How to Get and Keep Customers with AI and Personalization

Michael Krigsman: Welcome to CXO Talk 857. I'm Michael Krigsman, and we are exploring AI and personalization with a world expert, David Edelman. David's new book is called Personalized: Customer Strategy in the Age of AI. He teaches at the Harvard Business School, is a senior advisor at the Boston Consulting Group (BCG), was chief marketing officer at the insurance giant Aetna, and was a partner at McKinsey. And he's done a lot more than that.

David Edelman: I've been playing around in this space since 1989. Back then, before the internet, I was a young consultant at BCG. Three clients had asked us in a row, just coincidentally, "What can we start doing with customer information that we are collecting?" We talked to them about loyalty programs, targeted marketing, and new ways of doing customer service.

I thought there was something more to this, and that this was only going to grow. I came up with the term "segment of one marketing." I wrote an article about it, and it really took off; it became the launch pad for a big chunk of my career. As the internet came in and brought new capabilities, I was working very closely with clients on how to get more value out of the potential of customer information.

And of course, over time, I started to see how the nature of that value changed, but also saw practices that were, at best, creepy and inappropriate. Now we're at a point where AI can enable incredible opportunities, but can also completely turn people off.

So, it seemed like the right time to talk about defining what we really mean by high-quality, good, real personalization. I also wanted to talk about what it takes to make it something that really becomes a competitive differentiator, as opposed to just a marketing tactic. That's what the book really focuses on: elevating this to something that becomes your value proposition. The book provides lots of examples and a framework for doing that.

It's important to think about this from that perspective versus just using data to manipulate people.

Michael Krigsman: It's a very interesting perspective. For you, then, personalization and AI are the core of the relationship that an organization or brand has with its customers. It's not just a veneer to manipulate emotions and get people to buy more stuff.

David Edelman: That's right. Think about the brands that you, as a consumer today, tend to do more business with, especially some of the digital native brands: Spotify for music, Netflix for movies, Uber for transportation. A lot of the value you get from them is because of how they use information about you, information with which you are generally comfortable, and which, in many cases, you deliberately gave them. That information makes your experience better, faster, possibly cheaper, and it could also be more joyful, leading to discovering new things.

This is now becoming a basis of competition, not just among digital native brands, but in more traditional sectors as well.

Let me give you an example, Michael, of how, in a dull area, this can really make a difference. A few years ago, before generative AI became a big thing, I was living in Lexington, Massachusetts. They were offering an incentive for people to put solar panels on their homes; you could get a rebate on some of your property taxes. This opened the floodgates for all kinds of marketing, with everyone trying to pitch solar panels: "25% off this," "40% off that."

But buying solar panels has some complexity. You have the cost of the installation, the cost of the panels. Then, if you generate too much energy, you have to sell it back. There are tax incentives. So, it's kind of confusing. I received a piece of mail that said, "Edelman household, we've done the math at your address,"—it included the address—"and we believe you can save over 20% on your annual energy bills"—which is the number that matters to me—"with Sungevity. There's a personalized URL in this envelope that will explain how.”

I opened the envelope, typed in the URL, and immediately got a Google Earth image of the roof of my house with solar panels superimposed on it. There was also a calculation on the side showing how much energy that number of panels, at my longitude and latitude, with the east-west orientation of my roof, would generate.

Then, they used Zillow to get the square footage of my home and used that to estimate how much energy I would use over the course of the year. They had the numerator from the Google Earth calculation and the denominator from the Zillow calculation. It came out to 21.3%. Then, there was a "click here to learn more" button. I clicked, and I was immediately put into a video call with a young salesman.

He saw that I was Mr. Edelman and greeted me. He had all the information about me right there. I didn't have to explain anything. He immediately started to explain the economics, the leasing options, and how it works. He was already armed with two email addresses of neighbors of mine who were already working with them.

The economics made sense to me. The references from my two neighbors—whom I actually didn't know, so I got to meet them this way—were all positive, so we did it. We didn't even look at a competitor.

They made the process so easy. From that point on, it was all managed through an app: scheduling the installation, monitoring how much energy was coming through. When a squirrel ate one of the wires, I got a notification and used the app to schedule a repair. It was all seamless. That's personalization. That was adding value, empowering me to understand how to buy something. They used data that I realized was appropriate for them to have. It wasn't creepy, and it was adding value to me. It made a huge difference in helping me navigate the process.

Michael Krigsman: How is that different from 20 years ago, where a diligent salesperson, like a car salesman, would do the same thing? What's new today? Why should we care, especially today?

David Edelman: The reason today is different is because we can do this at scale. We can also unlock information that a car salesman probably wouldn't be able to get. Maybe it's easier with a car, but for other categories, you can unlock more information.

If you think about the Sungevity example, there's a lot of different kinds of AI, math, and calculations happening that they're stringing together and doing this on a very broad basis. They essentially did a Google Earth scan of every home in Lexington, Massachusetts, looking for houses with a minimum of 20% savings. They then used that to target those homes and generate this kind of experience.

There's modeling involved, being able to interpret the Google Earth data, render the images. I talked with the CMO, and they use a "test and learn, test and learn" approach, which they did for about 18 months to figure out how to get that experience right. A big part of this is being able to do it at scale. That's the positive.

The flip side is that, because you can do stuff like this at scale, you can also do things that are more creepy, like providing—or actually bombarding—people with marketing because you can, because it's easier to do at scale. You might slightly personalize it, but that's the risk we run. A big part of what we want to advocate for and talk through, using examples in the book, is how to do this appropriately so you're actually building trust.

Michael Krigsman: In other words, you can use AI, the power of AI, for good and build your reputation. However, you can also, in many cases, destroy your business and your reputation inadvertently because you can use AI so easily at scale.

David Edelman: I don't know how likely complete business destruction is, but you can certainly turn people off and dramatically increase your unsubscribe rates. Here’s a simple example: I recently moved from Lexington to Cambridge, Massachusetts, and I was bombarded with emails from Home Depot—absolutely bombarded—with all different kinds of emails from all different departments.

I was ready to unsubscribe, and then they asked, "What's your next project?" I thought, "Okay, this has promise. They're asking me, and maybe they'll actually use that information."

But then they sent a lot of stuff about bathroom renovation that didn't seem very well organized. They could have asked, "What's your budget? What are you interested in doing?" They could even, these days, have me take a picture of my current bathroom and use that to understand the dimensions, ask me what I want to upgrade, connect me with a designer, or show me options. There are a lot of capabilities and tools out there, but it takes focus and discipline to legitimately add value, versus just bombarding people with generic marketing blasts.

Michael Krigsman: I think the benefit is clear, as you were just describing: the benefit of being able to do this in an automated way, at scale. So, the question then comes up: how do we go about accomplishing this goal?

David Edelman: In the book, we lay out a framework of five different promises. Essentially, if you're using my information to give me value, you're implicitly making five promises. This is a good way to organize your thinking about how to design the experiences, capabilities, and processes you'll need to execute.

The first of the five promises, and the most important, is empower me. Help me to do something I could not do before, like figure out how to buy solar panels and know what's really right for me. Empower me. Help me do something that adds value. But in order to do that, you have to know me.

You have to have information that I, as a customer, am comfortable with you having. Maybe you just asked me for it, or maybe it's from public records—I'm pretty comfortable that those exist. It could be from my daily interactions with you. But, know me. There are a lot of AI tools now that can help you know me even better.

These tools can take the transcript of a call I have with a call center rep, digitize it, and use it to understand where I might be having problems using your product, for example. You can also use AI to stitch together lots of different data points about me and integrate data in ways that weren't possible before. So we have: empower me, know me. Once you know me, reach me appropriately. Don't bombard me. Orchestrate the right time to reach out.

And again, AI systems are very good at predictive modeling. They can identify triggers and determine the right time to send some type of outreach. If you have multiple outreach communications, these systems can also coordinate which are the best ones to send and when.

Then, when you reach me, show me. Make it easy for me to understand. Give me a personalized experience, like the image of my roof with the solar panels on it. Generative AI makes this incredibly easier to do.

One great example is from when I was chief marketing officer at Aetna. People didn't understand their health insurance, which was a problem, as it led to calls to the call center and using out-of-network services. We worked with a company called SundaySky to create personalized videos. They were simple animations, not very sophisticated, but they were personalized.

We could tell you, Michael, "This is the health insurance you bought. Who in your family is covered? Do they have primary care relationships? If not, here are some nearby doctors. Click to make an appointment. They're taking new patients."

That was a home run. People finally understood their health insurance. So we have: empower me, know me, reach me, show me. And then, delight me. As you get more information from me, from our interactions, make the experience better.

And that certainly is what happens with services like Spotify and Netflix. You get better recommendations over time. These five are what we call the five promises of personalization.

As I laid out, each promise has different capabilities behind it, all of which are increasingly available. But you have to deliberately decide that this is the direction you want to go in and that you're going to put those capabilities together to add value to your customers.

Michael Krigsman: Please subscribe to our newsletter and subscribe to our YouTube channel. Check out cxotalk.com. We really do have extraordinary guests coming up, so check it out.

Let's take a few questions in the order they came in. This is from Chris Peterson on Twitter. Chris asks, "In the solar panel example, did the neighbors opt in to be reference customers and have their contact information shared?" So, it's a privacy question.

David Edelman: The answer is yes. They weren't surprised when I reached out to them. Once I became a customer, I was asked the same thing: whether I would serve as a reference. By the way, I referred them to my sister, who ended up using them as well.

Michael Krigsman: We have an interesting question on LinkedIn from Anne-Mara Potts. She asks, "Why are we stuck in the bombardment phase?" She thinks it's an organizational challenge. Is that true in your view?

She goes on: "In other words, data is held in long-standing silos within the organization. Customer service data isn't accessible by anybody," and so forth. "Why are we stuck there?"

David Edelman: Take a typical retailer or financial services company—any organization that sells several different products. Each of these products has someone in charge, trying to drive profit and loss (P&L) and their own growth. So, each product manager ends up going after overlapping customers.

Whether it's Home Depot or a large bank, everyone is hitting you with marketing, and it's not orchestrated or coordinated. Usually, the same customers tend to be eligible for multiple things. They're likely to score high as good prospects.

And so, there's no coordination. As you said, everyone has their own database, and they're all coming after you. Other things happen that keep the data separate and balkanized. For example, if you call customer service with a problem, that information doesn't get to marketing.

I'll give you a great example, and I'm not afraid to share it because I've met their chief marketing officer and talked about it. I'm a big fan of Sonos speaker systems, but they recently did an app upgrade that isn't working well. I called and wrote to them. I'm not the only one having significant problems with the app. Yet, they constantly try to get me to buy more speakers. There's no point in that until they fix the app. But that information doesn’t seem to be shared across departments.

It certainly doesn't come across that way. A lot of this has to do with the balkanization of data, with the lack of anyone taking a customer's point of view. At a high level, no one is saying, "We really want to drive customer lifetime value as one of our main goals," instead of just bombarding customers with marketing.

At Aetna, we measured this. When I became chief marketing officer there, I also got Aetna health insurance. I was a customer as well as the chief marketing officer. Within the first two months, I was bombarded with all kinds of marketing from my own team.

I asked, “Why is this happening?” So, I requested the data on how often we contact someone, their open rates, and unsubscribe rates. It was absolutely clear that, beyond four touches a month, everything plummeted. People were unsubscribing or not opening emails. Less is more, especially if that "less" is smarter and more relevant to the customer.

Michael Krigsman: So, you're really moving far away from that spray-and-pray mentality where you just use the data, dump it out there, and send as much as possible because it costs almost nothing. By the way, taken to an extreme, that's the philosophy of the scammers who call us all the time.

David Edelman: Yes, and you don't want to be lumped in with them. I think, for the greater good, there will be questions about regulation and managing customer data. I think regulation of how customer information is used is definitely needed. But we also need to show people that it's not all invasive, creepy, and manipulative. There are many valuable uses of customer information.

For example, if you return to a website you frequently visit, you want the site to recognize you, know you've been there before, and what you've looked at. That makes the experience more relevant and speeds up checkout.

We need these examples of how using customer information adds value. The danger is the "tragedy of the commons," where so much irrelevant material is thrown at people that they become very uncomfortable. This puts even more pressure on limiting what's possible.

Michael Krigsman: Arslan Khan on Twitter asks: "With so much repetitive data being collected by so many organizations, would it ever be possible to have a central repository for all appropriate data? Could organizations just do data calls to pull that data out? Do you see organizations being willing to collaborate on something like this?"

David Edelman: I don't think we'll see a single, central repository, nor do I think we want that because of the risks involved, like single points of failure. But I do see two things happening. First, ecosystems are emerging where groups of companies work together to share information because sharing allows them to provide a solution.

So I'll give you an example. Marriott Bonvoy in the travel industry. They're moving towards a capability where you could tell a chatbot, "I want to go to Spain for 10 days. I want to use as many points as possible. I have two teenage children. Here's my budget. We only want to stay in three hotels over the 10 days. Give us itineraries and booking options."

Marriott can't answer that completely on its own. It has to coordinate with transportation providers, attractions, restaurants, and so on. It needs an ecosystem where it can move information around. We could get into a whole discussion about agentic AI, which is one way for that information to go out and make those calls.

But AI lets you combine information more easily and track it, so you have an audit trail and can control who sees what. I believe companies like Marriott—and possibly Home Depot or Lowe's for home renovation, along with others in healthcare and other categories—are going to become destinations tied to loyalty programs that hold a range of information shared across their ecosystems.

I think the other likely scenario is that people will have their own agents holding the information they want to share with others. Companies can then request and access that information with the user's permission.

I think we'll see Google, Apple, Meta, and Amazon all competing to provide capabilities like that. It will be a question of ensuring these operate appropriately and that people are comfortable with how information is managed.

Michael Krigsman: It's interesting that you mention agentic AI, as I was thinking agents seem like the natural way to accomplish the lengthy series of tasks you described in the solar provider example. It's elaborate personalization based on research and data.

David Edelman: Sungevity, like I said, uses a series of different AI techniques. Four years ago, when this happened, we didn't really have agentic AI, so there were manual steps in between. Still, much of it was automated.

Yes, you can now string these AI techniques together easily. Some of it is through agents. Some of it is what we call composite AI, where multiple AI tools work together. I'm seeing more companies thinking about delivering personalized customer solutions that require several different actions and using agents to bring those together.

For example, Home Depot could use a picture of a bathroom I want to renovate. They could analyze that picture for dimensions and placement of fixtures and pipes. Then, they could suggest products that fit those dimensions, calculate costs, and use generative AI to create a new rendering.

That could all be strung together using agents, as I doubt a single piece of software can do all of that. I think this is where things are headed.

The key, though, is solutions. It's about providing solutions to customer needs and figuring out how to automate the steps involved to make that happen at scale.

Michael Krigsman: I want to take a moment and suggest that you subscribe to the CXO Talk newsletter at cxotalk.com. We have live shows, and we'll notify you so you can join in and ask questions.

And we have more great questions coming in. This one is from Ritu Bandari. Ritu says, "Thanks for the great insights. I'd love to hear your view on leveraging AI to personalize customer journeys in B2B marketing." So, using AI to personalize the customer journey in business-to-business marketing.

David Edelman: B2B, I think, is an underdeveloped area for this, but there are definitely examples. The challenge is that there are several different routes to market in B2B. You can be direct-to-customer or have a salesperson involved. I’ll give you two examples, and it's interesting because both companies are named Cisco.

One is Cisco, the food delivery company. You see their trucks everywhere. Restaurants across the country use Cisco and its competitors to order food and supplies.

When a purchasing agent opens the Cisco app, within 300 milliseconds, they get a completely personalized experience. The app knows who they are, their restaurant, menu, price points, location, and whether they typically order in bulk or on the spot. It knows the nearest warehouses, what's in stock, and, since it’s food, what items need to be moved and potentially discounted. It could even suggest new menu items based on available ingredients.

All of this creates a completely personalized experience. If you think about the five promises—empower, know, reach, show, delight—it hits all of them. Cisco’s customers are amazed at how much easier this makes their lives. The app then connects with the supply chain, coordinating deliveries. All of that is managed.

Now, they don't use agents for the entire process yet, but you can imagine the entire experience, from opening the app to delivery and invoicing, being completely end-to-end managed eventually.

A different example is the other Cisco, the technology company, which sells primarily through salespeople. Cisco has over 100 products. Each salesperson has many different customers, and they can easily feel overwhelmed. They might ask, “How do I know which product to discuss with which customer?”

Traditionally, when Cisco announces a new product, they send the announcement to all customers, hoping it will spark a meeting. But most customers aren’t interested in that particular product.

And it led to frustrated salespeople and missed meetings. So, Cisco has created a CRM system on steroids.

They started with a core CRM, but then added information on how customers use their Cisco products. Which features do they use? Is usage trending up or down? Where are they located? What marketing materials have they viewed?

They even incorporate outside information. "Is this company expanding, acquiring others, entering new markets? Are they facing cost pressures?"

All of this feeds into a system with a content management backbone. Salespeople receive a weekly call list with who to call, what content to discuss, and why—meaning why the system recommended that specific action.

Since using this, they're getting way more meetings, cross-sell rates have gone up, and the sales force is dramatically happier. That’s personalization, both for the sales rep and the end customer, who receives dramatically more relevant information.

So again, it's not automated because there's a salesperson in the middle who does exercise judgment. Cisco is working on fully automating this process for the lower end of the market.

But it's a B2B situation where they're leveraging the steps in the sales channel and personalizing each one. These examples show some of the possibilities in B2B.

Michael Krigsman: It strikes me that in this case, additional salesperson training would be required to help them understand how to use this wealth of personalization data and determine what’s most important and relevant for each customer.

David Edelman: Yes and no, Michael. AI is simplifying what salespeople see by boiling it down to the customer, recommended content, and the rationale. Salespeople don't need to dig into the underlying data. Generative AI is making the interface simpler.

Now the "yes" part of training involves both learning to use the system and inputting data back into it. Salespeople need to record what happened during customer interactions: what was discussed, what engaged the customer, etc. Much of this can be captured through voice input after meetings, with the app translating it into data.

So the salespeople need to become part of the data loop, not just users. This often takes work and requires them to perform follow-up actions they may not be used to. But this capability is particularly valuable in organizations with high sales force turnover.

Michael Krigsman: It's interesting how you describe the role of AI in summarizing and simplifying a large amount of data. AI will only get better at this over time, doing an ever-improving job of pulling out the relevant information.

But there are potential changes to work resulting from this. Greg Walters on LinkedIn asks, "How do you see the impact on labor, from the warehouse to the C-suite?"

David Edelman: Speed is crucial for successful personalization. BCG and I created a personalization index, looking at companies' capabilities, the experiences they create, and who the leaders and laggards are.

The leaders operate differently. They recognize that AI requires a faster metabolism.

One is, when you get a signal from AI – because they spot a trigger, they see an opportunity – you should be jumping on it right then and there. Ideally, even automatically in real time.
So the faster you can act on it, the more value you can get.

Second is AI requires more information to learn and get better and deliver delight, essentially. And the more you can test and learn and get things out the door, the more data back you get to feed the AI and make it smarter. So that you understand better which are the kinds of experiences and creative that's going to get Michael to act versus David to act.

And so this whole concept of speed becomes absolutely critical. And in these companies, they are redesigning their marketing organizations to work in pods with a lot less specialization. So groups of five or six people – basically strategy, analytics, creative, operations and technology. Maybe somebody from compliance has to be involved as well.

And they are working together to come up with ideas of what to test, get those tests out the door, look at the data, and constantly improve things over time. Now, from a work perspective, which was your question, there's several things involved here. One is just that nature of work. Instead of taking six weeks to get a program out the door, you're getting five programs out in two weeks.

So it's a totally different dynamic because you're getting people to work together. The other thing, in order to do that, is you're reducing the amount of specialization. And AI is enabling and forcing people, I believe, to be more generalists.
A lot of the things in marketing that have led to what I believe is hyper-specialization – people who are just doing social, just doing search – is because of the quirky nature of those interfaces. You have to understand how Google operates, how programmatic buying operates, and all the execution systems.

AI is going to simplify all of that. The execution will be dramatically more automated. So that means you've got to focus on thinking, "What do you want to do? How do you want to think about the nature of the touches and the strategy?" And that's going to force people to have more breadth.

That's better for their careers because they're not pocketed in narrow specialties, but it does force people who have been very narrowly focused in an execution specialty to step back and think about how they can broaden their skills. And companies have to help them do that.
So one thing I absolutely see in marketing is a reduction in specialization, the use of small teams working together, and more generalist capabilities.

I think we're going to see that in a number of different areas. For example, in purchasing for supply chain management. This is not about marketing and customer experiences, but a similar thing where you can have regional or more centralized views because the AI makes things smarter to be able to manage and coordinate inventories.

That can have ripple effects on employment. It could. We haven't seen it yet. Most companies are using this to redeploy people, get more scalability out of them, recognizing the growth that they're getting, doing things they haven't been able to do before and getting smarter.

But there is a risk of that. I think a lot of what we're going to see is a tension between a pull towards more centralization and generalist capability versus – well, at the same time we have to think about – people who are very narrowly specialized, pocketed, and how we can redeploy them. So there will be a tension there.

Michael Krigsman: While we're on this subject of changes in work, what advice do you have for chief marketing officers at this juncture, where there's so much that's evolving and changing so quickly?

David Edelman: I think there's two things that chief marketing officers have to do. One is you've got to start experimenting with the capabilities that AI offers to be able to automate some more things, get your metabolism up, get faster, and get the per-touch cost down.

So be able to generate more creative and use that to test which is going to be the best, and do that much faster. Be able to get tests out the door faster. All of which can allow you to do more and to scale and to hit growth targets faster. Because it's not just about overall lowering costs, it's about lowering the cost per touch.
So that if you can lower the cost per touch, you can use the savings to drive more growth. And you've got to be able to prove that you can do that balance. That's one thing, which is to reshape your organizations to be more scalable using the AI capabilities.

The second, though, is at a higher level, getting away from just marketing tactics and thinking about personalized experiences, like I was talking about with Cisco, Sungevity, and there are many others in the book.

Marketing has an amazing opportunity to be the strategic leader at the executive table who is banging on the table for personalization as a strategic opportunity.

That certainly was the situation I was in at Aetna, and I see that in a number of other places. It requires the whole leadership team to rally behind that because there's so much cross-functional capability involved.
But someone's got to be the quarterback on the executive team. The CEO can't be the quarterback day-to-day.

The chief marketing officer, or evolved chief experience officer and or chief customer officer - however the label is – has a tremendous opportunity now to step up and say that these AI capabilities can enable us to compete differently.
And I'm working with companies for whom that is absolutely the case. And the organizations want the CMO to step up to that because that is the right person. They're closest to the customer, they understand customer data, they can coordinate how to create more value.
So I see it both as a pressure to get costs down and scalability up, but an incredible opportunity to be much more of a strategic leader.

Michael Krigsman: To what extent, broadly, are you seeing CMOs that possess the emotional, the flexibility, the intellectual flexibility, to adapt in this way, as you're describing?

David Edelman: I don't have an overall assessment of all, but recently, together with BCG, we did a survey of chief marketing officers on how they were using AI tools. As you would expect, the vast majority - 70, 75% of them – have been experimenting with AI, and mostly for the tactical advantages it can give them.
To automate some processes of creative, help automate things like media selection, a lot of the tactical stuff which I would say is kind of the easier stuff, they're starting to work on.

And there are organizational issues that they're facing in terms of pursuing that. That's part of what they're grappling with. It came up a lot. I was at the ANA, Association of National Advertisers, Masters of Marketing conference last week in Orlando.
How organizations are evolving was a key topic of conversation, moving more towards the pods, as I talked about before, whether that's easy or hard to do. But most are finding if they do, it's one of the best ways to get it working.

As far as the strategy side, really only about 10% of the CMOs, admittedly, stepped up to say, say that they're stepping up to really lead the way on personalization and using the AI capabilities for more value-oriented personalization.
Quite a few – about 30% – are saying it's on the horizon, they want to get there. About 10% really said they were there. And interestingly, when we did our personalization index, where we looked at 200 companies, it was only about 10% of the companies that we considered leaders.

What was interesting, though, and one of the things that surprised us through all of this research, is that personalization leaders are happening in every category except consumer packaged goods, because in consumer packaged goods it’s really hard to get end customer data.
But across any business that has direct customer contact – you are seeing B2B, healthcare, financial services, travel, retail, fashion – all of those businesses, you are seeing leaders. So we're seeing more of a range within categories than between categories, which means it's possible for any business to pursue this. It's really a question of focus.

Michael Krigsman: Arslan Khan comes back and says, “Data, as you described, is being used to streamline end-to-end processes for better customer experiences.” He says, “It sounds like consultant-speak. But what kinds of governance is needed for this type of data, and should IT be the only department responsible for this data?”

David Edelman: There's a few things loaded in that. Let me actually work backwards. First of all, in terms of who should be responsible for the data, that is the responsibility of almost any, any department that is touching the customer, whether it's customer service, marketing, sales, billing, product – all of which capture data from the customer.
IT enables a lot of the platform capabilities, the technology underneath it, but there are decisions in how you manage your business that are critical that determine whether or not you're going to capture data.

A very simple one is, when you are running a marketing program, for example, are you tagging – with metadata tags at a very granular level – all the aspects of your contact, such as the color, the nature of the words, the size, what time somebody opens it, whether they were touched beforehand to open it? When I was at Aetna, we had 44 different tags that we were using to understand what was going on in an interaction. That's not an IT thing. That's a functional decision to say "we want to capture that data." Then you work with IT to implement that and make that possible.

So the business has to decide that it wants that data. The business also has to decide that it wants to share data and bring together data in order to get a total picture of the customer. And that requires having a vision for what you want to do with that data and working together to make that possible.
Now, as far as governance is concerned, IT is very critical, but I've also seen roles such as chief privacy officers in some companies - especially financial services and healthcare companies – that are thinking about policies and guidelines, some of which are technical, but some of which happen during the interaction. "What questions do you ask somebody in a customer service call to get more information from them in a permissioned way?"

So are there guidelines, from an operational perspective, for example, in a call center, of what data somebody can or cannot access? So there's usually some role. It may or may not be in IT. It could be. It could be a chief data officer. There's a lot of different evolutions here.
But there are policies, guidelines, and definitely technology capabilities that do need to be in place. The important thing is to make it explicit that you're going to manage that, have somebody on the hook to lead the way. But they have to work with the executive committee to make that possible. It takes a village.

Michael Krigsman: Which gets right back to the beginning of our conversation, where personalization becomes the core of strategy, really.

David Edelman: Yeah. And this is something that isn't just in marketing. That's one of the huge points of the book. And in fact, there's a chapter in the book that goes through the different C-suite roles: CEO, even chief legal, general counsel, CFO, CIO, COO - all the different roles in the leadership suite - and how they contribute to a personalization strategy.
The thing is, it's a strategy, and it is a decision of how you want to compete. So for example, several years ago, Howard Schultz at Starbucks said publicly, in an investor call, he wanted Starbucks to be the most personalized brand in the world.

And Starbucks’ app - very similar to what I talked about with Cisco – you open that app, you get an incredibly personalized experience. Now they may have had some operational issues that have stemmed from that, that caused the stores to be overloaded.
But from a personalization perspective, you can't deny that they put things together to try to make that possible and actually did advance the ball. There's just a lot of systemic things they still have to work out.

So it is a corporate strategy that marketing may be the quarterback for, but it takes the C-suite working together.

Michael Krigsman: Share advice on really walking through those five promises so that marketing folks listening can get a little bit more of a concrete sense of how they can address the personalization opportunities and challenges that you've laid out.

David Edelman: Number one is, really the goal is to empower customers. But in the very short term, immediately, there are things you can start doing, two in particular.
One is managing your reach – so reach me – and orchestrating contacts so that you are not overloading the customer. So having some function, some role, somebody who's doing the analysis on what really is the next best action. There are certainly IT capabilities, and probably most of you have CRM systems that can do this if you let it do it. If you put the data in and actually let it help you make decisions, so that you can do "less is more.” Be more appropriate in how you're actually touching people.

The second, which is also tactical, is show me, and actually using capabilities like SundaySky to create personalized videos. That's a really easy thing to do. To take something that may be more complex, like helping you understand your loyalty program and what you can redeem points for, given how many you've accrued, given what your interests have been in the past.
All of that can be done in a personalized video. These are things you can just start with. They're tactical marketing things, but they can make a difference.

Then the broader thing is to step back and say, "Are there more systemic opportunities to address problems that customers have, or open up new possibilities, that will take a bit more coordination across the board to make possible?"
Thinking about things like the way Sungevity said, “Selling solar panels is ridiculously complicated or reductionistically manipulative.” So they said, “We can come up with a different way to compete.”

Thinking about those opportunities, putting those on the table, and seeing, even before you get all the AI, are there ways to do this at a small scale to prove the opportunity? And then as you start doing that, you can see what data and technology capabilities you need to scale it.

Michael Krigsman: Arslan Khan comes back again, and he asks great questions. He says, “If everyone is responsible for data, then no one is. In organizations, departments might view which data is important, which is not. Should we have a central understanding of what data actually means for every department? Should organizations somehow specify what data means in a standard way through the organization?”

David Edelman: Yeah. And that's why a number of companies are setting up roles like chief data officer, or chief data and analytics officers. Larger companies are absolutely putting in those roles to do exactly that and to work with department heads.
I mean, even legal: there's data in all the contracts that you manage. So they are working with functions to do that. Short of that, it's probably the technology team that needs to do that.

I would argue, though, that from an experience perspective, marketing should be at the table saying, “In order to deliver” – so not all the data for everything – “but to deliver the experiences we want to deliver, here's the data we need.”
And work with the executive team explicitly on how to unlock that data, have a roadmap for doing it, work together through the issues, the trade-offs, do problem solving to do it, and be very focused. Because you can go after everything and spin your wheels forever. But focusing on what's going to make a difference to the customer is where you should start.

Michael Krigsman: How do we measure the impact of personalization?

David Edelman: It depends a lot on what your goals are for personalization. Is it customer engagement, to cross-sell, more use of your products, growth in lifetime value, lowering the cost of servicing your customers?
It's important – we hadn't gotten into this – but personalization programs need to have a business-oriented goal. In the case of Cisco, the technology company, it's about sales. It's about getting in front of customers at the right time and driving growth and sales.

In others, for example, at Aetna, one of the parts of personalization was helping people understand their health plans so we could get them to take healthier actions, have lower calls into the call center, raise customer satisfaction. But there were measures that were around cost and managing customers. So it really depends on your business goals.

Michael Krigsman: Clarity on what you're trying to do, why you're trying to do it.

David Edelman: Yep.

Michael Krigsman: Are there obstacles or challenges or resistances that tend to come up as an organization is adopting, or trying to adopt, personalization? Can you briefly speak about those? And what are some of the solutions or antidotes?

David Edelman: One of the main things is the fact that it is cross-functional, and it does require different parts of the business working together in ways they have not before, which requires trade-offs. It requires making decisions about who has the big "D" for actually being the decision maker on things when there are trade-offs.

I'm seeing this in organizations, for example, where they want marketing, or a CXO, to take the lead in driving this. Well, then that CXO becomes the client for IT, for service operations, to do the investments and deliver against the specifications of the CXO who has made a business case that the organization is driving towards.

So there are definitely functional dynamics and new kinds of models that companies have to experiment with. And there's no question there are challenges there.

Michael Krigsman: Okay. And with that, a huge thank you to David Edelman. David, thank you so much for being with us. I'm very grateful to you.

David Edelman: Oh, my pleasure, absolutely. Hope everyone has a good weekend if you're seeing this on Friday.

Michael Krigsman: And thanks to everybody who asked questions. You guys are an amazing audience. You guys are so smart. Before you go, please subscribe to our newsletter and subscribe to our YouTube channel. Check out cxotalk.com.
We really do have extraordinary guests coming up, so check it out and have a great weekend, and we'll see you again soon. Take care, everybody.