I think what drift did a phenomenal job was building this incredible brand in Boston, especially. And for those of us who were [00:02:00] building other great companies in Boston, uh, you know, I think Boston is really known for tackling tough.

[00:02:06] Difficult, uh, both techno technology solutions, but also, uh, just, just business solutions. Right. And so I think, uh, I, I follow drift keenly for a number of years. And, um, and what I really was looking for in, in the GM role was a great fit was, you know, I'm a. I'm sort of try to be in the center of everything and connect the dots.

[00:02:25] Right. It's not, uh, I always say I don't, you know, I don't necessarily have all the answers, but I know when something needs to be connected to, to solve that when I need to bring Maggie in or when I need to put her in touch with a customer or somebody on our team. And, um, and so when, when the opportunity came, you know, I've been really interested in curious from, you know, the AI side and sort of connected.

[00:02:46] You even connected how standpoint from my perspective. Right. Right. And I think, um, it, it was just a great opportunity to come in and bring my experience running a company as a, as a CEO, before this, uh, down to, [00:03:00] you know, really running our fastest growing part of the business. Right. And, and connecting our, uh, our product teams.

[00:03:05] So our sales teams to our customer success teams, to our AI designers and, um, and it's been a phenomenal, phenomenal year. And, uh, it's a great team to work with. Yeah, and

[00:03:16] Maggie Crowley:  I think it's been interesting having this is the first time, at least on the product side, I've worked directly with a GM and I think it's been amazing to have, to be able to focus.

[00:03:25] I think what Jordan brought to the team was let us focus a little bit more on how do we get the tech, right? Because operationally AI is very different. From a product side. So I think it's been great to have, um, two people, you know, it's a split split some of the problem between between each of us. So we could actually make our customer successful while still building this thing.

[00:03:43] Paul Roetzer: So is Drift AI, a practice unit? Is it a division of like, how do you describe drift AI as an element of the company?

[00:03:53] Jordan Esten:  Yeah. I mean, I think of it as it's, you know, in five years or two years, right. [00:04:00] It's going to be, you know, our biggest product or our biggest selling product. And for now it's, uh, You know, it's, uh, it's what we can bring to, to both the enterprise and some of our, our, our, our fastest growing customers, uh, to, to expand beyond our current product line.

[00:04:15] But it's where it's going as the platform, right? Like it is the platform that, that will power. The AI itself will power a lot of different things that drift does. Um, but we, I think we've been very careful to separate it into its own business unit because then it just gets sort of set aside. Right. And, and it, it has to grow on its own.

[00:04:35] It's going to grow and it's going to power Drift growth more than it's going to just grow on it.

[00:04:39] Paul Roetzer: Yeah. I love that idea. I know Maggie, you and I have talked before and we did that webinar in the state of marketing AI. It's just this whole idea that. In three to five years, AI is infused into every element of marketing and sales software.

[00:04:49] So the idea that you have this standalone thing doesn't eventually make sense. And even some enterprises, like I'll get questions a lot of times from marketers, like, do we need an AI division or an AI strategy? It's like, you [00:05:00] need a smarter tech strategy. Like I don't, you can call it an AI strategy if you want.

[00:05:04] But what you need is a way to look at differently. Look at problems differently. Look at things you're trying to solve differently. Achievement of goals differently. And. You can do it smarter with better technology. I mean, it's, it's kind of a simplistic way to look at it, but yeah, I, I love the way you're talking about it.

[00:05:18] Yeah. And I think part

[00:05:19] Maggie Crowley:  of it is also it's allowed us to figure out how to operate. As a team and how to use the technology, both. I mean, drift is the customer is our first customer of our products, as well as, you know, obviously our customers. So I think for us, it's also allows us to kind of be a bit of a lab.

[00:05:35] So we do have a drift AI lab. And part of what we're doing is figuring out, you know, how do we incubate these new processes and systems and products before we bring it to everybody else. So we're not sort of. Disrupting the way that we all work.

[00:05:47] Paul Roetzer : [00:05:47] So for people who aren't familiar, Maggie, maybe give us a quick rundown of what is Drift, what is a revenue acceleration platform, which is how Drift positions.

[00:05:57] Maggie Crowley: So our whole philosophy is that the way [00:06:00] that people communicate is changing, you can see in the consumer market, how the rise of video, the rise of the rise of messaging and text-based communication. And we believe that business B2B buying is, is going to go in the same place as. As the consumer world, because, you know, we're just consumers that happened to be at our day jobs.

[00:06:17] Um, and so our, our whole idea is that we think that we can be, we can accelerate the path for buyers and accelerate how revenue comes in for companies by helping buyers get what they need faster. So answering their questions faster, being available to them, giving them personalized information at scale 24 seven.

[00:06:34] Um, starting with the website. So that's what we believe. That's what we're working on. And so we're just kind of riding that global mega trend and AI plays a big role in that because it allows us to provide those personalized experiences at scale, without just needing to add, you know, more and more and more sort of human labor on, on the backend.

[00:06:50] Paul Roetzer:  and Jordan, I mean, you've talked a little bit about why don't you expand maybe for a moment on just the overall point of view on AI? Because so many times like we'll guide people like, listen, if you're talking to an [00:07:00] existing tech company, you know, maybe it's part of your tech stack already. They better have a public point of view on their overall belief on AI and its role in the industry.

[00:07:08] If they don't, there's a really good chance they're years behind their peers. So how do you think about the point of view Drift has on AI and its role in the future?

[00:07:18] Jordan Esten:  I think what we've noticed very quickly is that it's a different benefit and solution for every company. Right. And it can solve different.

[00:07:26] Pain points. And you, you don't necessarily need AI for everything that, that you pain points. Right. But you want to identify, uh, where a match for AI is. Right. We have some customers that need to generate more leads or qualified leads, and it's about. You know, the, the, like let's get more, let's get more, let's get more.

[00:07:46] And AI can do that before they have to scale up a whole, a whole team of BDRs, multi bus yards. Right. It can help with that. We have other customers that they're actually getting too many bad leads, right. And so they want fewer meetings and they want to support it, [00:08:00] sort of deflect these bad questions. And, uh, and, and the AI can take that versus having, you know, their support team or their sales reps have.

[00:08:07] I have too many meetings. So it's, it's really about, I think, understanding what your pains are and then layering AI. Um, on top of that, it's a solution for the right thing. It has to be used for the right things. Right. And I think too many people just say, well, we're just going to bring in AI.

[00:08:23] And it's actually a shift we've made from saying, well, we're just going to layer AI over your entire site. Like, we want to be very specific and thoughtful about where it goes, where it starts. And, uh, and really what pain is solving.

[00:08:36] Paul Roetzer:  And Maggie, I know you've talked to customers a lot, so like where is it landing with them?

[00:08:42] And what's, what is your feeling on the understanding of AI and asking like, to Jordan's point, it's not about, can we just add AI to a bunch of stuff? It's, here's what we're trying to solve. Can we get there more efficiently? Can we accelerate success faster? Um, where do you think the market is? Is there in their [00:09:00] understanding of AI and what it can be used for?

[00:09:02] Maggie Crowley:  It really depends on the company and the team that we're working with. So I think there are some marketing teams and sales teams that are very familiar and they're, you know, maybe they even in their own products have AI. And so there is sort of get it. And for those customers, I would say you can kind of skip past the foundational understanding you have to build and get right to, okay, here are the problems we're going to solve with this technology.

[00:09:24] And they're kind of off to the races, but I would say the majority of customers, and this is a lesson we learned very early on. They don't necessarily understand the, like what AI is. There's tons of fear and uncertainty and doubt in the market on what the technology actually is. And it seems really scary and people don't want to sound stupid.

[00:09:41] And so they, you know, they don't want to ask questions and we had to learn how to be really explicit about what the product does and doesn't do, especially because there's, you know, there's like, Science fiction narratives out there, and people will assume that the tech, because it's AI, the technology will do things that it can't do or won't do, or we don't want it to do.

[00:09:59] And so I [00:10:00] think we learned very quickly that, um, what, what we try to do is get people to understand like what it is, what it will do, what it won't do, and then show them as fast as we can. An example of how a conversation using AI, something that you can't have, you can't program. Like by hand with a decision tree and once a customer kind of starts to get that and see how it's different and see how it can be more flexible, more buyer centric.

[00:10:24] That's the point at which they say, Oh, okay. I understand why this is better. And now I'm super bought in.

[00:10:29] Paul Roetzer: So we're in the product. Is it being used? Like how, how are you, is it layered in specific features or is it really starting to be applied across. If I'm a, if I'm a new drift customer, am I automatically going to be using some element of AI?

[00:10:47] Maggie Crowley: Our main AI product is isolated. It's totally isolated, but it's sort of like its own thing that you, that customers explicitly purchase. Because like we said before, we sort of started with this version, this one use case in a, for ourselves [00:11:00] following the advice that you gave earlier to say, okay, we're going to solve this one customer problem with this technology.

[00:11:06] We're going to end. We're going to figure out how to sell it, sort of end to end how to onboard customers, how to operate. And then once we understand how it works with the technology, we're starting to bring. AI it's elsewhere in our products, but we wanted to get that one use case right first. And that was our virtual selling assistant, which helps give buyers personalized answers at scale, as they're looking to get more information about product and that that's something virtual selling is. 

[00:11:25] Paul Roetzer: Isn't that a recent announcement around like the product? Is that what you want to explain a little more what that is?

[00:11:34] Maggie Crowley: Yeah. So it's really our view that, um, Rather than relying on sort of hand programmed decision tree chatbots that you might've seen on other websites. So what we could do is use conversational AI to learn from past conversations, to learn from, you know, people who have been chatting like an SDR or BDR, and to build a model from, for each individual customer that allows them to give personalized answers to their buyers at scale.

[00:12:00] [00:12:00] So that's what the virtual selling assistant does. Um, And yeah, it's, it sort of brought together a bunch of really interesting combinations of technology and human, a human in the loop training model, um, to provide that kind of answer.

[00:12:13] Paul Roetzer: Jordan. Why don't you talk more about the human in the loop side?

[00:12:15] Because I think people don't understand that it's that magic switch like, Oh, it's AI, we're done like no more human needed, but that's not the case. 

[00:12:22] Jordan Esten: Well, I think it's one of the reasons why, you know, Maggie said it's isolated, right? Like we were making very customized, uh, models and, and building for, for every different customer.

[00:12:32] Right. And, and of course we have this, this incredible base model that everything starts with, but like, we really. Uh, take the time to have an, an AI conversation designer work closely with the customer team to, to make sure we have their brand voice to make sure we're, we're under we're knowing the questions that the bots gonna receive, that we're proactive with with the qualifications that they care about.

[00:12:55] Like we're, we're really putting in the time, uh, to build a customized experience for them. [00:13:00] And, uh, and that isn't just done once. Right. It's done when we first build the model. And so there's a lot of interaction. And then we have AI designers that are, that are reading conversations daily. Right. And they're continuing to learn because.

[00:13:14] You know, the types of questions that are asked of the businesses change, um, the model, we don't get them all in there and right from day one. Right. So customers don't, and that's, that's a learning too, right? It's like customer expectations when you go live. Um, but the bot doesn't. Get dumber over time, right?

[00:13:32] We get smarter. The bot doesn't quit and, you know, go work for somebody else. It stays around, you don't have to re onboard it. So there's a lot of those benefits, but it's all really driven by having that human in the loop to continue to iterate

[00:13:46] Paul Roetzer:  it and moderate it. And for people are not familiar with the terminology, like that's the machine learning, but you hear about machine learning so much.

[00:13:52] That's the idea that the machine literally continually learns, but humans are often the ones. Training at the providing the inputs and the oversight to [00:14:00] make sure it gets smarter over time. Um, yeah, go ahead, Jordan. No, so,

[00:14:05] Jordan Esten: and, and it's, and it's obviously, you know, it's, it's based on the data that is coming through.

[00:14:09] Right. And so we're there, we're not just like randomly adding things. I mean, certainly we work with the customers to say, you know, how do you want to, uh, maybe train a new areas? Or do you want to, uh, you know, change what the virtual selling assistant is focused on or added to a new part of your site?

[00:14:25] That may be sort of a fresh start rebuild, but, but the rest of it is it's based on the conversations that are going on. Right. And it's being trained, um, based on, based on actual, back and forth with customers.

[00:14:38] Maggie Crowley: Yeah, I think it's really important. That's such an important call out, but it depends like any AI project that you're going to have at your company is going to be dependent on how good your data is, um, and how long are willing to train it.

[00:14:51] So I think that there's, there's definitely a balance between. Spending a lot of time, you know, training, whatever model is you're trying to use on the data that you have versus going live with [00:15:00] it and learning on live interactions or for us, you know, live conversations. And that's something that we also dial up or back, depending on how comfortable the customer is with how, like, how, how quickly they want to go live and how okay they are with the different variant, like how much the bot may know.

[00:15:15] And that was a big lesson learned at the beginning. And I think it's relevant for any kind of AI project, which is understanding the data that you have available, how good it is and how long it's going to take you to train the AI that you're trying to use.

[00:15:26] Paul Roetzer: And what about, you know, so again, like one of the great ways you talked about selling is like, here's your job today?

[00:15:31] Here's what it's like to use a manually powered. If then driven chat bot here is your conversational agent of the future. What does that mean to the user in terms of onboarding? Um, Training needed for their team to work with a more intelligent agent. Is that the kind of thing drift provides where you're actually upskilling your customers, teaching them new capabilities?

[00:15:56] Jordan Esten: Yeah, absolutely. I mean, I think it's a, you know, it's a [00:16:00] different experience. In some regards you have to have more patients and actually let the bot do its job. And, uh, you know, we have a lot of customers who have, you know, very. You know, anxious. I mean, they have been guys, they just want to jump in, right.

[00:16:12] They just want to have comments. Um, but it's, I think it's important that the bot can kind of, you know, go through its qualification. Um, and so we, we do quite a bit of training with our customers and their teams to make sure that, uh, You know, it's, it's working as, but also working as they want it to be working.

[00:16:30] Like it should depend on their environment. And so we want to both train them on how the AI is built to work, but then modify, uh, the flow and the, and the structure of the AI based on how their team wants to work.

[00:16:42] Paul Roetzer: So we'll often when we're talking about buying AI solutions, think about it like the autonomous vehicles scale of like the zero to five full autonomy.

[00:16:50] And we created this marketer to machine scale, similar concept where like all human, all the time to fully autonomous. If someone's building this overtime, if someone [00:17:00] like, you know, buys into drift and starts using the AI functions. How much of their job are you truly intelligently automating? Is it like you want to get to 50% of maybe what would have previously been done all by human as being done by machine?

[00:17:12] Or is there an end game to try and get this to where humans aren't even in the loop anymore?

[00:17:19] Maggie Crowley: I don't really think about it as automating someone's job, the way the pro from a product perspective, the problem that we're solving is we're saying, how do we make them more efficient? So what can we do to make the people that we have better focus on the things that humans are good at and not have to do repetitive behaviors that we can, we can automate for them.

[00:17:39] And so it's more about, okay, well, if we have. You know, for example, for our products, we have SDRs BDRs who are chatting and talking to leads. How do we get them leads that are better qualified to Jordan's earlier point? How do we make sure that we're deflecting people who aren't ready for sales or who are customers to a different place, or, you know, answering like lightweight questions and making sure that the time and the people has spent on the good [00:18:00] stuff.

[00:18:00] And I think on the other end of the spectrum, you know, The AI that at least our AI, um, isn't going to replace like critical thinking and the branding and positioning and all the like really good stuff that marketers are awesome at. And so, again, it's just more about like, how do we take that knowledge and like, make sure that we're answering good questions at scale and making sure everyone's like not having to do that repetitive behavior.

[00:18:22] Um, so you can focus on the good stuff.

[00:18:25] Jordan Esten: Yeah, I I'd add to that. I mean, I think we, there may be benefits so that customers don't need to continue to add as many, like they can scale faster. And so like yes, in the longterm you get, you obviously like get those efficiencies and maybe future jobs are, you know, are, are, um, are taken over in some recycler, the AI, but not absolutely like for us, it's about making them.

[00:18:49] Better more efficient, more, you know, more volume. Right. And that's a huge, that's a huge focus and then it, and then it goes to the insights, right? It's like, it's like maybe at first it's about [00:19:00] scaling from the sales perspective, more value, but then it's about, can we, can we help educate them on the types of questions that are, that are being asked consistently?

[00:19:07] Can we allow them to make more efficient decisions, um, that are broadly based on what we're seeing from all their customers?

[00:19:16] Paul Roetzer: And are there minimum requirements of data or of number of visitors to the site to get value out of any like how much data does it need to actually get really good? 

[00:19:26] Maggie Crowley: I think the first version of our own bot, we had like 50 conversations to a hundred conversations to be trained it on. So that's one of the benefits at least of the way our system works is that you don't have to have like 3 billion lines of data for it to work. Um, obviously the more data you have, the more accurate you can be, but, um, for us, you don't need a ton, but I would say you need to.

[00:19:47] Couple of thousand conversations to have to like be off to a really great start. Um, and then I think that traffic depends on what problems you're trying to solve with our use case. Like there are some companies that have less traffic, but they have a more nuanced [00:20:00] use case. Um, but it journey you'd probably speed it up.

[00:20:03] Jordan Esten:  I mean, I think, you know, traffic is dependent on. You know, what's your average selling price it's dependent on. Um, again, how many, how many reps do you have on your team that, that is, you know, do they have time to jump into chats or not? Um, you know, we, we see, we see across the board from, uh, from how, how much traffic our customers have.

[00:20:23] I think by definition, more traffic drives more value value from it in most cases. But I do think like Maggie and the team have done a phenomenal job of. Of like allowing us to stand these up much quicker with less data and get it going right. So we get it going. And then we, and then we build data to grow smarter and more specific to their domain.

[00:20:44] Um, but, but we don't rely on the traffic necessarily for the batch strength. We rely on it for the, for their ROI and for solving their, their pains.

[00:20:51] Paul Roetzer: And Maggie, when you said 50 to a hundred conversations, you mean like an existing, like human powered chat bot, where there has been [00:21:00] conversations, the machine can learn how your reps deal with it, that kind of stuff.

[00:21:03] Maggie Crowley: Yeah, that's the way our system works is we take either conversations. People are having with a bot, but much more ideally, and more typically we would take, if you have a rep who's chatting via chat tool or on the phone, and you have transcripts, we'll use that data to train the model. Because what we want to do is, is model it after what your best STRs are already saying, like you already have great salespeople who are really great at answering questions, and that's what we want to learn from. So that's what we look for.

[00:21:29] Paul Roetzer: And then we touched a little bit on this earlier, but how do you prioritize the roadmap for what does get infused with AI when you're looking at the product roadmap for drift? Obviously like there are so many applications of AI from NLP of evaluating the text to, you know, actually learning from training data on emails or texts.

[00:21:48] How are you looking at that roadmap and trying to figure out where to invest the resources next? 

[00:21:54] Maggie Crowley: Yeah, and this is like, we can get pretty nerdy on the product philosophy side on this. But I think for us, like [00:22:00] typically what I've seen over the past couple of years is that the challenge with AI is that there's too many things you can do.

[00:22:04] And too many things we want to do. So Jordan mentioned insights earlier, there's this like whole universe of things you could understand from a conversation to that. That would be cool and amazing. Um, and then there's also a balance of, you know, what is. Unlike traditional product development. There's this, there's this aspect of, you know, what's new in the tech, on the technology and what, what does that make possible that we could do?

[00:22:26] Um, but the way I try to think about it is what can we do to unlock more value for our customers? And we start there. So what's going to help us get buyers where they're going faster. What's going to help our customers do their jobs better. And that's really the thing that wins for us is where we think we can apply AI.

[00:22:40] That's going to help with that the most. 

[00:22:42] Paul Roetzer: Do you run into the innovator's dilemma at all, where it's like, you have something that, you know, would be better, but the market isn't ready yet, or isn't demanding it. And you have to balance between this is a transformational shift in how marketing could be done.

[00:22:55] And we think it needs to be, but. Customer may not be ready for the demand there. 

[00:23:00] Maggie Crowley: Yeah, I think the way I like to think about that, and this is regardless of AR AI or not is like the, the more you understand your customer and the more you understand the workflow that they're going through and their problem they're trying to solve, they don't, you don't have, you don't have ever have to wait for your customer to ask for something.

[00:23:15] If you know that they have a problem, you can solve it, that's going to be worthwhile. And so I think it's on you as a. As a team who is selling something to tell a good story and help them understand why it's going to be better and why it's going to solve a problem that they already have better. Um, cause I think as a product person, if you're waiting for customers to ask you for stuff, you're, you're probably, I would say missing the mark.

[00:23:34] Jordan Esten: Little bit. You’re building something without talking to them.

[00:23:36] Maggie Crowley: Right.

[00:23:38] Paul Roetzer: What about, uh, patents? I know. So, you know, again, there's always the build versus buy. I know drift has made acquisitions. Um, but I also know you hold patents related to AI. I mean, how do you think about just that growing from the internal and like really building this core capability versus like, you know, looking at ways because.

[00:23:56] As we talked about, like, these are all narrow use cases. Like there's, there's [00:24:00] hundreds or thousands of companies being built to do a specific little thing. And so there's, you got to assume at some point there's some consolidation where you start just, you know, finding those interesting tools that you can roll in.

[00:24:10] Is it something you publicly talk about just in terms of your strategy or thinking there around how you'll continually stay on top of the innovation, because it does move so fast in this space.

[00:24:23] Jordan Esten:  Yeah. I think how we, you know, I, I'm not. Not speaking like company-wide, but I think from an AI perspective, like we see so much runway with what we have, like w w when we acquired a giant Otter, like, it's just, it's the core and the base of not just what we're doing now, but it's already could be the core of the base of like so many things, um, that are inline and around our current customers.

[00:24:50] Right. We talk about it. Like, we're not trying to go. Solve issues that are, are on the other side of the company. Like we're trying to solve issues that are like, you know, with the marketing and sales, like, you know, really within that, that realm and stay on that line. And I think there's so much that, you know, there's so much runway in front of us with, with the, with the core technology that we have that we're just so heads down focused on, uh, on getting applications into the marke.

[00:25:16] Paul Roetzer: Yeah. My perception from the outside is you're not buying a bunch of plug and play things like just stacking a bunch of tools. The acquisitions were core to the foundation of the tech and it's like, it's going to actually enrich the whole platform by like, so that, that was just kinda my perception. Yeah.

[00:25:32] Um, real quick ethics. So again, a core thing that a lot of times marketers don't even know to ask, but does drift have a point of view on just the ethics of AI in the development and the use of it?

[00:25:44] Maggie Crowley: Yeah, absolutely. It's something that we, we care a lot about. Um, we care a lot about making sure that our team is as diverse as possible to make sure that we're thinking of all the right things.

[00:25:55] Um, we've had a couple of, and one of the practices that we have is bringing in [00:26:00] speakers who, um, are. Well-versed in the ethics of AI and how to make sure that you're thinking about it correctly. And so we've had, I dunno, a couple of speakers over the past quarter or two that have come in to help us on the technical side and on non-technical side, understand like what we should be thinking about.

[00:26:14] Um, and one of the things that I, that I really love, especially this is from Jordan's team. Is we have, again, we talked about human in the loop and I think one of the dangers with AI is that if your training data set is so large, it's almost impo it's, it is effectively impossible to know what's in it at some sense.

[00:26:30] And to know whether or not there's any, what the biases are that are in it, because there's going to be bias, biases, and basically all data. Um, and so we, our. Training art, AI conversation designers are really part of our team and we've care very deeply about the experience they're having and what they're seeing in the data.

[00:26:46] And so that's one of the ways that we think about it as like a educating ourselves on what biases can show up and how, and be putting in processes to make sure that we're listening so that when things come up, we can act on them.

[00:26:59] Jordan Esten: Yeah, this is so important. I mean, I think that, first of all, we're really lucky.

[00:27:03] Like drift is an incredibly diverse company starting at the very top and, um, you know, and, and working down through the entire organization, um, with, with incredible ethics, right? I mean, like, it's, we don't have to, we, we just, we know that like, we need to do the right thing.

[00:27:19] Paul Roetzer: Right. You don't have to have the paper that says here's the 10 things to know.

[00:27:21] Jordan Esten: And it really is from the top, I think from a, from a diversity perspective. Um, you know, I saw this in the elections industry actually, because we were building products for anyone and everyone, right. Our end users were everybody. Right. And so we had to make sure that we had products that could be, um, you know, used by disabled voters, by, you know, a whole different broad set.

[00:27:44] And so we made sure we built a team that had that. Background, right. We didn't want to just a team of a bunch of people in Boston. Like we wanted those experiences and, and that diversity. And I think we try to think through that, both at drift and as part of our AI designers, right. I don't, you know, [00:28:00] if we have AI designers that all look the same and they're training these models, they're going to be all the same and they're not going to really represent the site visitors on the other side.

[00:28:09] And the, and the users who are again, Anybody and everybody. And so we have to be really, you know, diversity is, is, is important, but, but it's actually important. Like it, it, it actually impacts and build better product when you have a diverse team and, and it, and it builds a better culture and it's so vital and we've just focused on it.

[00:28:29] I mean, I think our AI design team is incredibly diverse in the types of people and, and. And we're, we're so lucky on that, in that regard that I think the company supports that as well.

[00:28:39] Paul Roetzer: Yeah.

[00:28:41] Maggie Crowley: As soon as I, yeah, I totally agree with Jordan. I think you can tell what this is like such a hot button issue for everyone at drift, so psyched about it.

[00:28:47] But I think it's also like, you know, having to two founders of color, like we CA it's very much a part of our company's DNA and I think. Like the other big, like big [00:29:00] tech companies for us are not role models in this. Like we think we, like, there is a space for all of us to do better and like what's going on in the market is not good enough.

[00:29:07] And we want to be, we're holding ourselves to a higher bar than we've seen elsewhere because like, we think that that matters a ton, not just for us as a company, but for the outcomes like Jordan's mentioning for our customers.

[00:29:18] Paul Roetzer: Yeah. No, and I, I think it's, you know, again, Maggie, you and I have had conversations around this stuff.

[00:29:23] Like I, my, my. My mission is to make enough marketers care, to learn what AI is, so that they start to ask these questions, because right now the average market would have no idea to even understand the role diversity plays in the creation of the AI. And that's not good enough, like as an industry, we need to be demanding that of the companies that we buy from.

[00:29:44] And we're not there. We're not even close to being there. 

[00:29:48] Jordan Esten: So I think, and Maggie made a really important may give me a really important point about. You know, we, our AI designers are incredible, incredible members of our team, and they're not sort of siloed [00:30:00] building models that, that no one is reviewing.

[00:30:03] And, uh, like we really go through that Maggie and her team really then with the designers to, to look at their, you know, look at what they're building and, and, um, and we just, it's just incredibly collaborative because if you have somebody just building in a silo, To Maggie said like you don't, you don't know what's being done there and then you want to be, you want to make sure we're really have some checks and balances on, on the team as well.

[00:30:25] Paul Roetzer: Right? Well, that's, I think that's a great place for the, the majority of this to end. I want to end with our usual rapid fire questions for Maggie and Jordan.  Alright you ready? Maggie. So I alluded to this at the beginning. If you could have been an Olympian in another sport, besides speed skating, what w what would it have been? Did you have another sport?

[00:30:48] Maggie Crowley: Obviously? Um, I started as a figure skater, couldn't cut it, switched to hockey, and then was, uh, went to speed skating. So I think figure skating probably would have been the one I would've [00:31:00] picked, although as a half Canadian person, hockey is also very close to my heart. So like any of those, maybe skiing.

[00:31:09] Paul Roetzer: All right, Jordan, if you could have been an Olympian, what sport would it have been? This is the first time I've heard

[00:31:19] Jordan Esten:  Maggie was an Olympian.

[00:31:20] Paul Roetzer: I learned it when we were doing a webinar together.

[00:31:23] That's a joke.

[00:31:26] Jordan Esten: Um, um, Oh, wow. Um, man, I was a saw, I was a college soccer player and uh, I told, I remember telling Maggie this, when I started, I was like, Oh, you were to look at it. I was a college soccer player as a kid. Everyone loves to tell me like what sport they play, girl. Um, but no, I actually, I grew up, uh, Holland for a couple of years in Europe and, and, uh, and fell in love with soccer.

[00:31:51] And, uh, I don't play anymore because I get injured, but I think that would, that would probably be the sport I would've stuck with.

[00:31:58] Paul Roetzer: Nice. All right. For both maybe [00:32:00] curling

[00:32:01] Maggie Crowley:  curling,

[00:32:02] Jordan Esten: you could still do that.

[00:32:03] Paul Roetzer: All right. A voice assistant you use most Alexa, Google assistant, Siri Cortana. Don't use them.

[00:32:12] Jordan Esten:  I've used them all.

[00:32:13] Uh, I think Google assistant is probably the one that I use the most. Uh, Think mine is my best, like music mix that I, that I love. And I think I just set it up. 

[00:32:24] Paul Roetzer: I always do stupid. I ask that question. My watch will go off. I don't use them. You don't privacy reasons are just functional. 

[00:32:32] Maggie Crowley: I just don't, I don't need something like listening to me in my house, but, um, maybe one day I'll get there, but not, not yet.

[00:32:40] Paul Roetzer: I took the Alexa out of my house. My kids wouldn't stop talking to it. Um, alright. More valuable than 10 years. Liberal arts degree or computer science degree.

[00:32:50] Jordan Esten: I think computer science. I mean, I think, I, I think that, you know, liberal arts is incredibly important, but there's so many places to learn that. Um, I think getting that [00:33:00] core, um, I mean, I went to Carnegie Mellon, so maybe I'm a little bias there, but I wasn't a science major.

[00:33:06] Maggie Crowley: I totally disagree. I think you don't need a computer science to read and to learn how to code, but I think learning how to think and how to think critically is going to be the most important skill in the future. Love it.

[00:33:18] Paul Roetzer: Biased. Uh, Mark Cuban is the infamous, like he's on record that he's, he's a believer that the codes will code themselves and that liberal arts, the human thinking is the key to the future, uh, net effect over the next decade. More jobs eliminated by AI or more jobs created by AI or no meaningful impact. 

[00:33:39] Jordan Esten: I think, created, I mean, if I look at certainly at drift and that our customers uh, it's, it's created

[00:33:45] Paul Roetzer: okay. Yep. I agree. I'm with you. All right. Well, any final thoughts for our, you know, again, our audience tends to be the beginner level.

[00:33:54] They're just trying to figure this stuff out, looking for the first couple of pilot projects, any final words of wisdom for them [00:34:00] to help them kind of figure out this new frontier we're heading in. I think

[00:34:04] Jordan Esten:  it'd be patient. Like if you're going to do it, pick, pick the pain you want to solve and then be patient and put the time and effort in to, to, to really, um, You know, bring AI and make it, make it a keen part of like that pain and that solution.

[00:34:18] And don't expect the results to come on day one, right. That they're going to  you got to keep training it. You've got to keep learning and you're going to all of a sudden, see, like, we talk about customers, it's like this, and then it's like that. Right. And I think that's what, uh, you know, have that patience and willingness for the AI to learn.

[00:34:35] Maggie Crowley: Yeah, I agree. I think my advice would be, um, Pick one problem. And maybe it's the same flavor of being patient, but pick one problem that you really want to solve, that you have good data for and work on that versus getting overexcited, trying to do too many things at once and sort of seeing your AI project fizzle out.

[00:34:51] Because I think if you get that first one, right, and you understand how, how it works and how it can be helpful, the next ones are going to be so much more easy. Okay.

[00:34:58] Paul Roetzer: And Maggie, you have the build with [00:35:00] Maggie Crawley podcast. Um, where else can people follow you? 

[00:35:05] Maggie Crowley: follow both, both Jordan and I on Twitter. Um, retweeting, each other's tweets.

[00:35:10] Jordan Esten: we just go back and forth. You just go back and forth. And I, and I, uh, you know, I am. Maggie's biggest podcast listener. And I always send her screenshots from the car or whatever, and not there I'm listening. So that's my big shout out is for part though.

[00:35:23] Paul Roetzer: Well, I really appreciate both of you taking the time to do this. This has been an awesome conversation. I would love to go down the ethics route one of these days. Maybe we just do a whole podcast dedicated that, get a couple of other people it's just. It's not talked about enough and I really appreciate you guys offering the perspective there. So, um, again, Jordan, Maggie, thank you so much for your time.

[00:35:41] And for everybody listening or watching, we really appreciate you being a part of it. Uh, this has been the Marketing AI Show and until next time we'll, we'll see you then. Thanks so much.