#279 Customer Strategy in the Age of AI | David Edelman, Harvard Business School Fellow

How are we going to empower customers? Where are we going to get the information? How are we going to make sure it's trusted? When's the right time to reach out to them? How do we make it a personalized experience? And how are we making sure we're getting smarter and better over time? Those are useful questions that can really guide the process of pursuing personalization.

Hi, Dave. Welcome to the show.

Thank you, Richie. Pleasure to be here. Cool.

So just to motivate this, can you give me some examples of where personalization is used and might be useful?

Yeah, sure. Let me actually give you a really good example of where it came together for me, which was actually one of the motivations to say, I better write a book about this. So a few. Just before COVID So this is before Generative AI, I was living in Lexington, Massachusetts, and they offered an incentive for people to put solar panels on their homes. You would get a rebate on your property taxes. And when they did that, the floodgates opened of all kinds of marketing, people throwing flyers under our door, emails, everything. 40% off this, 30% off that. It was really hard to understand how to evaluate any of that. I got one actual physical envelope sent by direct mail saying, we've done the math and at your address, we believe you can save over 20% annually on your energy bills by putting on solar panels with there's a personalized URL in this envelope that can explain how. So I open the envelope, sure enough, there's a URL with our address embedded in it. Type it in. I get a Google Earth image of the roof of my house with solar panels superimposed on the roof, and then a calculation down the side of how much energy that number of solar panels would generate. Also factoring in my longitude and latitude and the north, south, east, west orientation of my roof. So that says how much energy it would generate. Then they went to Zillow, got the square footage of my house from that, and use that to estimate how much energy over the course of a year we would likely use in our home. So they have the numerator and they have the denominator, and it came to 21.3%. And I was impressed. That was quite a personalized, specific thing. And it also focused on what I really cared about, which is saving energy costs and being greener. Then it said, click here to learn more. And I'm immediately put into a video call with the young salesman who greets me, sees, I'm Mr. Edelman, has all my information right there, says, I'm delighted to tell you how you can save over 20%. And then he immediately starts going through the different leasing options in very simple math. So I could see how the economics works. He also was armed with two email addresses of neighbors of mine who had used them to serve as references. And the economics made sense, the references made sense. So we did it. We didn't even look at a competitor. And then from that point on, it was all managed through an app. So the app we used to schedule when they would come out to track how much energy. When a squirrel ate one of the wires, I got a notification and we scheduled for them to come out and fix it. When I was over producing and selling back into the grid, I could see that. And so it was all seamless. Solar panels are a commodity, but the experience around them matters a lot and can be quite complex. That's personalization. That's creating value for me as a customer by using information about me in a productive way.

I love that because it involves so many different facets there. So you've got things like thinking about what does the customer actually want. You've got saving time for the customer because they didn't ask you a lot of stupid questions, which they could have looked up. And then it's making use of a bit of data and some software there just to figure out, like, well, you know, what is your house and what savings could you actually get there? So very cool stuff there. That's a great example. So I guess we're going to figure out how you go about implementing that in your own business now. And I know in your book you have a framework of sort of five principles for personalization. Can you just give me an overview of what those five principles are?

Yeah, sure. So when you're thinking about doing something like what Sungevity did, you've got to. You're implicitly making promises to the customer, you're promising to the customer that first and foremost, you're going to empower them. You're going to enable them to do something they could not do before because of the way you're using information. In my case, it was figuring out how the heck to think about the economics of solar panels, which were actually quite complex. So you got to empower someone, which means really thinking about what does a customer need? What are solutions for them, not just how you can manipulate them into buying what's the solution for them? So empower them. The second is know them, have data about them from a source that the customer feels is going to be appropriate. Now Some people might have thought what Longevity did was creepy, that it got all this information. I found it very clear that that information's out there. I thought it was a terrific use of it. I did not think about that as creepy at all. And their growth and success led me to believe most people did not think that was creepy. So we can get into different kinds of information and gathering information. I think we will, Richie. But first you have to know me in a way that I'm going to be comfortable. Then you've got to reach me. You've got to figure out when and how to contact me or to make yourself available. And the danger today with so much AI and a lot of the personalization stuff that's more tactical is that companies are going to create all kinds of content and just bombard the hell out of you. And that's not appropriate. Personalization is not. So, for example, we just moved. I bought an espresso. I treated myself to a Nespresso coffee maker. It comes with a hundred pods for the coffee machine. Within two days after I got the machine, I am getting bombarded by Nespresso with offers to buy more pots. I just got 100 pods two days ago. How much coffee do they think I drink? They may think I'm a small business, but it's a residential address. They should know these things. And they could even ask me. And then a week or so later, they're sending me offers to buy a second machine. And this was just ridiculous. So I just shut the emails down. I just unsubscribed. And, you know, that's an example of pseudo personalization gone bad. That's just simply bad targeting. So you have to reach someone appropriately, then you have to show them. You have to give them content and experience and offer in a way that's going to resonate. And of course, in the Longevity example, the way they showed the roof of my house with the solar panels superimposed with the simple math, that was all a personalized content experience, which now actually generative AI makes a lot easier and then delight me. So as you get more and more information, make it better. And this is something certainly you see with Netflix, with Spotify, as it looks at what you do and you get thumbs up, thumbs down, it can tailor what it offers to you. Netflix, for example, makes over 100 different trailers for most of their. Most of the films and shows that they create. And all of that is based on the information that they get from individuals. So empower me, know me, reach me, show me, delight me. It's a good framework and I, I do a lot of independent consulting with companies who are starting to get into this and using that as a structure, asking the questions, how are we going to empower customers? Where are we going to get the information? How are we going to make sure it's trusted? When's the right time to reach out to them? How do we make it a personalized experience? And how are we making sure we're getting smarter and better over time? Those are useful questions that can really guide the process of pursuing personalization.

That's very cool. I like that it's broken down into different stages of the customer journey. So I guess, do you always start with the first one with Empower me? Is that the place to begin if you're just starting to think about personalization initiatives?

Yeah, it is. So you have to start with Empower me because that's the goal, business wise, is to create an experience for a customer that's going to give them value, that then presumably gives value back to your business. And then all the other stuff can focus on enabling that versus just for example, if you just pursued know me like, let's get all the data we can on customers. Oh my God, you could spend months and months and months collecting all kinds of data towards what end? So focus it, focus it so that you can get business impact. I'll give you a really good example of this. So Cisco, the food delivery company, S Y S C O, not the tech company, so they have trucks all over the place that are delivering food to dining establishments. And they created an app for people to use to order food for their establishment. Now they could have just simply had a really good app that makes it easy to find things and order things and just focus on that. But they felt that these are often people who are so busy they don't want to spend a lot of time ordering. They also want to make sure they're getting really good value. So Cisco brought AI and personalization into the picture so that when you order, when you open that app in 300 milliseconds, it knows who you are, your eating establishment, your menu, your price points, what you've ordered before, whether you order in bulk or on spot, where you are geographically, what's the nearest warehouse to you that services you, what do we have in that warehouse and what might we have? Because a lot of this is food that we want to move and discount. And if we want to offer those discounts, can we give recommendations for menu items to the buyer that could take advantage of those discounts and all of that happens with a completely personalized interface in 300 milliseconds. So that's. And so we. I've interviewed several of the folks who use that application, and they say it completely saves them time. It makes it ridiculously easy for them to order. They get new ideas, they get ways to take advantage of price promotions, and that creates a bond that's really hard for another distributor to break.

That's also a very cool example. So just from, well, I guess my personal experience, so I'm vegetarian and like, trying to order food from a restaurant, it's like, well, just how do I hide the meat options? I just wanted a personalized experience where I'm going to find things that I might actually want to buy. And so few businesses do this. So I can certainly see how having that AI sort of information about, well, I guess just what your past purchases are, that's going to be really, really a simple way to encourage people to buy what they actually want.

Yeah. And actually, let me start, because this is a discussion focused on a data community. And there are some interesting things, actually, if we break down the Cisco and even longevity to some extent, but even more Cisco of some of the things that are going on with data there. So think about. For that app, Cisco is pulling in all the data on what you have ordered. It's pulling in product data, it's pulling in warehouse data, all of that into one. Now, often integrating all of the data from different silos around a company is not an easy thing to do. And companies spend millions of dollars on consultants and outside folks who can spend the time data engineers bringing that together. AI now can make that dramatically easier. Generative AI writes code, and we don't talk about that enough. Everyone's obsessed with how it creates content and generates answers, but it can also write code. And for example, there's a company called Narrative AI that offers a specific capability called Rosetta, as in the Rosetta stone, that looks at one data set, understands its schema, looks at a second data set, understands that schema, and writes the code to combine those databases, to normalize the data, to dedupe the data and bring that all into a third repository, which could be, for example, Snowflake, which uses narrative on the front end to bring data together. And so for Cisco, using AI to bring that data together was a critical step in making that happen, and it required thinking through new ways of managing data and taking advantage of AI capabilities.

Yeah, that's very cool. And certainly data engineering, dealing with all these sort of different Data setups, that's a sort of persistent problem for a lot of companies. And I do like that AI is making that easier. So since you mentioned data, we talked about some of the data types that you might want to use. So for example, previous purchase history and in the solar panel example you gave, like the location of your house and things like that in general, are there any types of data that you might want to know about your users in order to provide these personalized experiences?

Yeah, one thing that companies don't do enough of is just simply ask people, like the Nespresso example I gave earlier, you know, if they couldn't look at my address and see it's a residence, can they at least just ask me, you know, ask me what? You know, are you a residence? How often, how many pods per week do you think you're going to need, you know, and set up maybe a subscription or something. But that's called zero party data where you are asking somebody directly, you're not inferring it from actions they've taken or anything like that, you're just simply asking them. It's like when Netflix says thumbs up, thumbs down, that's zero party data. So that's a whole area which is trusted because the customer knows they're giving it to you, is something that can be quite timely and accurate and it shows intent often as opposed to just factual demographics or something like that. So that's an area in particular that companies who are pursuing personalization, well, are spending more time, for example, a bank asking you, what's your financial plan, what are your goals, what are you saving towards, how do you think about budgeting and then providing you with notifications and ideas that help you manage towards that budget, that getting that budget, that's zero party data. So that's one important thing. Second thing that's happening is the reduction in availability of third party data. So third party data is where somebody else is doing something and they are offering you that data to use. And this is classically true in terms of online marketing, where data that was picked up by cookies because of the way you're browsing the web could be used to target you. But increasingly that data is not available. Most Apple users now are going for private mode and the data is not there. So you need other ways to get that data. One of the ways that's emerging is what I would call second party data, which is where somebody like Home Depot says we have all of these interactions with customers and we understand what they're buying, who may have recently moved and is now doing renovations. And you can use our data, but we're not going to give it to you. You can use our data to target marketing. And these are called retail media networks. And this is happening. Retailers are offering it. Banks are starting to offer this because they have such rich data. And especially for companies like in consumer goods where they don't have a lot of data, using the data of a retailer or a bank to target customers is going to become one of the few ways to get access. And these retail media networks are growing enormously fast. Home Depot is making over half a billion dollars in incremental revenue. And that's almost all to the bottom line from this. It's a whole new area of business for them. But that's second party data. So in the book we talk about zero first, second, third party data, how to collect it, how to use it, the different trade offs in terms of prioritization and access and privacy. But that's a very important area to have a command of. As you're thinking through the know me strategy.

Okay, yeah, it certainly seems very important to distinguish these different types of data. And it is interesting how there is an increased focus among consumers about privacy and maybe they don't. Not many people want every company to know everything about them. So yeah, you need to rely less on third party data. There's so yeah, certainly the idea of first party data where you just simply ask, ask a customer what they want or what, what they need or what you need to know about them, that was news to me. So yeah, I think surveys are underrated as technology. So yeah, that's very cool. All right, so in terms of know me then, are there any more suggestions you have for like what are some good ways to get to know customers and how might you implement a personalization strategy from them?

Yeah, so I think one of the most important things is getting customers to interact with you more through asking questions, through getting them engaged to take actions on your site. Because first party data is absolutely critical as well. So the data on interactions that somebody is having with you is really valuable. The problem for a lot of companies though is that the data that you get from a customer is all over the place. As I was starting to reference in the Cisco example, you may have data, product use data, call center data, billing data, marketing data, sales and e commerce data. Those could all be. And for many companies they are in completely separate silos and they're not easy to be. It's not easy to bring them together. And so from a data perspective, having all of those interactions is only as good as your ability to activate all of that data. So there are solutions, like I mentioned, Narativ and their Rosetta product before, to bring things together. And there are other tools now that are starting to not only bring the data together, but provide insights about that data. So give you another example of this. There was a company called Pointillist. They've now been bought by Genesis, the call center software company. And they have a tool called Journey Analyzer, Customer Journey Analyzer. And what it does is it looks across all the different touch points that you have with a customer, and it says, richie tried to use the mobile app, then tried to use our product, he tried to pay a bill, it didn't work. And then now he's calling into the call center. And so you've had a journey that touched four different things that isn't going well for some reason, and now you're calling into the call center. So what the AI does is it makes the match of Richie across all of those databases, strings together your journey, timestamps it so we can understand that journey. And so when you call into the call center, that call center rep knows you were just on the app trying to pay your bill and couldn't immediately. The rep knows that. And if that's a common problem, they may already have a fix for that, that the rep gets a prompt and tells you. And probably now with generative AI, you might not even need to go to a rep. They may just simply say through a chatbot, well, you just trying to pay your bill online, you didn't have it, here's the fix. And it's all done automatedly. But if it's something that's been done, the rep can get. Even if it's something that hasn't been a common problem, at least the rep immediately knows what your challenges have been. And so this was something, for example, Comcast, when they launched Xfinity Mobile, their challenger in the mobile phone business, they used this capability to constantly manage and stay on top of all of their customer service issues. And, and they ended up winning the J.D. powers Award for best wireless service provider for three years in a row. And so that's thinking about all that data and activating it in a very deliberate way.

That's very cool. So I was just thinking about, like, I guess the end goal of any personalization is you want to get that fifth. That fifth principle is about delighting the customer. I'm trying to work out how you go about measuring that, but it seems like customer service or customer Support is one good way of doing that. So if you are getting people, if you're solving their problems on the first call, rather than having to have three calls with the customer service rep, then that's a big improvement. Can you maybe talk me through some other ways in which you might be able to measure, like, the delight of your customers?

So how do you measure the delight side of how you get better? And so a lot of that is about getting better interactions with customers and seeing better engagement. So a good example of this, Brinks Home Security. So they have alarm systems that they sell. It's all on a subscription basis. And every year, people, they would like people to renew their subscriptions. And historically, they have found that in order to get people to renew, they would need to give them a discount. So, 25% off in order to renew. And that was generally effective but costly. So what they did was use a tool called Offer Fit, which allows them to do massive multivariate testing of all kinds of ideas to figure out the best way to get Richie versus David to renew. And so it could be the offer, it could be the time of day, it could be the way we send it, it could be the visual, all different kinds of things, the wording. And what Offer Fit does is allow you to set up test cells to try multiple variables instead of a simple a B split test. And so for renewals, they started using this to test many different variables. And they started learning things such as people who had used their alarm system actually were happy that they're alarmed. And especially if they had an incident where the alarm system helped them, they wouldn't need a 25% discount. So you can bring in that data and you can remind somebody of the incident that they had and how Brinks helped them, and you can offer a dramatically, if any, discount in order to get them to renew. Plus, people are feeling better because they're reminded about the value of their alarm system. You can also look at people who were just using the alarm system for certain things, but not for others. And you can inform them about capabilities that they could take advantage of in their alarm system that they hadn't before, and that creates value for them and so on. And so they did that and came up with a whole range of different ways of connecting with specific customers and, and ended up getting a 400% ROI on the changes they made to their renewal by doing all of this test and learn, test and learn, test and learn. And from that, figuring out that, you know, for Richie, who uses his alarm regularly, it's this message for David, who never really bothers putting on his alarm, reminding him that there's actually been some crime in the area and he better start using his alarm. And he's got a great alarm system. So all of that requires data bringing that together. It requires nonstop test and learn and you can start to see the better response rates that you get as you improve over time.

That is very cool. I think a lot of companies, they don't necessarily experiment that much with their offices. It's like, well, I guess it's sales season, let's give a standard discount to everyone and just by using data and trying to understand your customers motivation. So I guess it's both sides of things. It's data and, and the business side of things. That way you're going to get a lot more money. And yeah, certainly 400% ROI is not to be sniffed at. I guess one of the interesting things you mentioned though is like you're saying that you have to go beyond just a B testing. You need to do a lot of different experiments in this area. Can you just talk me through how you go about this? Like do you start with simple a B test and then gradually get more complicated or how do you implement this?

So let's just also start from the premise that AI eats data. So AI needs constant feeds of data to get smarter. That is how it becomes smart. And the more variants you can feed AI, the smarter it can get because it starts to learn the differences between Richie versus David in terms of our interest. So in order to do that, you've got to test and learn, test and learn. So when I was chief marketing officer at Aetna, one of the programs that we started was trying to get people to take healthier actions, things that would improve their health. And if we did that, they would save money, we would save money, they're healthier. It's all a win. The challenge is I may know the healthy action that I want Richie to take, but boy, I don't know how to get you to do it. And what's going to motivate Richie versus what's going to motivate David could be very different. And so you've got to try a lot of different things. And so we set up what we called a pod that operated in a similar way to Agile software development. So if you're a data person, you're probably familiar with the principles of agile. Two week sprints, really structured with a backlog and staying on top of what you're doing. So we applied this to marketing of Getting a cross functional team of six people, creative, a strategist, analytics, operations. We actually needed compliance in there as well, given that we were a health care company. And so together they are coming up with ideas and they're saying, okay, we want to motivate people who, we know, our diabetics who have not had their levels checked in over three months to go in and get their levels checked, to get actual, get a blood test and okay, so we know, and we know who those people are. So we've got to try different things. And so you do gradually start first with some simple split tests of saying, well, what if first we just simply appeal to people from a health perspective and then maybe try something that has more to do with their families and the value to their families of them being healthy. And you start with that and then you start layering on more and more variations. So as we, are there other motivators? We couldn't do any offers. So with healthcare we couldn't do offers. But we started thinking about creative renderings, the language that we used. And this was before offer fit was in existence. So we had our own crude multivariate testing tool, but nonetheless we did. And we started testing four, five different variables at a given time. And over the course of a year we had over a billion different impressions that we had made from different people and started using that data to be much more sharper and personalized. And over 50 different actions that we wanted people to take, we were running programs for. And that led to dramatic increases in people actually taking the actions. And that led to cost savings in the Medicare side. It led to improved STARS performance. So it's a win win. And we gradually expanded that over time. So a lot of it comes from setting up, it's an operational question of setting up a team who ready to do fast cycle. If it takes you 12 weeks to get an email out the door, forget it, don't bother. But if it can take you two weeks to get it out and you can even get five different tests out in two weeks, that's a whole new ball game. And the faster you can learn, the faster the smarter your AI gets and the better precision that you can have.

Okay, that does seem very important that you're probably not going to get this right on the first go. You do need that iterative process to just keep experimenting and testing, trying new things and just learning from your mistakes. It seems perhaps in healthcare this is particularly important. So I guess with your first example about the solar panels, it's like, well, okay, if you don't personalize it quite right, then whatever. But with healthcare data, there's a lot of like, very sensitive personal information in there and if you do something wrong, then there's going to be consequences. Can you just talk me through, I guess, the trade off between personalization and privacy and what happens if you get things wrong?

First off, let me just start with some data. So I wrote the book together with the head of the marketing department, marketing practice at bcg, Mark Abraham. And BCG does an annual survey of consumers attitudes towards businesses using information about them. And the Latest results say 70% of people, and this is across all ages on average, are comfortable and appreciate doing business with companies who use their data appropriately. And in fact, most of them expect companies to have data about them that they use appropriately. But on the flip side, 75% of them have said they stopped doing business with companies who they felt use their information inappropriately. So there is a bar there, people are comfortable with it. If so, the whole issue is appropriately, trustworthily. That's really the challenge. So now how do you do that? How do you make sure that happens? So there's three things that we've kept in mind. One is you have to look at it from the customer's perspective. So you are a customer, you receive some kind of interaction. What are you thinking about the data that was behind this? And to me, do you feel it is manipulative? Is it appropriate? So you have to, the same way that I talked about empowerment, you have to think about adding value to the customer. You have to take the customer's perspective on what they're going to see. Second is you got to have some guidelines on what is and what isn't appropriate. So for example, in health care, you can't just simply send an email to somebody that has all this information in it because the email could be intercepted, somebody else could see the email. So the emails have to be at a pretty high level with a link that then goes to an authenticated page where with identity you can get more information that's a bit more cumbersome. Not everybody clicks through. But look, it's just what you have to do to manage privacy. And the third is to absolutely stay on top of unsubscribe rates and anything where you're getting pushback or calls and what's happening in terms of people just tuning you out. So when I came in as the CMO of Aetna, I also got Aetna health insurance. So I was a customer at the same time. And within six weeks I saw my inbox just bombarded with all kinds of stuff and why am I getting these things? Are they even appropriate at all? And so I said we have to look at what's going on with our customers in terms of their willingness to interact with us. So we did a very simple chart saying how often have we touched somebody and what's the rate of them actually engaging and opening up the emails. And you saw a curve that after more than four emails a month, it dramatically fell off. People were seeing us as just simply spam and and while they may not subscribe, they just hit delete all the time. So we had to get on top of data like that. So thinking about the customer putting in guardrails, looking at the data, gotta stay on top of all of that with a real customer frame of mind.

That does seem incredibly important. And I have to say email in general seems to be done very badly by most companies. It is just a blanket email to everyone and you almost always don't care. So not clicking through becomes the norm. All right, so I guess some sort of segmentation is needed here. So how do you go about personalizing your emails then? Can you give us some advice on how you get this personalized email strategy?

Well, I think a lot of it has to do with being smart about why you're actually contacting somebody with when you're contacting somebody and how it renders. So again, are you empowering people? Are you reaching them appropriately? Are you showing them something that's going to be valuable to them from an interaction perspective? So all three of those, you really have to challenge yourself and say, is this adding value? The problem that we have is in a lot of companies is just volume based people just hitting the emails. And now with AI automating more, creating content, it's going to be easy to just hit the button and just get stuff out the door. And that's going to lead to a tragedy of the comments. I mean you look at the data, 75% of people saying they've stopped doing business with brands who use their information appropriately. You look at my graph that I did and say after four emails a month, people just completely tuned now, now people, you know, they go the blunt way of just sending stuff so less is more and the more you can pull back. Be smarter, use information to figure out the right time and the right way. So think about the subscription renewals that I talked about earlier with Brinks where they're sending an email for you to resubscribe. But it's first of all it's timed about when you would subscribe so they're not bombarding you all along. And they're doing it in a way that knows you. They know that you use your alarm, you had an incident and they helped you and you got value out of that. And so now here they give you some tips on other things you can do to get most value. And by the way, click here to renew. So that's a very different experience than just simply saying time to renew, 25% off. It's a totally different attitude that's recognizing you as a customer. AI can help you do the right thing. At the same time, it can also help you just simply bombard. So think about the ways that you can connect with somebody and make sure that you're seeing it as an experience versus just simply a push to get somebody manipulated to act.

Okay, I like that. Is think about the holistic experience of the customer. Okay, so you mentioned the idea of a subscription business asking you to renew. I'm wondering are the different personalization use cases for subscriptions versus one off businesses or B2C versus B2B or any industry specific use cases like what do you need to think about in order to I guess personalize your personalization initiative?

Yeah, I mean like I said, subscription business businesses where you have first party information of some kind because you are directly interacting with the customer are the ones that have the most potential for personalization. If you're a pack consumer packaged goods company, it's really hard. You're selling through channels, you don't have a lot of information about your customers. There are services here and there and loyalty program things that try to help you get information. But the reality is it is hard, it is costly, it's not going to be as easy as people who have first party. I think a lot of the consumer goods companies, anyone who sells as an oem, OEM through a third party is going to use things like the emerging retail digital media networks that I mentioned earlier to try to target people using the channel's data because they don't have it themselves. But where you do have data, there's tremendous opportunity to start doing things. And you know, there's, there's so many different variations, you know, from Cisco, the food delivery company where they have a direct, essentially it's B2B but it's E commerce. And so they can totally personalize that interface. But I'll give you a different example of the other Cisco, the technology company who they have salespeople who call on accounts. Cisco has over A hundred products. And it's just too much for a salesperson to comprehend. Which product do I talk about to which customer? And so what they would do is whenever Cisco came out with a new product and a new announcement, they would send that to their customers. But the reality is most of the new products that Cisco comes out with are really not relevant for a lot of customers. They're relevant for a few. And so customers were just getting hit with all these random product announcements. Salespeople were not getting an opportunity to talk to their customers because customers weren't responding and they were feeling challenged. So Cisco stepped back and said, we have to make this more, more personalized for the rep and for the customer. So they created essentially a CRM on steroids. They took their basic CRM system and added more data. They added data about how customers were actually using their products. Was it trending up, trending down? Which features were they using? How many people in the company were actually using it? They took outside data. Is the company overall growing, shrinking, involved in a merger, expanding into new geographies? They looked at what was all the information that anyone in that company was looking at on the website. Were they looking for help, information, whatever? And through all of that, created an intelligence system where every rep on a single pane of glass can get a weekly call sheet. Here's the customers you should touch, here's the content you should use, and here's why, what the logic was. And so with that, the salesforce could be dramatically smarter about who they contacted with what content. Response rates started going way up. They were getting more meetings, they were doing more cross sell. Salespeople were dramatically happier and it was a win win across the board. So that's another completely other form of personalization where it's through a rep to a customer. So there's lots of opportunities, different ways of thinking about this, but it all comes from thinking, who do you want to empower? Where is there a compromise that a customer or maybe even your channel is facing where using better data, activating data about that customer can be used to improve their experience?

That just seemed like a really amazing use case. I'm sure like in a lot of companies it's like, well, the salesperson has a name, job title, company title, and they're just cold calling, going, hi, would you like to buy this? And actually having all that information, well, this is what the customer actually wants and these are the specific products they might be interested in that's going to dramatically increase the close rate. Wonderful. Okay, so just to wrap Up. What are you most excited about in the world of personalization?

Yeah, I'm most excited about companies not bombarding me, companies being smarter, creating experiences that are going to add value. And I think there's going to be as agents in AI, what they call now agentic AI, where because AI can write code, it can write code to take actions and do things on your behalf. Now again, there's issues of trust and will you let things do it on your behalf? But imagine this scenario from Home Depot. Since I just moved, I want to renovate my bathroom. So I take a picture of the current bathroom. Home Depot is able to figure out the dimensions of everything and can see where the pipes are and all of that from that picture. And I tell them my budget, I give them maybe some pictures of styles that I like and then I say give me five options within my budget for a bathroom remodel and how much will it cost and can you coordinate making that happen? That is coming. They are working on that. There are a whole lot of things that have to come together, including the fact of Home Depot coordinating with other parties to make that personalization possible. The installers, the product manufacturers, all of that ecosystem has to be coordinated. But AI now can help them control the flow of data across parties. So I just see more personal solutions being offered and that's going to create opportunities for brands to have really powerful relationships. And I think companies are going to vie to be the hub for solutions in an area because they can give much more personalized solutions to you. So I think looking forward, there are interesting strategies of are you going to be the center of the solution? Are you going to be a player in somebody else's solution? How are you going to manage the information flows? But this is happening and these companies are working on it.

That's cool. So certainly I feel like shopping is hard work sometimes. I know maybe I was a terrible consumer, but yeah, having technology make shopping easier for you, that feels like a win win both for consumers and for the businesses that are trying to sell stuff. So that is indeed a very exciting thing. All right, super. Thank you so much for your time, Dave.

Oh, sure, my pleasure. And please pick up the book Personalized Customer Strategy in the age of AI. And I'm accessible through LinkedIn and would love to hear your stories of things that you're doing. Always on the hunt for great case examples. Thanks, Richie.