Ep. 113 | The New Customer Segmentation with Forrester’s Brandon Purcell

This week Brandon Purcell, Principal Analyst at Forrester Research for customer analytics and artificial intelligence, joins Allison Hartsoe in the Accelerator to talk about customer segmentation. Brandon has seen a collision of customer segmentation schemes between traditional consumer research segmentation and company-owned customer segmentation.  Learn why it is difficult to take action on traditional customer segmentation and how you should think about segmentation now.  

Please help us spread the word about building your business’ customer equity through effective customer analytics. Rate and review the podcast on Apple Podcast, Stitcher, Google Play, Alexa’s TuneIn, iHeartRadio or Spotify. And do tell us what you think by writing Allison at info@ambitiondata.com or ambitiondata.com. Thanks for listening! Tell a friend!

Podcast Links:

The Best Of Times And The Worst Of Times For Segmentation

It’s The Most Wonderful Time Of The Year For Segmentation

Read Full Transcript

Allison Hartsoe: 00:00 This is the customer equity accelerator. If you are a marketing executive who wants to deliver bottom line impact by identifying and connecting with your revenue generating customers. Then this is the show for you. I’m your host Allison Hartsoe, CEO of ambition data. Every other week I bring you the leaders behind the customer-centric revolution who share their expert advice. If you are ready to accelerate, then let’s go. Welcome everybody. Today’s show is about customer segmentation and the many forces changing how we think about it. To help me discuss this topic is Brandon Purcell. Brandon is a principal analyst at Forrester research covering customer analytics and artificial intelligence. Brandon, welcome to the show.

Brandon Purcell: 00:48 Thanks for having me, Allison. I’m happy to be here.

Allison Hartsoe: 00:50 So I think we all think about Forrester as a research company, but tell us a little bit more about your background and maybe the breadth of what you do at Forrester.

Brandon Purcell: 00:59 Yeah, sure thing. So I’ve been an analyst at Forrester for four and a half years now, and I sit on our customer insights team at Forrester, meaning that I write my research for folks who have access to a lot of customer data and every day they’re trying to turn that data into insights. And so, my research focuses on customer analytics, how to transform massive amounts of customer data into insights, so these people better when serving retain their customers. Now, in addition to that, because there’s a machine learning element to a lot of the techniques that I write about when AI became a thing again about four years ago, I took on some of our seminal AI coverage, and I continue to manage our overall AI taxonomy. What are the different technologies, component technologies that make up artificial intelligence? And so I go very deep into this world of customer analytics and customer insights. And then I look very broadly at the AI tech landscape. My background is, before I came to Forrester, I worked for a small service provider out here in the Bay area and I led their data science teams. So I did a lot of the modeling and the analytics that we’ll probably be discussing today.

Allison Hartsoe: 02:04 Excellent. Well, it sounds like you’re well-positioned to talk about this particular angle and really anything about customer analytics. Fantastic.

Brandon Purcell: 02:12 Certainly something I’m passionate about it.

Allison Hartsoe: 02:13 Yeah. So let’s say that I have my own business intelligence team and they’ve clustered all my customer data, and they’re trying to tell me who I should target. How is that traditional approach to segmentation, maybe not the right way to approach it anymore?

Brandon Purcell: 02:30 Well, first of all, I would say congratulations if you have a BCI team that was able to pull all of your data together and unify it and then cluster it. I talked to a lot of clients who can’t even do that, right? Like they’re stuck in what I call the data doldrums or data drama, and they can’t unify their customer data. But even if you do have that kind of segmentation, obviously with today’s ever-shifting expectations of customers, there’s an opportunity to do better segmentation. Segmentation that’s much more dynamic to create these more relevant personalized experiences for customers. And I really think about segmentation in two types. I just published a report on segmentation called the tale of two segmentations, and it was the report was a response to the demand for best practices around segmentation last year among our user clients. So non-vendors meaning brands.

Brandon Purcell: 03:20 The number one theme in my inquiries was customer segmentation. 15% of my inquiries from these folks were about segmentation best practices. And that’s not because segmentation new obviously segmentation’s been around us as long as it’s really customer data has been around. But what has changed is one exogenously, the increasingly empowered customers in the age of the customer demanding personalization and these better experiences. But on the other side internally, companies just have a lot more data on their customers and so may need to find better ways to use that data to segment customers. And so I embarked on this research, and I found when people talk about segmentation, they’re really talking about potentially two very different things. One thing in the report I call consumer segmentation. And now this is the more traditional market research-based approach to segmentation where you are doing focus groups, or you’re doing surveys, and you’re identifying people’s attitudes, their values, their motivations to find these psychographic groups of people.

Brandon Purcell: 04:18 And that’s very useful segmentation from a strategy perspective, from a product enhancement and product development perspective, and from a communications perspective, then you learn a lot about both your customers and also potentially customers you don’t have yet. But it’s really hard to take action on those segments on your customer database. And that’s why there’s a second type of segmentation that’s much more actionable, which is your customer segmentation, grouping your customer based on shared attributes, shared behaviors, shared patterns of transaction or channels. And this type of segmentation is more analytically driven, probably driven by maybe a BI team or a data science team where a segment classification is actually assigned to each customer in the database. So you can start to group these people in furnish, really differentiated experiences for people who fall into each segment. This is the type of segmentation that really needs to change because you can’t get away with doing demographic segmentation anymore. Obviously, all people in one demographic aren’t the same, really needs to be looking at much more granular data about them and about what they expect and want from you in that moment.

Allison Hartsoe: 05:23 Oh, that makes a lot of sense. So I want to circle back to two things that you said. First, why is it hard to take action on the traditional market research?

Brandon Purcell: 05:32 Sure. So I talked to one company, it was a large North American insurer, and they worked with the vendor to create that type of segmentation and took about six months, and they did a very rigorous job with focus groups and surveys and, and ended up with I believe eight different segments that were described and articulated very well. And they were very happy with it until a business leader came in and said, okay, how many Dave and Susan’s do we have in our customer database? And they spent the next 12 months trying to answer that question. And the reason that they ultimately couldn’t answer that question is because the different variables that go into a psychographic or attitudinal segmentation, the things you’re asking about when you’re creating those segments don’t endemically exist in your customer database. So how do you actually find those Dave and Susan’s within the customer database?

Brandon Purcell: 06:22 Now, they tried lookalike modeling, and because they knew who the customers worked with, filled out the surveys and so they tried to do some propensity modeling to figure it out, but they couldn’t, they couldn’t get it to a point where the accuracy of this model is better than a coin flip. And so they actually just jettisoned the segmentation. And so the segmentation project they spent a lot of money on and a lot of time on just sits as a report in an executive staff. Unfortunately, that’s happened to quite a number of companies. So the question really becomes, what do you want to do with the segmentation? Because I don’t mean to say the consumer segmentation isn’t actionable at all. If you’re trying to figure out, okay, what should our company and customer strategy be? That’s a very useful type of segmentation. But if you’re trying to actually address individual customers, you really need to think hard about whether that segmentation makes sense.

Allison Hartsoe: 07:08 Okay. Yeah, and I completely agree with you. I remember an example in the class at Wharton where we were talking about how a news story had used demographic information to come up with this ridiculous assumption that all Latinos were more likely, or the Latino audience, in general, was more likely to buy CDs than any other, and it was just like, clearly they only had demographic segmentation to work with, and that’s all they could find in common. But it was like making a connection between sheep farmers in Afghanistan. It looked like it meant something, but it really meant absolutely nothing.

Brandon Purcell: 07:46 Right. That’s great. That’s a great example.

Allison Hartsoe: 07:49 Okay, so the second half of what you were talking about was about the customer data on an individual basis and when you talked about segmentation assigned to each customer, to me that reminds me of a propensity, like a propensity to buy or propensity to use a channel. Is that more or less what you’re thinking in that spot?

Brandon Purcell: 08:08 Well, yeah, I’m glad that you’re thinking in that direction because it really is. So I think of propensity modeling as the type of segmentation because ultimately you’re coming up with these propensities of customers to take a certain action and then when you address the customers, you’re pulling a group of them that meet a certain threshold propensity, right? And so that is ultimately a type of segmentation, and I think in an analytic, analytically advanced firms that have multiple different propensity models, each of those models can be used as a segmentation at a different point in time. For instance, churn propensity, it’s kind of the canonical customer analytic technique. Looking at past customers who turned and then using machine learning to find statistically significant behaviors and patterns of behavior that lead to churn. Here you can use this of model towards the end of the customer life cycle, right? To save those customers who have a high propensity of turning and maybe also in conjunction with the lifetime value model to ensure that the, you’re actually spending your retention budget on high-value customers, whereas a different propensity model or propensity to purchase model and will probably be used at the beginning or in the middle of the customer life cycle.

Allison Hartsoe: 09:15 And what I like about that model that you’re saying, but I’m not sure it’s coming through, I want to underscore it for the audience, is that when you pull that data, you’re almost pulling people in a dynamic fashion. In other words, I might have the same address and certainly be the same race and sex, but I will always change my propensity to do different things and have different behaviors and where I am in the life cycle with a company. So we’ve changed from these old static models to these fairly dynamic, almost intuitive models. Would you agree?

Brandon Purcell: 09:50 Yeah, I agree. And the intuition comes in on top of the models and by that I mean an advanced company will likely have propensities to, for different products, propensity to churn, propensity to respond. Maybe even service propensity, the likelihood a customer needs a specific type of customer service, that’s great, but at the end of the day, those propensity models aren’t decisions. There needs to be some sort of decisioning apparatus on top of the model that says, okay, when these different conditions are met, when branding is at this stage of his customer journey with us, here is the right experience to deliver based on the propensity model that makes the most sense at that point. So if I’m at the end of my journey, if I have a very high propensity of turning, obviously you’re going to send me a retention incentive. So it looks like I’ve been saving up and I’ve been looking at let’s say mortgage information. Well, you’re probably going to send me a mortgage offer.

Allison Hartsoe: 10:40 And I think this automatically takes a company into testing and I’ve always thought of that as kind of the Canary in the coal mine. If a company doesn’t have a testing program dialed in, they’re probably not ready for anything stronger.

Brandon Purcell: 10:52 Yeah, I am so glad that you said that. So testing is so key to all of this. I mean, we talk a lot about AI and machine learning and everything, but at the end of the day, we have to actually make decisions based on these models and then test new experiences to see which experience is actually optimal and different experiences are going to resonate with different segments of customers. And so we need really needs to be continuous. And that changes over time too, right? Because you’re dodging dodginess variables like your competitors are making movements in the space too, and so you have to be continually testing communications, content experiences.

Allison Hartsoe: 11:25 That’s true. I think what you said about customers, their expectations change, but also competition will copy what you’re oftentimes doing or what you’ve spent a lot of hard-earned data science time figuring out and they’re like, Oh, thank you very much. We’ll try that with our customers.

Brandon Purcell: 11:42 Yeah, exactly.

Allison Hartsoe: 11:43 Yeah, so it’s hard to say on top of it, but I think some companies seem to be doing it better than others. Are there particular examples that come to mind for you where you’ve seen them really perhaps use LTV or use better technology or experiments to be more, I’m going to call it customer-centric, but what I really mean is that heterogeneity that sits underneath the data.

Brandon Purcell: 12:05 Yeah. Well, when I think about lifetime value, in particular, I always think about the Royal Bank of Canada case study, which is a little dated now, but I’d like it because they’re using LTV and the way that I think companies should use it, which is to try to find not just your current high-value customers, but also these new customers who may not look particularly high value, but exhibit signals that they’re going to become high value over their lifetime. And so they think the same. They built a lifetime value model, and they found these folks who were actually not just broke but massively in debt. And then we’re going on to become 3.7 times as valuable as their average customer. And we did a little more digging, and they found that these people were really meant to dentists and doctors and so they just graduated from medical school, and they were in debt and they were taking on more debt to open their practice, but they were answering this extremely lucrative profession. And so Royal bank of Canada sent this information and created a bunch of programs and services directly to these individuals in Canada and ended up growing their market share in this sub-segment from 2% to 18% and again, these people worth almost four times the average customer. So incredibly successful for them.

Allison Hartsoe: 13:12 Really nice. I oftentimes think that companies already know this. They should understand if they’ve got all the transactions in the bank, and they should see that there’s a certain amount of debt and what was the debt for, but they don’t. And I think that’s largely because the information is broken into multiple databases or somebody is just not putting it together correctly.

Brandon Purcell: 13:31 Yeah, you’re totally right. And that’s one of the things I really like about out of the different customer analytics techniques, lifetime value and journey analytics. Because both of those techniques require data from a bunch of different domains within the enterprise. And so one of the kinds of the Trojan horse of doing these different techniques is at the end you have this unified data on customers, and that becomes incredibly powerful as a foundational asset for analytics. For example, USA, I put up there as an example of a company that does customer insight probably as well. When John Hershberger took over as their head of insights team wanted to understand the product journey of different customers as they form relationships with the USA. So what’s the first product, second product, third product, and in order to you, so obviously he has a unified product and information at the individual customer level.

Brandon Purcell: 14:22 Yes, a bank and insurance. They have probably 30 35 different quote-unquote products, but still like the order in which customers can own those products is their pretty mutations is about is probably in the billions, right? What they found is that five different product journeys accounted for 90% of their customers, and 80% of their customers were on one of three different products during the user sequences. And so just knowing that the new next best products pretty easily, but he said the real value of that was after we’ve done that now they have this unified database at the customer and product level to build all sorts of other.

Allison Hartsoe: 14:56 You know what surprises me though is I remember a USA example probably ten years ago where they were using the loyalty or the membership number that they had to try to accomplish a very similar task and I think they have that advantage and that the membership number is used on everything that the customer touches. Right. A huge advantage for them.

Brandon Purcell: 15:18 Huge advantage, yeah, for sure. And as I mentioned in my intro, I did a lot of this work before, and it was mostly with large banks, and you think that in large banks you have these identify the key, so it’d be really easy to do essentially an inner join at the customer level. It can be really difficult from people in process perspective as well as from a data perspective because just because data I of data hygiene issue.

Allison Hartsoe: 15:41 Yeah, I think that’s right, and in most of the examples of that I’ve heard where somebody has been wildly successful, they almost inevitably had two things, probably more than two things in place, but one thing they had was a senior-level position so they could reach across their colleagues to get what they needed either at the management team level or a step below the management team level and then behind that they had air cover from the CEO in order to be able to resolve these issues where one of their colleagues was like, yeah, I’ll get to that. Nah, not important to me. They needed that. I’d even heard a story where the executive took a letter from the CEO to the other executive to say, yes, this is okay to give me, and we’re going to do really great things with it. But that level of the people in process I don’t think can be underestimated.

Brandon Purcell: 16:33 No, not at all. That’s so nice. When it comes from the top down. You’re trying to build this from the bottom up. It’s going to take a lot longer, and it’s going to happen in a piecemeal fashion, meaning two lines of business at a time working together.

Allison Hartsoe: 16:45 So have you seen it though, had live into success in places where you can build from the bottom up?

Brandon Purcell: 16:51 Yeah, I have seen it. I did speak to a couple of banks who are doing what they call next best conversation where you have your marketing team and your customer service team working together towards common goals. One bank, in particular, had their mining team building out propensity models, propensities for different brokerage accounts, and financial services products. And then they were actually showing these propensities to the customer service teams so that when a customer called in, let’s say I called into that bank, there were no, first of all, who I am. And then, the customer service representative would actually see the three next best conversations, and they could be conversations based upon my likelihood to need a new product for my likelihood to meet my type of customer service. Um, and so how do you measure the efficacy of that? Right, when you have like marketing goals and customer service goals, and so they continue to use the plastic kind of call center goals, average handle time, first call resolution, but they were also looking at the efficacy of their cross-sell and upsell capabilities as well and then the older arching goal to ensure the success of this program and really to determine whether or not to jettison the program was NPS, net promoter scores.

Brandon Purcell: 17:57 So is this actually having a positive impacts on the customer experience or a negative one? But ultimately, that’s the case where marketing and customer service or customer success work together, share data and insights and operationalize something that for them was ended up being better for the customer.

Allison Hartsoe: 18:11 That’s a great example, and it also reminds me when you say off sell and upsell, immediately I start thinking personalization. Have you seen examples where personalization is the immediate application?

Brandon Purcell: 18:21 Yeah. Really ultimately, for all of this talk about segmentation and customer analytics in general. I think that the goal is a type of personalization that we had forced her called the next best experience. So ultimately taking everything that you know about a customer, distilling it down, using different analytical techniques to identify what is the right experience to furnish for this customer in this moment, either the next time we see them occur or maybe even proactively. And so that means not just personalization, parts of personalization in terms of cross-sell and upsell, but also looking at their likelihood to need customer service or their likelihood to need some sort of education about their products and services. For instance, I’m a Charles Schwab customer, and they’re really good when there’s market volatility like there has been these last couple of weeks. They’re really good at proactively reaching out with their own analysis or take on what’s happening in the market and prescriptive advice of what to do with your funds. Ultimately there’s no short term win in that type of personalization or customer experience, but they’re trying to build longer-term loyalty that’s going to increase my lifetime value over time. So I think personalization is the goal of this, but personalization with a lens towards what the cost, what the customer would like to get from us, and also incorporating insights from multiple different business domains. And that’s what we call next best experience.

Allison Hartsoe: 19:39 Well, and it kind of reminds me of what you said at the very beginning when you verified that there’s still a place for this traditional almost consumer side of segmentation. And when I think about these examples, they’re oftentimes very company-centric. They’re talking about the customer, but it’s what can we as a company give to the customer at the right time as opposed to what the broader consumer need is and how that’s changing, which is picked up from that other style of segmentation. And then then the question becomes how almost consumer-led or customer-led, where customer has a need and we are able to respond. Is that within our business domain, or should we be thinking broader into areas of innovation?

Brandon Purcell: 20:24 Yeah, it’s outside in.

Allison Hartsoe: 20:26 Yeah, and I could see how the two work well together, particularly around personalization and segmentation. So earlier, you mentioned that a couple of times. My favorite three-letter acronym, sometimes we call it CLV, sometimes we call it LTV. What role do you see that metric playing when companies are trying to figure out what is the right thing to do?

Brandon Purcell: 20:47 So I believe, and I’m with you on lifetime value. I think that’s an incredibly important metric and can be used as the common currency across business domains that they use to identify, okay, what is this next best experience? Because right now, let’s say we have different propensity models on a customer, and we see that a customer has a high likelihood of turning, they also have a high likelihood of needing a product, and they just had a negative customer experience. We could send them a retention incentive. We could send them a product offer where we could fix their problem. Which one of those things do we do? Right now? It’s kind of like first in, first out, whoever noticed at first maybe they’re going to receive an offer for something at the same time that their service is shut off and that’s just going to anger them even more.

Brandon Purcell: 21:32 So there needs to be something to arbitrate between those different candidate interventions or experiences and lifetime value can be that currency to see given back to the conversation about testing earlier on to look in the past, which interventions have led to the biggest increase in a customer’s lifetime value, and that is the intervention that we should deliver. And in some cases, it may actually be reaching out and fixing that customer service, which actually may have an immediate cost to us so that there’s a short term loss in revenue or bottom line, but in the longer term that may create the type of loyalty that encourages the customer to increase their product portfolio with us and ultimately increase the length of time that they’re our customer and therefore their lifetime value. And this is one thing that I think is going to be very difficult and already is with the company score, the piloting this is moving from short term mindset to a long-term mindset actually sacrificing a conversion for something that may actually prove more profitable down the line.

Allison Hartsoe: 22:28 I so agree with you and like jumping up and down say yes, yes, yes, that is exactly right. But I think it’s very difficult for companies who are under pressure from wall street who’s always saying, what have you done for me lately? And it’s like they can’t escape this dichotomy. Is this putting pressure on wall street to turn around the way they look at companies and be more accepting of these longterm moves?

Brandon Purcell: 22:53 Yeah, it absolutely is. And there are some companies where lifetime value is owned by the CFO of course, who is furnishing all of the metrics for wall street and from my experience, those tend out tends to be the companies that have the most success with lifetime value because they’re able to find this balance between lifetime value for business needs and lifetime value for kind of a gap reporting mindset. And I do think that maybe in the future we’ll start to see some sort of approximation of a future customer equity value among financial analysts who are making big bets on enterprises. We’re not quite there yet, but I do think wall street needs to change its mindset.

Allison Hartsoe: 23:30 Well, I’ll give a quick shout out to Pete Fader’s company because data equity has been working pretty hard to help companies see that and you can see a lot of their cheeky posts about the companies that are out there and what they’re really worth using the lifetime value propensity modeling.

Brandon Purcell: 23:47 Yeah, sure. So I mean obviously Pete Fader is kind of the lifetime value guru and contributed a good deal to my lifetime value research and really grateful for that and will continue to help beat the drum.

Allison Hartsoe: 23:58 Yeah, good. Are there any other examples you’d like to share?

Brandon Purcell: 24:01 So I do want to share one example because I was talking about this difficulty in matching consumer segmentation to customer segmentation, attitudinal psychographic segmentation, behavioral and demographic segmentation. And I did talk to you a pretty large financial services company recently that was able to have some success with reverse segmentation. And so what reverse segmentation is, and you may already know this, but potentially the listeners don’t, is essentially you take the do the attitudinal segmentation on existing customers. So you also have structured behavioral and transactional data on them, and then you transform that data where each row becomes a combination of their demographic channel, et cetera factors. So let’s say just some very simple database. If you have gender, male, female, you have the channel that they primarily transact on. Let’s say you have five channels and then you have age, and let’s say it’s been then to five different age groups.

Brandon Purcell: 24:59 So you would have two male and female times five times five 50 different rows that represent each different potential combination of the values of those variables. And then on the column level, you’d have the average attitudinal score to the different questions that you ask in your attitudinal survey. Then you can perform clustering on that data set to identify, okay, what are the statistical significant groupings of people based upon their attitudes. And now, since you have that at the value level for these different demographic channel values, you can start to actually force that onto your entire customer database. And this was a financial services company that had some initial success doing this. Ultimately they moved towards more propensity modeling to try to identify who these customers were in their database, but they were able to build the business case just using this reverse segmentation techniques.

Allison Hartsoe: 25:53 Oh, that’s a great insight. Thank you for sharing that example. As I think a lot of people would struggle with generally, okay, I’ve got data over here and data over there, and I’ve got different owners and processes. How do I make the bridge? So does this kind of database have a specific name? I think we should give it one.

Brandon Purcell: 26:09 Yeah, that’s a great question. I just reversed segmentation is, uh, the common technique. I don’t know about the database name though.

Allison Hartsoe: 26:16 All right. We’ll call it reverse segmentation magic.

Brandon Purcell: 26:20 Yeah, exactly.

Allison Hartsoe: 26:21 Good. Okay. So, in addition to reverse segmentation, if somebody wants to get started with the proper way to think about segmentation, what should they do first, second, third, what’s their best next step?

Brandon Purcell: 26:33 Sure. So first is determine and articulate your business objectives. Unfortunately, that step just kind of gets missed, or people believe that they already have an inherent understanding of that that’s shared across their multiple stakeholders. And my experience is that’s commonly not the case. And forcing yourself to articulate your business objectives helps get alignment, of course, but also helps you to ensure that you’re not mistaking a means for an end. And so many times I’ve talked to companies where a segmentation is confused and an end as an end, and it’s not just a means to some sort of business objective. So are you trying to grow your customer base? Are you trying to grow your high-value customers? Are you trying to retain customers? Are you trying to differentiate your products and services in increasingly commoditized marketplace? These are the things that you need to really see clearly.

Brandon Purcell: 27:23 And then since segmentation and analytics, in general, is completely predicated upon the data you have. Pick up the right data in bits, freight, what data do you have at your disposal? Make the data wishlist to what data would you like to have? Assuming that whatever projects you take on is successful, well, then you can build the business case to invest in more or better, cleaner data based upon that business objective and the data you have available to you. Then you can start to enumerate the different types of segmentation or analytics projects that are possible. Once you have that list, then you can start to prioritize them. I usually use a pretty simple prioritization framework that looks at feasibility, expected value, clarity. Just assigning each of those attributes a one to five score and seeing what has the highest score at the end.

Allison Hartsoe: 28:10 Well, with CLV, you could actually put a number on it.

Brandon Purcell: 28:12 Yeah, that’s very true. Again, being used as the currency to arbitrate between different options at the enterprise level.

Allison Hartsoe: 28:19 All great points of focus and even though you might in step one have multiple business objectives. I think what you’re saying is that it’s really important to pick what’s the first use of it, even though later on. Sure, you could use it for all sorts of different things. Get everybody on the same page as to what the initial application is. Right. Those are all good steps. Three steps. Do you have any more steps?

Brandon Purcell: 28:41 Well, there’s, there’s action.

Allison Hartsoe: 28:43 Take action.

Brandon Purcell: 28:44 Yeah, and the action is also planning for action too, right? Because yeah, some of this stuff can be difficult from an analytical perspective, but you’re really going to be changing the way that you treat customers, and so there’s a content or communications aspect to that. There’s also in process changes you have to make as well. In some cases, this is going to be delivered through humans within the call center or within a store or a branch, and so there’s training that needs to happen as well. Those things can take a lot longer than building a model.

Allison Hartsoe: 29:14 I think that’s a point worth underscoring because especially if you have a group of people who are changing very quickly, like your call center is constantly turning over staff. The heaviness of that lift to get the training in and get it to stick or deliver it in such a way that it’s very easy for a new person to pick it up is critical. Otherwise, you really don’t have an action.

Brandon Purcell: 29:35 Right. Unfortunately, I’ve seen a lot of these projects fail where the action is going to happen either in the call center or with a sales team and the call center because of the turnover in the sales team just because of a kind of adherence to the status quo, and I think general distrust of analytics just aren’t willing to pick up this new way of doing things.

Allison Hartsoe: 29:53 Yeah. Not invented here.

Brandon Purcell: 29:55 Yeah, right, exactly. Who are you to tell me data scientists about my customer? I know my customer. I talked to them every week.

Allison Hartsoe: 30:02 Yeah, and I don’t want to not see Bob for golf or whatever.

Brandon Purcell: 30:07 Right. Exactly.

Allison Hartsoe: 30:08 All right, Brandon. Well, this has been a really great conversation. If people want to reach out to you, how can they get in touch?

Brandon Purcell: 30:14 Yeah, sure thing. So feel free to email me at bpurcell@forrester.com and then follow me on Twitter at BCPurcell.

Allison Hartsoe: 30:24 Oh, excellent. Cause we will do a lot of posting of this episode through Twitter. We’ll be sure to reference you there as well. Fantastic. Great. As always, links to everything we discussed are at ambitiondata.com/podcast. We will link to Brandon’s paper as well as to his profile, and thank you, Brandon, for joining us today.

Brandon Purcell: 30:43 Thanks so much for having me, Allison. This was a lot of fun.

Allison Hartsoe: 30:45 Remember when you use your data effectively, you can build customer equity. It is not magic. It’s just a very specific journey that you can follow to get results. See you next time on the customer equity accelerator.