Marketing technology is undergoing dramatic transformation as companies seek greater personalization to engage buyers across the customer lifecycle. This episode examines how AI is changing marketing. Our guest are Sameer Patel, CEO, and Andrew Eichenbaum, Lead Data Scientist, at Kahuna software. 

Sameer joined Kahuna in August, 2016 as CEO and board member. Prior to Kahuna, Sameer was GM/SVP at SAP/SuccessFactors where he led the team that was responsible for making SAP a leading provider of Collaboration and Customer Engagement solutions with 35 million subscribers. Prior to SAP, Sameer worked consulting and systems integrator business. Sameer has a BS in Finance and Economics from Babson College and a Masters in MIS from Boston University.

Dr. Andrew Eichenbaum is a leading Data Scientist specializing in artificial intelligence and machine learning. Andrew has been analyzing large scale data sets for over 15 years and leading data-centric projects and teams for more than 8 years.

In industry, Andrew has worked with a range of large organization including Orange, Intel, and Intuit. He has also worked in early to mid-stage start-ups including Shopping.com, MyBuys (now Magnetic), and was on the founding team of Yummly. He has particular expertise in creating, leading, and scaling high-performing Data Science and Advanced Analytics teams. He is a sought-after speaker around the topics of Data Science, team dynamics, and performance management.

Transcript

Michael Krigsman: Welcome to Episode #209 of CXOTalk, and on this CXOTalk, we are celebrating the holiday season. And how are we celebrating the holiday season? We’re going to have a great conversation. That’s our way of celebrating. I’m Michael Krigsman, I am an industry analyst and the host of CXOTalk, and CXOTalk brings together people who are shaping the future; the most innovative people in the world, for in-depth conversation. And you are part of that conversation, so as we’re talking right now, there’s a tweet chat taking place on Twitter, with the hashtag #cxotalk. And I’m so thrilled, because today, we’re speaking on the topic of artificial intelligence - AI - in marketing. We’re here with Sameer Patel, who is an old friend - I’ve known Sameer for years - who is CEO of Kahuna Software, and his colleague Andrew Eichenbaum, who is Kahuna’s lead data scientist. Gentlemen, how are you?

Sameer Patel: Fantastic! How are you, Michael?

Andrew Eichenbaum: Doing great.

Michael Krigsman: I am excellent. And you know, I was thinking we should have brought, I should have worn a Santa hat.

Andrew Eichenbaum: Yeah. The Starbucks-branded Christmas cup here…

Michael Krigsman: Alright, well that’s good. So we have some Christmas cheer happening here.

Sameer Patel: Mhmm.

Michael Krigsman: So Sameer and Andrew, please. Sameer, why don’t you tell us about Kahuna Software.

Sameer Patel: Sure. So, thanks for having us. So, Kahuna software is a B2C marketing automation provider. We have built a real-time platform that allows brands to be able to understand what the interests and preferences of their consumers literally within seconds, be able to make sense of it, add it to the profile, and put meaningful offers in front of them, be that for a commerce vendor or if you’re a media company and you need to engage your prospects, this is the new way of using artificial intelligence to engage with your consumers on the right device at the right time.

It’s in California, we’re a four year old company, we’re privileged to be funded by Sequoia Capital, and SoftTech Ventures, and Tenaya is our lead investor, and a whole bunch of other amazing people supporting us. And, we’re about 60 people, mostly in California, some in New York and the others in Vancouver. Andrew?

Michael Krigsman: Hey Andrew, you’re Lead Data Scientist, so what is it that you actually do? What does a Lead Data Scientist do?

Andrew Eichenbaum: So, that actually is an interesting question, because when I was looking around at places, I joined at the Kahuna just about a quarter ago. I actually had not looked at marketing places. I had actually done consulting work with large companies; Intel, Intuit; with their marketing groups, and suggested things like recommendations, personalization, and they said, “This is great,” well, everybody had sort of left it at that point. When I was contacted, this was the first company that really thought of marketing as data, and then what can we do with this data? How can it better help us understand people and better mail to them; better send to them; know “when,” “where,” “how,” and pretty much not spam them? So, the data scientist sort of stuck in the middle of making sure the data’s all good, and then being able to figure out all of these things.

Michael Krigsman: Fantastic! So, why don’t we kick this off with a discussion of B2B, B2C marketing; your business to consumer marketing? But you’re selling to … Your end customers are consumers, but you’re selling to businesses. \

Sameer Patel: Yeah, I mean our end-customers are consumer brands who engage with you and me as consumers who buy stuff, right? So one of the great things about taking over this business is that it’s very easy to keep yourself honest and know if you’re getting useful stuff, because all of us are consumers often, or any of the recipients of what Kahuna does, right? So, you know

Michael Krigsman: Welcome to Episode #209 of CXOTalk, and on this CXOTalk, we are celebrating the holiday season. And how are we celebrating the holiday season? We’re going to have a great conversation. That’s our way of celebrating. I’m Michael Krigsman, I am an industry analyst and the host of CXOTalk, and CXOTalk brings together people who are shaping the future; the most innovative people in the world, for in-depth conversation. And you are part of that conversation, so as we’re talking right now, there’s a tweet chat taking place on Twitter, with the hashtag #cxotalk. And I’m so thrilled, because today, we’re speaking on the topic of artificial intelligence - AI - in marketing. We’re here with Sameer Patel, who is an old friend - I’ve known Sameer for years - who is CEO of Kahuna Software, and his colleague Andrew Eichenbaum, who is Kahuna’s lead data scientist. Gentlemen, how are you?

Sameer Patel: Fantastic! How are you, Michael?

Andrew Eichenbaum: Doing great.

Michael Krigsman: I am excellent. And you know, I was thinking we should have brought, I should have worn a Santa hat.

Andrew Eichenbaum: Yeah. The Starbucks-branded Christmas cup here…

Michael Krigsman: Alright, well that’s good. So we have some Christmas cheer happening here.

Sameer Patel: Mhmm.

Michael Krigsman: So Sameer and Andrew, please. Sameer, why don’t you tell us about Kahuna Software.

Sameer Patel: Sure. So, thanks for having us. So, Kahuna software is a B2C marketing automation provider. We have built a real-time platform that allows brands to be able to understand what the interests and preferences of their consumers literally within seconds, be able to make sense of it, add it to the profile, and put meaningful offers in front of them, be that for a commerce vendor or if you’re a media company and you need to engage your prospects, this is the new way of using artificial intelligence to engage with your consumers on the right device at the right time.

It’s in California, we’re a four year old company, we’re privileged to be funded by Sequoia Capital, and SoftTech Ventures, and Tenaya is our lead investor, and a whole bunch of other amazing people supporting us. And, we’re about 60 people, mostly in California, some in New York and the others in Vancouver. Andrew?

Michael Krigsman: Hey Andrew, you’re Lead Data Scientist, so what is it that you actually do? What does a Lead Data Scientist do?

Andrew Eichenbaum: So, that actually is an interesting question, because when I was looking around at places, I joined at the Kahuna just about a quarter ago. I actually had not looked at marketing places. I had actually done consulting work with large companies; Intel, Intuit; with their marketing groups, and suggested things like recommendations, personalization, and they said, “This is great,” well, everybody had sort of left it at that point. When I was contacted, this was the first company that really thought of marketing as data, and then what can we do with this data? How can it better help us understand people and better mail to them; better send to them; know “when,” “where,” “how,” and pretty much not spam them? So, the data scientist sort of stuck in the middle of making sure the data’s all good, and then being able to figure out all of these things.

Michael Krigsman: Fantastic! So, why don’t we kick this off with a discussion of B2B, B2C marketing; your business to consumer marketing? But you’re selling to … Your end customers are consumers, but you’re selling to businesses. \

Sameer Patel: Yeah, I mean our end-customers are consumer brands who engage with you and me as consumers who buy stuff, right? So one of the great things about taking over this business is that it’s very easy to keep yourself honest and know if you’re getting useful stuff, because all of us are consumers often, or any of the recipients of what Kahuna does, right? So, you know and in the B2C of the consumer space, I think in the last four or five years, we’ve seen probably the most tectonic shifts in terms of how you and me as consumers want to engage, and be engaged by brands, right?

And, you know, the specific area where Kahuna plays is the area of retention, right? So you might use ad technology and other ways to drive new customer acquisition. Where Kahuna comes in place is how do you get existing customers to engage more to make that first purchase go from first, to second, to third, and really drive loyalty for [...]? So, that is a very specific place where we end up, where the majority of our folks and all of our folks as a company.

You know, one of the most interesting things that when we look at some of the convergence and a need for consumer brands to begin to rethink how they engage, and transact with the consumers … If you look back at the last four or five years and see the changes that have happened in terms of how we expect brands to work with us, you look at some of the most common - and I apologize because I overuse examples, but I think that should matter. You know, when you look at the notion of how you procure a ride on Uber, for example: You go from this point of them making an offer to you consuming the service, and finishing the service, and at no point during the whole lifecycle did you ever think someone was marketing to you. You thought you won: you needed a taxi, they provided supply at that point and location where you needed it, you used it, you consumed it, you moved on.

You know, when you, and Michael you know I’ve talked about this, I love when you look at Amazon for example, right? And you see a wish list, and you see things that Amazon might recommend to you, for getting to a point where I don’t think - at least I will speak for myself, I almost never find what Amazon is recommending to be annoying. I may not buy, but it super relevant. And we’re getting to the space where if offers can be as relevant and as timely as they seem to be becoming, for some of the more digitally native and proficient brands, you know, we are willing to accept more and more of what might otherwise have looked like spam if it were not well-targeted, right?

So Kahuna is focused on taking that kind of an experience to every single consumer brand out there. How do we get brands to truly be able to engage, just at the right time, with the right offer, on the right device, based on the digital breadcrumbs that you and I today are leaving, every single day, you know? You’re on your mobile phone, you’re browsing stuff; you’re putting stuff in wish lists; you’re almost buying; you’re looking at goods ten times. We never had this kind of information as brands to be able to truly understand what you are interested in; how we put that to use in a way that respects how a consumer wants to work; and really, that is what gets us to work every day. We try to get better, and better, and better at [...].

Michael Krigsman: So Andrew, this notion is very interesting, what Sameer just said, that when you are buying a product, say from Uber, you don’t think that they’re marketing to you, you don’t interpret messages as spam, and so, is relevance the key for consumers to have that kind of open, warmth, and acceptingness of marketing messages so they actually seek it, rather than push it away?

Andrew Eichenbaum: So, let me take a step back to put this in context. To start off with what Sameer is saying, we’re in this sort of new era, where you can market to anybody, probably 14-16 hours a day. People are that connect to their cell phone, it’s always there, there are multiple channels to reach out to them, and that’s all through one device. This is something which has become ubiquitous, at least in the US market over the past sort of five years. Now that being said, it’s easy enough to spam them, and nobody wants to do that because people have become hypersensitive to the whole situation of, “I’ve got one bad message, okay, you’re in the cutting block. Second bad message two days later, you’re gone.” And the reacquisition cost to pull somebody back in is amazingly high.

So, the question is it’s not just not sending them spam, it’s knowing what to send them, when to send to them, how you send to them, because there’s a whole range of things. What message do you want to send to them? And, it just extends out.

One of the most interesting things you could think of is not to think about goals, like I want to sell this person a product, I want them to move the next step down my path. We’re now in an area where we can think about I want to increase my expected long term value of all my customers. I want to increase their overall engagement stake, and this is what marketers can now reach to. This can be what seemed to be a more nebulous goal before, is now something that we can actually move forward and try and act on.

Sameer Patel: You know, Michael, one of the things that’s interesting to just sort of note as we have this discussion is try to understand the baseline first, right? We always will talk about better ways to do stuff, but I always like to spend a little time just talking about where does the market sit today? So, against this backdrop of this very sophisticated consumer and their expectations that I had just described, and that we talk about, it’s really also important to say, “Okay, how does the existing technologies in place stack up to that very increasingly demanding customer,” right? And it’s pretty daunting when you start to think about how marketing automation created a decade ago stacks up to that, right? The market’s over ten billion dollars in size, yet there is over two hundred and eighty billion dollars worth of goods left in abandoned shopping carts every single year. Two hundred eighty billion dollars, right? That’s how much you and I go and we almost buy, and we put it in the shopping cart, and we leave it there, and you’re just effectively nudging the consumer to the finish line, or providing them with a handholding and the information and the research that might be required to persuade them to finish buying. So, you’re left with almost … If you look at some of the research that is out there right now, the conversion rates are 2-3% on all this expense on commerce. That’s how bad it is.

Michael Krigsman: When you say the conversion rates are 2-3%, which conversion rate specifically?

Sameer Patel: E-commerce. So when you look across e-commerce today, all the investment being put into what seemed like the right offers lead to 2-3% of conversion, right? And, you know, we’re now finally at a point where you can begin to bring together both the art and the science of marketing, given the advancements in technology which is what I’m going to talk about on the show later today, to start to say how do I take those meaningful messages; but that’s not enough, I’ve got to find A) as meaningful as they can be, B) how they’re being put out at the right time on the right device that we were saying earlier, because the consumer is telling you, “Look, I’m sharing my location with you on my phone, I’m sharing with you things that I’m interested in, this is not those days of email marketing where you knew nothing about me and you spammed me. You better well be doing much better than you are right now,” right?

And how does the backdrop where you start to truly understand the difference between the technology that 90% of consumers are using today, and increasingly what the brands they’re using today, and increasingly, the new persona of this digitally-connected consumer. Does that make sense?

Michael Krigsman: Okay, so we’re in this situation where if you can more accurately tailor the content, the message, the timing, and the channel to the consumer, to the online consumer, that person will be more receptive, because “Oh yeah, I’m getting something that I care about.” That’s the bottom line here, right?

Sameer Patel: That’s the bottom line! I mean, it doesn’t have to be commerce only, right? This is deeper on all sorts of messages that various online companies want to put in front of you. It could be about engaging for a show like CXOTalk where you’re going after consumers. It’s the same problem [...] too. Seriously, right? The thinking about how to put meaningful content won, but secondly, did they hear you? And this is why I love this business, because we are all consumers when walk out the office in the evening.

Michael Krigsman: Okay. So Andrew Eichenbaum, again, you’re a data scientist, and artificial intelligence comes into play here to solve this problem. Tell us about that.

Andrew Eichenbaum: Artificial intelligence can solve many problems. The question is can you define what you really want? You might ask the system if you tell your data scientist to go off, and I want to increase our revenues by X amount, and the data scientist goes off and increases your revenues, and then three weeks later, everybody goes away, and nobody’s bothering to see your site anymore and they’re asking “What happened, data scientist?” “I increased your revenues. We never talked about lifetime value and things like that, but you were able to increase your revenue by 50%.” [Laughter] It’s a double-edged sword.

Michael Krigsman: “So I spammed everybody, I increased your revenues like you told me, but you didn’t tell me you wanted me to keep these people coming back!”

Andrew Eichenbaum: That’s right. So, it’s a double-edged sword. Data science is sort of half science, half black arts, much like the rest of marketing. You have these fields and understandings of what’s going on and what you think about the consumer, and how you think they work. The only difference with data science, is data science is always asked to prove it with numbers. You know, looking back at data, and looking forward. But, the interesting thing is that all the rest of marketing is following along those lines too. People are being asked to say it’s not just hearts and minds, it’s what numbers … What were the returns? What was our ROI on the spend for these sets of campaigns? So on and so forth.

Michael Krigsman: But how does the AI work? Can you just dig in a little bit and make that AI connection? So, what do you mean by AI in this context? What’s the kind of data that you look at? How does this all …

Andrew Eichenbaum: Sure, yeah. That’s a good question. So, let me give an example: when a baby is born, they don’t understand the idea that something is hot. So, it takes them a while: they’re standing, they can see, say there’s a pot on the stove on the burner, and it’s boiling away. And they don’t know anything so they touch it, and they burn themselves. But, in coming closer and closer, you start getting in those things, “Oh, that could be hot.” Or you start learning, you put your hand a little closer, a little closer, a little closer, and constantly getting warmer and warmer without actually having to touch and hurt yourself. It’s a feedback system.

Most of modern AI is exactly the same thing. It’s what are called “supervised learning systems”. We have large amounts of historical data, where we know the outcome. So, we say we believe to get this outcome, we should look at all of this data beforehand. And so what we do is we have these programs, which we say, “Here’s all of the inputs, here’s what we want,” it’s either good or bad by the definition, or however you want to define your outcome, and then you pass it through the system and it spits out something like, “This is good.” Well you know it was good, so you reinforce and you say, “Oh, and it’s all happy.” And you pass another thing to the system and it says, “This is good,” it’s like, “No, it was bad.” You pretty much slap it and you say, “It should really have been marked bad.” And you do this over and over again. So, it’s learning the exact same way someone would learn that something is hot, or something is cold. You get a better understanding over time from a whole range of input variables.

Sameer Patel: That’s a really important point, right? Because I think there’s so much hype around this topic right now, and one of the things is … And when we talk to customers, this is whether in the sales cycle or the customer success programs, you have to start to have a very open discussion with customers about where the state of AI is today, promise them that you’re kind of at the edge, or the bleeding edge of it, but make no mistake, it’s a constant process of refinement and improvement and improvement, right?

And so, you know, maybe another way to also build on what Andrew said in terms of where the AI come in, there are a couple of places where the AI comes in. And one is, you know, what is the point of all of this? The point of all of this in plain English is how do you begin to move from what I like to call “lazy segmentation and coding,” because the technology just has never been sophisticated, so we kept putting random people into buckets just to make ourselves feel good that these segments actually matter; and starting to move in a direction where you’re starting to really engage and transact with an audience of one.

Now, to do that, you’re going to have to do … There are different places where AI gets injected. Maybe you can talk about it from the standpoint of how do we do it from a [...] standpoint, to understand how you cooperate with the end-user. And maybe the second piece here, Michael, that we’re talking about to where we .... And this is again where … I’ve written a blog post about the things I learned in the first sixty days of coming here, and one of the things I learned, coming from the outside in, was the amount of peanut butter in AI that happens in the industry, right? And you have to also talk about the integrity and the real-time access to data before you apply any AI to it. So one of the things that we do in Kahuna is within five seconds of you making a gesture within a mobile app on your phone, Kahuna will record that [...] a profile; and within thirty seconds we can put out an offer.

If you don’t have that level of immediacy, no amount of peanut buttering … That’s just putting AI on … It’s just making your bad data smarter, in its own way, right? So I don’t know, maybe you just want to talk about those aspects of where in this process you would specifically inject these concepts to actually [...].

Andrew Eichenbaum: So let me touch to both of those points that Sameer brought up. First about the data: Data is good. Data is great. It’s sort of the centerpoint of data science. But, if the data is junk, the data science coming out will be junk as well, so a large piece of any data scientist’s lie for the team is making sure the data’s coming to the system, is properly being stored, planned, verified, so that when we finally get to have fun and play with these advanced systems, we can believe in the results coming out, because if not, why bother doing it?

Michael Krigsman: So, how is this different from traditional marketing? So again, from the techniques that you’re bringing to bear with AI and machine learning, how is it different from the way marketing, digital marketing was handled up until this point?

Andrew Eichenbaum: Yes, so let me give some direct experience. So, I worked at a company called “MyBuys,” it’s now part of Magnetic; this was about eight-ten years ago; and we were providing Amazon-like recommendations of service - various midsize companies. And, one of the things we did there, or I did there, was to figure out what the optimum time to send out to a client base, and that being an entire consumer base, and you can plot it out over time for the entire base [...]. It’s best if we send out the last email at 10AM on Tuesday morning. And so, we backed out, starting a little beforehand until then, and it was all good. And then we were able to see significant lift.

But, now we have the ability to actually do this on an individual basis. We’ve seen an individual come in and respond to messages, or not respond to messages for various channels over the past couple of months. And we know how they respond to what type of message. So, we’re no longer blasting out to the entire group at a single time of the week. We can actually set up our campaigns so that an individual user will be sent out right before the expected time or act, and in a public channel. So this is the real thing. And you say, “Well, that’s a hell of a lot,” when you’re literally giving the orders of magnitude more data to do this. And the other bit is that Moore’s Law has helped us. We have a huge amount more processing power, and it’s not the limiting factor anymore.

Sameer Patel: What is the number? Is that half a billion events today?

Andrew Eichenbaum: Yes, so Kahuna on its own process has half a billion events today. And, it’s literally linear scalable system we can rerun in the cloud and simply add more computerization as we get more data.

Sameer Patel: So I think there’s another way to also look at this. Michael, I know you’ve been a follower of enterprise technology for so long, so if you also start to look at the … If you look at the predecessor to what Kahuna is today, the predecessor to Kahuna were email delivery systems that were built a decade ago. And again, no disrespect to that model, that was state of the art back then. The way to not have to put a flyer in your mailbox, email was a great alternative to do that at scale for much of it, right? So you totally [...].

We’ve now reached a point where the number of engagement and touchpoints for us as consumers has gone from one, which was email, to many. And we haven’t even seen this play out yet, right? So today the dominant ones are email still, and mobile; SMS. We’re going to have beacons tomorrow, we’re going to have IoT after that, we’re going to have chatbots, right? It’s going to go … The places where we engage are going to increase.

The other amazing thing about where we are right now is that every one of those engagement touch points are going to start sending different events to us, that again, email jsut never sends back. So if you do not have a system today that can accept these signals from these different engagement touchpoints, make sense of them, add them to Michael’s profile, but still have the luxury and the [...], if you will, to say, “I may have learned something new by Michael through a gesture on a mobile phone, but the right way to engage with Michael based on machine learning, is 7PM on Thursday nights, via email, because that’s when he seems to be on his laptop and he seems to want to engage and buy stuff.” You have to start as a marketeer, decouple the smarts of the system, and where you want to engage with people. This is where the technology has fundamentally shifted from what were truly nothing more than email delivery machines. You know, batch and blast, batch and blast, batch and blast, right? That allows marketeers to engage in a fundamentally different way.

Michael Krigsman: Andrew, can you drill down a little bit, and give us some concrete examples to kind of maybe walk us through how you look at the data, in order to personalize to that level of the individual that Sameer was just describing, using AI?

Andrew Eichenbaum: Sure. We can … Let me talk about a project we’re working on right now, and it’s a question of whether we should send an email or not. So we set up this batch, you optimize everything, each individual has a send time, but the question is: Should we be interacting with them at all? You can ask yourself the question, do we expect them to interact with this message, or do we even need to send the message because we expect them to interact with the site, the media anyhow, so that the cost of this message isn’t lost because there was no reason to.

I guess the best way to put this is we look at their history. We look at all the ways that they are interacting with the site over the past roughly hundred days. We look at all the different messages that they receive, how they interacted with those messages, or had they not interacted with those messages that we see, but that their site usage has gone up.

There is a bunch of other things going in there, probably a bit too much at this point; it’s trying to understand the expectations of using the site and whether people want to pay the costs of the email, where they say it’s not worth it at this point in time.

MIchael Krigsman: So you have this body of historical data, [on] an individual, about what they’ve done, their open rates on that email, is that correct? And then you’re looking at that historical data and doing analysis on it?

Sameer Patel: So there’s two pieces though, right? There’s the data that Kahuna collects, and then because of how the system is architected, if a customer -  if it’s a commerce example,we can even pull in point of sale data enriching the profile, right? So that’s again the benefit of having built an API first-model of four years ago, and not 20 years ago where the system [...].

Andrew Eichenbaum: So also, remember we’re not looking at just email. We’re looking at email, we’re looking at push notifications, SMS, a whole range. Because when that message was sent out, even before I ask a question - the sort of last-minute question - “Should we send the thing out or not?”, we need to understand what’s the best way to reach out to them. And [after we]look up the message and the time of day, and the requirements from the marketeer, we might want to send out a push notification, which is completely separate.

This comes back to a more interesting question that even the other modern marketers aren’t sort of attacking, in that this is really a derivative problem. It’s not just how many times have you opened your emails and moved forwards? How many push notifications like, “Ok, you that your last two messages were push notifications over this past two and a half weeks, and your interaction for this, and a number of other things, how should we message?” There’s this whole encompassing profile of where you are in the short-term, the mid-term, and the long-term, and how that defines how we should interact with you going forward.

Michael Krigsman: So it’s not just a matter of counting up times. [Laughter]

Andrew Eichenbaum: You can do that, but you don’t get the lists that you do. You’re back to where we were five years ago.

Michael Krigsman: Okay. So now, you’ve got this body of data, you’re making these inferences essentially about individual consumers, the type of content they like, when they like to read it or receive that content, and which channel, and so forth. So we’ve got all of that, and we’re applying these advanced AI and machine learning techniques to it, what do we get as a result? Why are we doing this?

Andrew Eichenbaum: You’re not spamming your consumer, is the easiest result. You’re telling them what they want to know on their timescale. My goal is not to market to people, it’s to make suggestions to people. If they never feel like they’re marketed to, but we’re sending them messages, then I feel that we’ve reached the ultimate goal. We’ve influenced the people without them ever thinking that they were being influenced. They were being passed information that is relevant to them at the point in time that it would be most relevant to them.

Sameer Patel: And that’s a very good point. I think one of the things that, you know, Andrew very astutely said, is being … The rhetoric right now, which is kind of disappointing; and again, because I came from the outside into this world, it’s one of the first things I’ve picked up when I took the role as I tried to understand how the industry was speaking. And we kind of went from this “Let’s drop the email, email’s dead. It’s all about mobile,” and you get to a point where none of us wants to get excessive notifications on our phones. That certainly doesn’t make it correct, because it’s become [...], right?

 So, this notion of being cross-channel at your core, where you can actually engage with people in a way that respects the pace at which they want to go through this journey, we’ve got brands, we’ve got big consumer brands that have massive segments of their customers, who say, “Look, the primary engagement model needs to be email, and mobile once in a while.” That same customer has very lucrative segments of people who want to buy from them that say, “Absolutely not. I need more mobile, some SMS.” We have to let every brand decide what is that right combination for every demographic or user type that you have, and respect the pace at which those consumers want to engage with these brands and transact. Sometimes, it helps the brand figure out, “You know what, it’s going to take four, or five, or six engagement touchpoints before you get to a place where you can get to whatever you define as your goal, be that conversion or what-have-you.”

We can’t assume steep cliffs, nor can you assume gradual rises. You have to start to respect the pace at which you run the most. So these are some of the things we think really hard about, and brands are starting to understand that we can’t switch completely to the other end, and just start hammering people with mobile notifications, right? It’s an interesting problem that I’m not seeing everybody solve, but I feel like we spend an adequate amount of time thinking about it. Do you agree with that?

Andrew Eichenbaum: Yes.

Michael Krigsman: So, when you deliver information that is cued again to the type of content, who it’s being delivered to, tailored to, how they like to receive it, when they like to receive it; in a sense what’s happening is you are bypassing the mental filters that the receiver sets up to block out that we all have, to automatically try to block out the meaningless barrage of marketing noise, that we get all the time.

Sameer Patel: Yeah, and I mean you know, yeah.

Michael Krigsman: Because I care about it. I care about this. I happen to be interested in things like cameras and microphones, and plug-ins for audio, and all that kind of stuff.

Sameer Patel: Of course you are!

Michael Krigsman: And when they … and isn’t everybody? And when I look at websites, and I see these things, I don’t look at it and say, “That’s creepy. I hate that.” I look at it and say, “Oh, that’s interesting.”

Sameer Patel: You know, yeah. I think you nailed it, and I think this is where… You know, unfortunately as consumers, we have normalized in some ways to a world of batch and blast that was created ten and fifteen years ago, where we get an adequate amount of spam and it’s not until I think now, in the last two or three years, where there’s actually enough technology advancement for us to sort of wake up and say, “Why are we dealing with the second grade way of living and getting hit, and getting spammed? And why can’t every commerce retailer be … Why can’t the recommendation engines be as targeted as some of the big brands that have been invented in the last four or five years?” And that to us is success, when we get customers to begin to sort of …

Every e-commerce vendor, every media company, every travel company, all of the industries that we play heavily in; if we can get them to that place, our job’s done.

Andrew Eichenbaum: For me, those banner ads are whatever pushes or emails that you get are great, but I hate them. Like I just purchased this a day ago. I don’t want to be seeing these. These are now the most annoying ads I could ever see. I purchased it, why are you bothering me? And being able to understand that sort of level I think is really the next step for most marketing out there, that even knowing where your best intent is most likely spent.

Sameer Patel: And you know Michael, I think the other interesting thing is: We always talk about, “The world is change and it’s a new world, and it’s a new consumer,” but there are many things that have not changed. Let’s be honest about that, right? There are things that continue, like when we talked about the conversion rates, two or three percent e-commerce conversion rates, this is a problem that every marketeer, and frankly every CEO and consumer brand has been waking up to for decades, right? The amount of money they plow into customer acquisition costs, to drive new customer acquisition, has only gone up. But the amount of focus and available technology, once they’ve acquired that customer, to get from that, from the point where you are now a customer to the first purchase. What is the actual work required to get you from first purchase to second purchase?

We all know, and I think you would attest to this, that as a consumer, forget all the gobbeldygook technology for a second. I would say, in my own buying pattern, if I buy from a given mobile e-commerce app for the third time, my mind starts to get training around certain categories. I associate that brand to certain categories. And I would not go to a search engine. If I’m looking at products within these two or three categories, I will see if this vendor actually has it. And that is something we think really, really deeply about. And I don’t think the retention part of marketing automation technology has advanced ever to a point like it is now, where you can begin to really, really say, “First point of first goal, get the customer to download the app; second goal, get them to make the first purchase; third goal, get them to make the second purchase.” And having that level of discipline, we’re now at a point where you can do that if you use machine learning-based technologies, right? So it’s going to change how we interact in the next decade. We’re just getting started.

Michael Krigsman: Now, what about the flip-side of this, which is the fact that you’re looking so closely at consumer data; and yes, they’re my digital tracks, and they’re not particularly secret, I’m going to these websites; but, when you start aggregating this kind of data, what about the privacy implications?

Andrew Eichenbaum: So, this is a question that comes up in every company that deals with marketing and AI. They say, “Well now that we can draw these correlations between all this data, what does it mean? Are we going to spoof our customers?” I can tell you numerous stories. And what’s really interesting is it’s actually based upon the older the age range, the more worried they are about security and privacy. Say their generic internet traffic. Now there are other places like medical, and banking, where people just don’t want to share. They’re extremely secretive. And, these are where things like personally identifiable information comes in, and making sure that nobody can see these things easily out on the web, or advertised to that specifically. But if you look at …

Heck, just look at social media. How much do people post, and they’re specifically sharing it with the world. Is it that much more than just viewing your … I don’t view my web history trail as any less personal than what I’m putting up there: pictures of my kids, conversations that might be going on… People have become accustomed to a certain amount of sharing, and they understand that that sharing will be used to profile them by multiple people. What the smart companies are doing is they are saying, “Yes, we’re going to do this, and we’re going to make it better for them.” You can see this on Google, on Facebook, it’s like, “Is this ad valid to you? Do you want us to advertise to this?” They are up there saying, “We’re going to profile you based upon your activity. What pieces do you do not, and do want to be marketed to? How do you want us to use this information?”

Michael Krigsman: So basically, opt-in, essentially.

Sameer Patel: I mean, yeah. You’re leaving certain digital breadcrumbs proactively, openly, right? This is no more than, you know, what you have agreed to leave open. Now usually, also understand in our context, because we sit on the retention side, you know. There is an association with a brand already, you have the app, we’re helping you just go through a more meaningful purchase with that particular brand which you’ve always had business with. There’s no denying that this stuff can be used in ways that are not kosher, right? It really boils down to, you know, you having the faith in the brand and the customer success teams, and the people you work with day-in and day-out to … because again, I think every brand knows they’re recovering from brand [...] that’s far more expensive than the initiative use, right? So, stay within the lines of what the consumer has agreed to expose, and you’re okay.

Michael Krigsman: So, maybe this is an important point, that essentially what you’re saying is smart brands will engage in behaviors that will engender trust among consumers. And you started to talk about this, but maybe give us some advice to brands for what are behaviors that will engender trust? How do you engender brand confidence and trust?

Andrew Eichenbaum: So, one of the easiest ways that data science has been viewed as important is explaining why we do what we do. Giving the user, if they ask, the reasons why we’re marketing this to you. And you see this on Netflix, if you go on their streaming. If you go down lower and lower in the recommendations, they say “movies because you watch this-and-that,” films based upon this genre that you seem to like, based upon your viewing history. Just give a brief explanation which seems plausible and at the root of the reason, or the way you were targeting them, and people find it much easier to ingest, and understand and trust those [recommendation].

Sameer Patel: I want to come at this from a different side as well. This is an indirect way, but I truly believe that there is going to be more and more of this over the next coming years. One of the things that is least talked about in this concept of trust between a brand and a consumer. So one is actually the things that we talked about, there’s a level of “Will you screw me over or not?” There is definitely that side. But there’s another discussion too, which is I start to have more trust and faith and respect for a brand, as a consumer, when I see that they have taken the extra effort to understand that. You know, you’re going to … Do you think Amazon breaks your trust when you see five things that they suggest based on your interest? No, I don’t think you - this is again, shrinking that distance between what an offer is, what valuable information is. And I think as we start to do that in a meaningful, respectful way, I’m not going on and telling Michael Krigsman’s friends what Michael’s interests are. I’m sticking to Michael, and only Michael gets to benefit from that. You start to build a mutually respectful relationship with a brand. And I think that implicitly starts to go a long way.

The problem has been the opposite, which is, brands have had to try to get very cute with the consumer because of really crappy, excuse my French, the really, really bad, shitty marketing technology. And that time is done. The consumer’s done, the state of the technology has changed over like “Clap, clap! Wake up! We’re done with all that!” It’s time to move on, right?  And I truly think that better relationship with every single consumer by putting meaningful stuff in front of them, and not screw them over with the data that they give you. That’s important, just to stay there.

Michael Krigsman: So we’ve got about three minutes left. And, how about for our closing, if each one of you just offer your continuing thoughts on what must a brand do to, in the immortal words of Sameer Patel, not screw over their consumers? Andrew, you want to start with that one?

Andrew Eichenbaum: There are so many ways you can screw over your consumer.

Michael Krigsman: … Or to do the right thing. What should brands do, to do the right thing? That’s another way of saying it.

Andrew Eichenbaum: It’s not as much doing the right thing. Marketers want to increase their market share, and it’s inherent in the system. The real piece is not to go above and beyond to the point where people become distrustful of you. It’s really just building up what Sameer said. You can easily go off, given your consumer information, and go to a third party, spend seventy five cents a person, and literally get their full history in there: their employment history, where they’ve lived, so on and so forth. And you can use that and it becomes extremely powerful. But if you start doing that, and people realize you do that, they’re going to go nuts. You’re going to make the front page of the news, and there goes your profits for the next wherever.

Michael Krigsman: And we hear about companies that make the front page of the news because they do that kind of stuff.

Andrew Eichenbaum: But if you keep it at the level that you’re interacting with my site, you’re buying stuff. You are doing this and that, and from that, we are moving you forward, you’re at a much safer space. Now, there are places like Target, which sometimes overreach a little; if somebody gets a congratulations on being pregnant, you might say, “What?” [Laughter] Maybe because they haven’t told anybody yet. But the interesting thing is, you hear more of those stories than the opposite.

Michael Krigsman: So Sameer, you have the last word. We have one minute left.

Sameer Patel: Yeah, look. I think it’s really simple. I think we are at a really critical, and an amazing time in terms of revitalizing and rethinking every piece of the enterprise technology stack and B2C marketing automations, and all different things. Marketers, for the first time, have the ability to start thinking about B2C marketing automation as workflow automation technology, and truly put the consumer in the center of the entire experience process, and figure all the business processes that actually emanate from it. How do I want to engage with an audience of one, Michael verus Andrew versus Sameer, and what are the kinds of campaigns and things I need to be running that have to follow that? This is the opposite of how ten year old technology was designed. That was designed around scaling and batching and blasting stuff out there, consumers are way smarter than that. Unfortunately, or fortunately for you, Netflix and Amazon and AirBnB and all the amazing digital brands have risen.

You know, just the tolerance level of short-e batch-and-blast technology is just not going to cut it anymore. A very simple way to consider this is to say, “Everything I do on a day-to-day basis: Is the consumer at the center of the process, or is the convenience of my marketing automation just at the center of the process, and you will immediately be able to figure out where the gaps are in the technology.”

Michael Krigsman: Okay. What a great summary in the end: Is the consumer at the center of our process, and our life as a seller, as a marketer, or are we doing things because…

Sameer Patel: For our own convenience, right? You know …

Michael Krigsman:

For our convenience.

Sameer Patel: It’s your convenience. It has to be about the consumer, and everything has to surround that.

Michael Krigsman: And on that note, this has been a very fast conversation. We’ve been talking with Sameer Patel, who is the CEO of Kahuna, and Andrew Eichenbaum, who is the lead data scientist at Kahuna Software. I am Michael Krigsman. You have been watching Episode #209 of CXOTalk, talking about the role of AI, artificial intelligence in marketing. Next week, there’s no show because of the New Year’s holiday, and everybody, thank you so much for watching. And Sameer Patel and Andrew Eichenbaum, thank you so much.

Sameer Patel: Thank you for having us.

Andrew Eichenbaum: Thank you.

Michael Krigsman: Everybody, have a great week. Bye-bye.