collage of team working together with colorful overlapping circles in background

How Advanced Analytics Boosts Customer Engagement For USAA

by Radius

USAA relies on a strong and specific advanced analytics program to optimize customer journey, customer experience, communications, and more. Watch this webinar to learn more about their approach and why it brings so much value to their research program.


Bari Weinhausen:

Welcome today to our Radius webinar. Today we are going to be hearing all about advanced analytics and how advanced analytics can be used to heighten your research insights, heighten your results, broaden your learnings. We’re going to hear more and more about that from our experts today. We have Michael Patterson with us, and we have Jay Bigler from USAA who will be joining us. I will be moderating a conversation with these gentlemen to understand how advanced analytics is used, when it’s used, why it’s used, what it really brings to the table, and really help you understand some opportunities that maybe you’re missing out on in terms of incorporating advanced analytics and studies that could indeed benefit from them.

So with that, I’d like to have folks introduce themselves and then we’ll get started. As for myself, my name is Bari Weinhausen, Director of Qualitative Research here at Radius. I’ve been with the company for a shocking 27 years and when I’m not moderating or leading some sort of qual expedition I like to spend time with my dogs and on the beach. I’m in South Florida, so hence the beach being a natural outlet. Love to start with you, Mike, and then we’ll throw it over to Jay, and then we’ll get started with our conversation.

Mike Patterson:

Sure. Thanks Bari. Mike Patterson. Glad to join you guys this afternoon. I’m director of research analytics here at Radius, and I get involved in a lot of different studies. I work and consult with clients. I also head up our innovation space, and then I also have the opportunity to do lunch and learn and webinars with Radius. So we’re hoping you’ll enjoy this session.

Bari Weinhausen:

Thanks, Mike. Hi, Jay.

Jay Bigler:

Hello, Bari. I’m Jay Bigler with USAA and I’m a lead research consultant for them. I joined USAA about five years ago, and initially focused on more enterprise level projects. And then the past couple years I’ve segued over to the banking space and I’m currently doing marketing research for the bank business and do things like manage the brand tracker programs and a number of other things that we’ll get into here a bit later.

Bari Weinhausen:

Perfect. And right before we dive in, just a couple housekeeping items. For the folks who are listening in, I do believe you all know that this is being recorded. We will have an opportunity for you all to share questions. Please, as you’re listening to the conversation that we’ll start just momentarily. Please go ahead, send questions into the chat, and as we wrap up, we will do our best to address as many of those as we can. Our conversation today with Mike and Jay should be about 30 minutes of our time, and then we’ll have about 10 or 15 minutes for Q&A as we see them come in. So, to get us started, Jay, tell me about the research that you tend to conduct. What kind of projects do you work on?

Jay Bigler:

As you might expect it’s a full marketing research function, so we’re not pigeonholed into any particular aspect of marketing research. Our projects span the complete scope of what I would refer to as the classical marketing mix. So the four Ps of marketing: price, promotion, product, and place. We do a lot of research in the way of product and concept testing, a little bit in pricing, a little bit in channel or place, and then quite a bit in the promotional space. Our research function is housed in the marketing and brand management org structure of USAA and so a lot of our work focuses on marketing tactics, marketing brand, advertising campaigns and that sort of thing. So we do a lot of that type of research and then we also do a good deal of strategic research which supports the overall strategies, marketing strategies, and business strategies. So, for example, the brand tracking research I mentioned would fall within the strategic domain as would segmentation research that we recently completed and some customer journey things like that.

Bari Weinhausen:

Fantastic. And do you have a sense of whether it’s the percent or the proportion of studies that you do work on that involves some sort of advanced analytic component?

Jay Bigler:

Yeah, it varies quarter to quarter, year to year but I’d say probably a little over half, Bari, involve advanced analytics.

Bari Weinhausen:

All right. I’m definitely going to hit you back up to understand the why. I mean, that’s certainly a lot of advanced analytics that you’re incorporating, and I’m going to want to dig into that. But before we do, Mike, you’re in a somewhat different role. You’re on the supplier side, so I’d like to hear a little bit from you about, we’re talking about advanced analytics. What are they? How do they work?

Mike Patterson:

Yeah, good question, Bari. So really in a nutshell, advanced analytics, I think of them as being sophisticated techniques where we’re able to look at the relationship between two or more variables. So, for example, imagine that we want to understand how important different factors are when individuals are trying to consider or choose a credit card. We could simply ask them to rate the importance of those different factors using a one to seven scale of importance but if we did that, what we would likely find is that they rate all of those factors being the same.

So if we’re testing things like the brand of the credit card, the annual percentage rate, the rewards that they might get for using the card, chances are, because all of those things are important, we’re really not going to find much differentiation between those items. And so instead what we could do is use advanced analytics like a technique such as, say, driver analysis to measure how much impact the different factors had on an individual’s willingness to adopt a particular credit card. And if were to do that, I suspect what we would find is much better differentiation and really, frankly, just better results.

Bari Weinhausen:

Great. So you can really see what’s rising to the top. Tell me a little bit about the types of projects that you tend to work on. Like, as Jay said about half of his projects do incorporate some kind of advanced analytics. Give me a sense of the kinds of things that you see a good fit for advanced analytics.

Mike Patterson:

We really have a wide range of projects that I get involved in here at Radius. So I mentioned key driver analysis, and that really involves a variety of different regression or regression-like approaches. I also get involved in segmentation studies. And segmentation studies allow us to use different clustering techniques to group together individuals that are similar in terms of, say, their needs or their attitudes or their behaviors. So we’re able to find individuals that all are somewhat similar.

I also get involved in a lot of choice models. And choice models are approaches that allow us to understand the importance of different features, including brand and price, when individuals are evaluating different products or services and trying to choose amongst the different offers. And so we’re able to understand what that decision process is, and then we can use the results to provide our clients with guidance so that they are really able to design those products and services to maximize adoption or maximize the revenues that they get.

And then the final technique I think that I get involved in a lot is called MaxDiff which is like a trade off technique that allows us to understand the relative or the absolute importance of different features. So for example, going back to that credit card example, we could conduct a MaxDiff study to understand how important those different factors were when individuals are considering a credit card.

Bari Weinhausen:

Perfect. Thank you. And just to round out our start point to understanding advanced analytics, under what kinds of circumstances do you tend to recommend and/or involve some sort of AA?

Mike Patterson:

Yeah, so it’s really when we’re needing to get deeper insights into data, then we might be able to get it simply by asking more relatively straightforward questions. So in other words, advanced analytics really allows us to explore data to a much greater extent and drive those insights that you simply can’t get using simpler descriptive statistics.

Bari Weinhausen:

Perfect. And now I want Jay, coming from your perspective on the client side, tell me why you do like to use advanced analytics in your studies. What does it bring to the table for you?

Jay Bigler:

Yeah, well, Mike just took all my talking points.

Mike Patterson:

<Laugh>, <laugh>.

Jay Bigler:

Just kidding. Yeah, it would be a bit redundant. Personally, I prefer if you can answer your research objectives with just good old straight descriptive statistics. I prefer that because it’s just easier to digest and obviously costs less and can be more clearly communicated. But often that’s not the case. So particularly when you have a lot of variables involved in your questionnaire and you want to solve for something that is not just readily available from the straightforward descriptive analysis from the research. So, as Mike said, fundamentally you’re using advanced analytics to look for relationships or patterns in a range of variables that you’ll have in your dataset. But you’re doing that to solve for something.

So for example, key driver analysis, you may have a study where you have 15 or 20 brand attributes and you want to determine what brand attributes are most important in driving brand consideration? Well you can ask for stated importance, but chances are every one of those is going to be stated as extremely important. That gets you no good, and you can’t really rank effectively 15 or 20 items. So often we’ll use, and I’m sure the audience is quite familiar with this, but derived importance and derived importance calls for the use of some sort of advanced analytic technique. As Mike said, there’s numerous ways to solve for derived importance, whether it’s ordinary least squares regression or just simple correlation analysis. There’s nothing wrong with that if that gets you to where you need to be but there are also some additional techniques that are worth <inaudible>.

If there’s high multicollinearity in your independent variables, then things like ridge or lasso regression can be used to help mitigate that impact of multicollinearity. And then there’s also base analysis, but basically the fundamental, and we’ll get into some more of that later, but basically the fundamental reason you use advanced analytics is to explore and solve marketing questions that involve a large range of variables.

Bari Weinhausen:

Perfect. Mike, anything to add to that in terms of the reasons one might want to use advanced analytics?

Mike Patterson:

Yeah, I think just adding to what Jay was just saying, I think really because advanced analytics are these powerful tools, it really allows us to probe our data to look for relationships and explore in depth so that we’re trying to maximize the value that we get from the data that we’re collecting.

Bari Weinhausen:

Great. Perfect. So what about in terms of advanced analytics, what does it provide that you don’t get on studies where it’s not utilized? I keep hearing it allows us to see more. What does it allow you to see? What are the key benefits? What does it bring Jay?

Jay Bigler:

Well, take choice analytics, so discrete choice or MaxDiff is a choice technique. There’s no way to tackle when you’ve got a discrete choice application where you’re trying to design an ideal product. There’s no way to have a respondent address all the combinations of product scenarios just in a descriptive fashion. So in that case, you need to set up an experimental design that shows subsets of those combinations. But under the hood of all that is discrete choice modeling and hierarchical bayesian modeling that gets you to where you need to be with that. They’re incomplete block designs, basically. And that’s the case with MaxDiff and with conjoint or discrete choice. So that would be one reason you would need to use an advanced analytic technique because you simply can’t get there with just a straightforward questionnaire.

And then segmentation, which I can get to and explain how we use that later but used advanced analytics for our segmentation work. But when you have a postdoc segmentation based attitude and behaviors I mean, you can’t get a solution unless you employ some sort of an advanced analytic algorithm to help you develop that solution to that.

Bari Weinhausen:

Perfect. And Mike, tell me a little bit more about some of the benefits that you’ve seen across the board.

Mike Patterson:

Yeah, so I think one of the real key benefits is that when we use advanced analytics, it allows us to better test hypotheses that we might have concerning either a market or data. So for example, Jay just mentioned segmentation. So we might hypothesize that there are different groups of consumers in a market, and so we can use advanced analytics, different types of clustering techniques, and approaches to determine both which groups exist in that market, and then understand how they differ from one another. So this allows us to confirm if these groups exist and if they do, what do they look like? And so we can either confirm or refine those hypotheses that we have using advanced analytics.

Bari Weinhausen:

Great. And the other side of the coin, Jay, what are some of the watch outs when using advanced analytics? What are some of the things that you keep an eye out for, that you look out for?

Jay Bigler:

Yeah you always gotta be watchful, that’s for sure. Well I guess the first thing is personally I hate black box techniques. And you’ll have suppliers that will engage with you that offer the latest and greatest but aren’t real transparent about what’s under the hood on that technique and that’s not going to be any good in my opinion. You need to understand the technique and have a supplier that will explain that to you and explain how the algorithm works and the pros and cons and strengths and weaknesses. So I think that’s an important watch-out. And so you also need to have a supplier.

A lot of times particularly with large research suppliers, they’ll sell in with their A team and then when you get the project execution, get the B team and the B team really doesn’t understand a lot of the time the analytics behind what they’re delivering. So you need a good research partner that understands, is going to take the time, is not concerned about spending the time with you for Q&A and taking the time it needs to explain the different techniques and the different options that you can bring to the table. And then beyond that, once you get the results, the watch-outs are you need to make sure that you can take the insights delivered from that technique and transform those into a story that you can deliver to your internal stakeholders that are not researchers and they don’t care really about the technique.

Particularly some do, but generally speaking, they don’t because that’s not what their job is. So they’re interested in “how does this solve my marketing problem or my business question.” And so basically you have to make sure that the deliverables on the analytic techniques are communicated in a way that is clear these are the insights that address the marketing question. Now, you’re always going to get somebody in your audience, not always, but almost always that is a bit more analytical and/or just a bit of a contrarian and they’re going to pop up and ask, “How’d you get there? What are you showing me? This kind of looks funny. How did you get from A to B?” So you gotta be prepared. And so I always encourage the research folks to do their homework and anticipate that question and make sure they can answer that because what they are going to do is say, “I don’t know, I’ll get back to you.” Because it hurts your credibility.

Bari Weinhausen:

Got it. For sure. Mike, what do you have to add? What do you watch out for? What do you look out for?

Mike Patterson:

Yeah, so because these are really sophisticated techniques, there are assumptions and stipulations that underlie the type of data that you have. So you really have to have a good understanding of the assumptions that underlie the techniques, and then know how to also address different issues that are going to arise otherwise, what you’re going to end up with are erroneous results. So to me, the big watch-out is really make sure that the vendor, the statistician, the data scientist that you’re working with, really has the skill-set and the knowledge in order to use and conduct these types of analyses because really, there’s software out there today that allows pretty much anyone to run different types of advanced analytics, but if you don’t know what you’re doing, you get yourself into some real trouble. So you might estimate a model and get results that really are just not accurate and that could lead you in the wrong direction.

I guess the other thing I would add is that these advanced analytic techniques generally require larger sample sizes, which can then increase the cost and because of the complexity of the analysis can also add or expand the timeline. So basically because of the complexity and the time involved in running the analyses, the analytics themselves can increase the cost of the project and just expand the timeline.

Bari Weinhausen:

Perfect. Jay, why don’t you share with us some recent studies where you did commission advanced analytics. I just want to understand the big picture, broad brush strokes in terms of what were some of the study topics, what was it that you were trying to learn, what was the approach that you took in your research, and why you chose to include advanced analytics and what it delivered? Give us a little mini case study, if you will.

Jay Bigler:

Yeah, happy to do that. I’ve got a few examples here I can quickly go through for you. One thing I’ll interject is these advanced analytic techniques, they do involve both judgment and the science, if you want to call it, behind the actual algorithm. So that’s why you need to bring your research skills to the table that bear judgment upon the data and the results from these advanced analytic techniques. So I just want to drive that point home. But we’ve used advanced analytics quite a bit, well, in particular MaxDiff quite a bit for prioritization of items.

Now those items may be product features and benefits, they may be advertising messages, they may be a long list of names or taglines that you’re trying to evaluate. And so again, MaxDiff falls within that choice methodology and experimental design methodology. I’m sure everybody in the audience has used MaxDiff and we take it for granted but there are some watch-outs for that too. I’ve seen some cases where the items you include in your MaxDiff set are really not apples to apples. And because everything is evaluated relative to one another in the item set, if you got 15 or 20 items in a MaxDiff exercise or sometimes more you’re not going to get good quality readout unless those items are all on the same level playing field. So I would encourage the audience to really scrutinize the items that they’re including in their MaxDiff exercises.

And the other thing, it’s not always possible, but the shorter and more concise those statements can be, and the more written in consumer understandable language, the better off you’re going to be. So we’ve used MaxDiff a lot. We’ve talked about segmentation quite a bit. We recently completed a segmentation and it was a post hoc segmentation versus priori, if you remember your research textbooks. But so we had attitudinal and behavioral questions in the questionnaire. We had life stage questions in the questionnaire. We had financial health questions as well as demographic questions and things like that.

So anyway, on the back end, you have to have some algorithm to group those respondents into segments. Now those segments need to satisfy some conditions. They need to be reachable segments. They need to be actionable segments. They need to be segments that will respond to your marketing messages, and they need to be sustainable because it doesn’t do good to go through one of these segmentation studies that can take you three months and put in all the time if it’s fleeting and reasons why that could be done. So anyway, you want to solve for segmentation solutions that get you there. And again, it’s a lot of art or judgment and science mixed together. But some suppliers will just have an approach to their segmentation studies and throw out a K-means segmentation to you and a [inaudible] solution or there’s some other tech latent class or whatever. Actually we worked with Radius on ours, and we tried probably five or six different segmentation algorithms before we landed on the final one we selected, which we thought told the most coherent story.

My experience has been, and Mike can give his point of view, but some of those traditional clustering algorithms, I’ve just never found that much success with them. So we landed on one, it’s non-negative matrix factorization, which we felt did a good job and Mike can always provide a paper on that if you’re interested and you’re really into the algorithms and the tech [inaudible]. But it’s interesting because it really combines a factor analysis approach with a clustering approach simultaneously and I think it gets you into a good place. So conceptually, I like that. So anyway, there’s that for segmentation and on the back end, keep in mind for a segmentation to be actionable, you really need a good solid typing tool, and that’s going to require advanced analytics as well. So in our case, we used a random forest typing algorithm coupled with a backup discriminate analysis algorithm.

Finally, not to make this section too long, but we use advanced analytics a lot in our brand tracking program and just briefly for example, we’ll use it to look at key driver analysis. And so that again, gets at things like, what’s the best approach to do that regression, correlation, et cetera but with high multicollinearity, we had a lot of binary variables as well. So we landed on base analysis, which is a probability based approach, which works well with binary data and intercorrelated data and that’s what we’re using for our key drivers analysis. And then one final thing on the brand tracker is we’ve used advanced analytics as well to look at the brand structure brand attribute structure.

So again, if we’ve got 15 or 20 or 25 brand attributes, how do those all interrelate with one another and how do they work to drive or how are they related to driving a key dependent variable like brand consideration? And so, you could do that with a structural equation modeling approach, for example but that requires a priori imposition of theory on that, which I don’t like. So we tried the Bayesian Labs approach which some of you may have heard of. I was hopeful and optimistic that it might work quite well, but it ended up not telling the greatest story. So we ended up using a graphical lasso approach, which is a combination of factor analysis and then lasso regression to look at that brand structure analysis so that’ll give you…

Bari Weinhausen:

So what I’m hearing is it might take multiple approaches and it’s about working through to find what delivers the best solution for you.

Jay Bigler:

That’s absolutely correct, and Mike can speak to that. But yeah, if you just go in with one canned approach you’re not doing a thorough job.

Bari Weinhausen:

And let me dig in a little further. Joe, you mentioned, and I’m glad to hear that you do turn to Radius for these needs some of the time, maybe more than some. But in any case how did you land on Radius or any of your other vendors? What are some of the considerations, the important factors to be thinking about when choosing a partner to provide advanced analytics?

Jay Bigler:

Well, I’d encourage you to work with some different suppliers and find one that you develop a trust and confidence in and a strong partnership and working relationship, number one. But you’ve got to vet out your partner based on technical competency number one, and also vet them out that they have access to a range of tools to bring to the table. And that’s not only an understanding of different approaches, but the software to execute those techniques. And then, as we spoke at the outset, you want a partner that’s going to not be concerned about spending time with you and logging that as hours on the project and coming back and saying, “hey, we got 10 hours of consultation here, so we gotta up the charge on this project and that.”

Now, obviously, if you get outta scope with something, then it’s perfectly expected that a supplier’s time is money and all that, or scope is money, but you’ve got to find a supplier that’s willing to work with you to spend the time to explain those things. And again you’re going to get approached, folks in the audience, you get approached every day with suppliers trying to sell you the latest and greatest. And I would be a hard sell. I would probe and probe and probe and clarify and clarify and clarify on all that stuff before you go off and bite on something like that.

Bari Weinhausen:

Great. Mike, what would you add to this list in terms of recommendations that you might make to folks who are considering including advanced analytics in their next study? What should they be thinking about?

Mike Patterson:

So we’re really always cautious when we’re recommending advanced analytics because we really do want to ensure that the tool technique approach that we’re going to use is really going to give us additional value or insights beyond just simply asking more straightforward questions. So we don’t want to include these types of techniques or approaches simply to include them. Because, as I talked about before, anytime you include some sort of advanced analytics, it’s probably going to require larger sample size, additional cost, and additional timelines. So, we really do take a critical eye when we’re recommending to make sure that we’re going to get the value from the technique that we want to.

Bari Weinhausen:

And anything additional to add for those folks who may be listening who haven’t really ever considered advanced analytics, it’s maybe not on their radar, but maybe they should be thinking about it.

Mike Patterson:

Yeah, I definitely think that there’s just huge value in including advanced analytics in a study. But if you’ve not used advanced analytics or you’ve not used a particular technique when you’re working with a vendor, just as Jay was saying, take the time and have that vendor explain the approach to you. Explain the benefits that it’s going to provide in the study, what are you really going to get out of that technique? And then also talk to you about the drawbacks, because there’s always consequences to including these types of techniques.

Bari Weinhausen:

Got it. Excellent. Folks, we’ve been talking for a bit of time already. It always goes by fast in qualitative research. But just one last question as we do begin to wrap up our time together. At the end of the day, what is the number one reason why one might want to consider including advanced analytics in their research? Really just kind of summarize what we’ve heard today. Jay, if you would get us started summarizing, and then Mike, go ahead and add.

Jay Bigler:

I think I’ll repeat myself a little bit here, but it’s basically the need driven by the need to look for those underlying relationships and dig deeper into the data and to solve for a research question that is just not surfacing at face value from the descriptive statistics. So, that sounds pretty lame maybe, but that’s why you would do it. And we have a lot of trouble. Field costs are going up all the time, and particularly for us, because we’re a membership based organization and the incidents of our target audiences are pretty low out there. So our field costs are quite high, and it takes us a lot of time and money to collect the data. And so we need to make sure we’re making the most use of that data. And advanced analytics offers the opportunity at least to leverage that data in ways that maybe didn’t immediately come to mind or to dig deeper into those insights.

And then finally, the end of the road is not the research itself, but it’s how those research insights impact and help and support marketing strategy and tactics or business strategy and tactics. So the calling we have is to deliver the insights that circling back to the end are going to bring value and support those decisions that have to be made.

Bari Weinhausen:

Got it. Mike, last couple words, and then we’ll turn it over to Paul for some questions.

Mike Patterson:

Sure. Yeah, I totally agree with Jay. I mean, not to reiterate what he said, but really advanced analytics allows us to maximize the value that we get out of data and then in turn the insights that we’re able to provide to our clients based on that data. So it’s just a collection of very powerful tools that really add a lot of value to research.

Bari Weinhausen:

Fantastic. And it looks like we have some questions coming in. I want to thank both of you, gentlemen, for your participation and your time. I am going to open this back up to Paul who can start to share with you some of the questions we’ve received. Hi, Paul.

Mike Patterson:

Oh, Paul, you’re on mute or something.

Paul Donagher:

Hi folks. Sorry. There you go. Yeah, I’ve been monitoring the questions. We’ve got a lot of questions here for you guys. I’m going to jump straight in. Jay, there’s one, I think for you in particular. You had mentioned having shorter statements either in choice studies or as attributes or using consumer language. One of the questions was, “do you advocate for using qualitative to help produce those before you get into quant? Is that something you guys consider at USAA?”

Jay Bigler:

Yeah, we’ve done that before, and Paul more often than not, we’re using the qualitative to come up with the statements to begin with. Whether it’s a focus group or discussion board or whatever format you’re using to parse the language in that. Although we have, I mean what’s the best way to express this and all that. And then you can glean from that maybe ways to pair down the wording. But, this may not be that helpful to the questioner, but I think the burden more falls on the brand team, you as a researcher, working with your internal stakeholders to say, “okay, well, I know this is kind of the marketing speak and this is how we’re articulating it, but you’ve got a three or four sentence reason to believe here and we’ve gotta shorten that down.”

So how can we communicate the essence of this message or reason to believe, but get rid of the extraneous words and superfluous words and just shorten it and also push back and say, “Hey, assume you don’t know anything about banking Mr. Stakeholder, Mrs. Stakeholder, if you were to read this statement, would you understand what it means?” And almost always I get that, “well, no, honestly, I would not”, so, well, that’s the point. We’re serving this up to respondents, so we’ve gotta tweak this language so they’re going to understand it.

Paul Donagher:

Got it. I do have another one for you here. This one’s from Tim. Jay, you said a segmentation should be reachable and sustainable, and Tim says, is there anything else other than those two attributes that you recommend for any segmentation to make sure that it checks off that box?

Jay Bigler:

It needs to be reachable, meaning you can identify who these people are, and that’s why it gets to be tough because if you’re doing an attitude and behavioral segmentation, it’s going to tell you attitudes and behaviors, but unfortunately, it’s not going to tell you a lot about who they are. So you’ve got to roll in demographics. And so there’s always a trade off between the variables you’re throwing into your model. But you almost have to throw in some demographics because when it comes to targeting you can target to some extent on attitudes and behaviors, but you’ve got to have some other variables to try to target. So that’s reachable. And the typing tool that I mentioned also comes into typing. So I don’t know that he asked about reachable, but I wanted to further explain that.

So you got reachable and sustainable, but they have to be actionable. And what I mean from that is whatever attitudes and behaviors and the value proposition and messaging associated, what’s going to resonate with those segments has to be able to be delivered by the business. And so if a particular segment is saying in effect, “I want this”, but you know it’s not for your business to be able to deliver on that, then it doesn’t mean you throw that segment out. But it means that segment is clearly not going to be one of your priority segments.

And, the last one I think I mentioned was you have to have segments that are differentiated enough that the whole idea of segmentation from a marketing standpoint is that they’re going to respond relatively better to customized messaging that’s targeted to those segments respectively. So you need to make sure that the messaging you’re going to deliver they’re going to be receptive to that. And that’s the relevance or the receptivity of it. I hope I answered that.

Paul Donagher:

No, I think that that answered that pretty fully. Mike, anything that you would want to add to that?

Mike Patterson:

Yeah, just one thing. So, the other dimension I would say with segmentation is measurable. And, essentially what I mean by that is we want to make sure that when we come up with segments, that those segments are not too large, so they don’t comprise a 50% of the market, because if you have a segment that’s 50% of the market, it probably needs to be broken apart. So it’s just simply too large. Nor do you want segments that are so small that they’re really not actionable. You don’t want a segment that’s only 5% of the market, because unless there’s something just incredibly unique or valuable about that segment, they’re not sufficiently large enough to justify going after. So you want sort of a sweet spot. So you really need to be careful when you’re extracting segments and look at those segment sizes.

Jay Bigler:

Yeah, that’s a good point. If you got a market of a hundred, just hypothetically, right? The most minute segmentation would be a hundred people because we’re all unique but that kind of defeats the whole purpose of segmentation. So if you extrapolate that to a market of 40 million or whatever your target market is there’s a fine line between how many segments, back to Mike’s point. And so that’s where that kind of art and science or judgment and science is mixed together. Yeah, because once you get beyond a segment of one, no segment is truly a hundred percent homogeneous.

So you’ve got a segment, and the idea is there are more homogeneous within and heterogeneous among. And so it’s kind of a probability game. So we’ve got this segment, they tend to be different from these other segments that have unique needs. And not every single one of them are we going to hit spot on, but the lion’s share of that segment is going to respond to this message that showed up in the segmentation results,

Paul Donagher:

Understood completely. Can’t be a hundred percent but you want to get as close to that as you possibly can.

Jay Bigler:

Right.

Paul Donagher:

That’s all we have I think for today, folks. Thanks for all your questions on the screen. My thanks also to Bari for moderating, for Mike, for your, and Jay, particularly for you to be involved in this. This was our third webinar of the year. We’ll do one more in Q4 and we’ll get together and we’ll send out some introductions and invites to you all then. So thanks for joining this one and we’ll be good for the next one that we put on. Thank you.

Jay Bigler:

Okay, everybody, take care. Thanks all.

 

Bring greater impact on your research program with Advanced Analytics.

Contact Us