The Truth About Segmentation

By Ginger Conlon | 7.10.18

Ah, segmentation. Most marketers swear by it. That doesn’t mean they agree on, well, anything about it: what makes it valuable, how to get the most from it, the best way to segment customers, whether one-to-one marketing is scalable. And that’s OK. The beauty of segmentation is that it’s unique to the organization using it. That differentiation allows marketers to get a distinctive view of customers and prospects, demonstrate their understanding through more relevant communica-tions and interactions, and even create a competitive advantage in the process.


“Marketing is much more effective when the message it targeted. We’ve proven that time and again,” says Michael Cohen, VP of marketing at eRelevance. “And, today, customers expect you to know them. They can easily get offended if you’re off target because they think you should know better.”


In fact, customer engagement is one of the common themes that emerge from all the varied perspectives on segmentation. The majority of marketers we spoke with for this article, for example, agree that segmentation provides value to customers. “Being able to understand individual customers, their preferences, their needs, their aspirations — there’s a value for the consumer in terms of relevancy and trust,” says Patrick Tripp, VP of product strategy for RedPoint Global. “And there’s value to the brand in terms of being able to drive revenue, relevancy, and loyalty.”


Most marketers also agree that, as much as you’d like to communicate with every customer individually, you can have too many segments. Many marketers we interviewed emphasize that segmentation is an ever-evolving process. They agree that there’s always more to learn about what you can accomplish with it and how it can improve your marketing.


“Segmentation is really about optimization and understanding what’s working for your business,” says Victor Wong, CEO of Thunder Experience Cloud. “It’s about finding where you’re doing best and doing more of that, as well as, obviously, finding where you’re doing worse and doing less of that.”



What’s the value of segmentation?

In today’s environment of ever-increasing customer expectations, one of the biggest reasons to use segmentation is to provide a better customer experience—which, in turn, will lead to better marketing outcomes.


“Without segmentation, everything just gets lumped together into one mythical average,” says Thunder Experience Cloud’s Wong. “As an example, if casinos and mobile games didn’t use segmentation, then every player would just get the same treatment, which means that your high spenders feel underappreciated, and it’s possible that your low spenders are just completely unprofitable, so you’re leaving money on the table.”


That’s why segmentation is so valuable for identifying commonalities and differences among customers — especially high-value customers. It enables marketers to help drive sales through improved messaging not only to those best customers, but also to those most likely to become customers, notes Joe Camacho, CMO of Sabio Mobile.


Segmentation’s value increases as marketers move beyond basic forms of segmentation, such as demographic segments. Segmentation using demographics is easy to understand and implement, it works, and it cuts down on waste, which is why nearly all marketers use it, Cognitiv CEO and founder Jeremy Fain points out. Behavioral-based segments are more advanced, but often are still boiled down to one descriptor, such as sports fans or coffee drinkers, he says.


Segmentation becomes most valuable when the segments are based on multiple factors. “I’d rather spend my money delivering my message to people who actually have a chance of being interested in my product,” Fain says. For instance, if a marketer is trying to encourage baseball fans to purchase her company’s sport apparel as gifts, instead of creating a segment defined as “baseball fans,” she might create one called “baseball fan gifters,” and include baseball fans who buy apparel in-store and online in multiple sizes and during the holidays. “Segments, especially online segments, have given us an opportunity to be more efficient and precise with our targeting,” Fain adds.


Segmentation increases efficiency and effectiveness when targeting users. Marketing today is about reaching the right consumer at the right time in the right place. “Without segmentation and data to support that targeting, you can’t do that unless it’s a very broad awareness campaign,” says Parker Morse, founder and CEO of H Code. “Segmentation allows marketers to be more effective and reach that right user, as opposed to running against probability-weighted or inferred models, where they might reach this user here or there.”


That increase in efficiency and effectiveness can have a significant impact on the business. Xanterra Parks & Resorts is one example of a company seeing increased revenue as a direct result of segmentation, RedPoint Global’s Tripp points out. The travel and hospitality company created segments based on nine personas — everything from the wanderlust-adventurer to the budget-conscious empty-nester — that were derived from unified data from across its numerous brands, including Windstar Cruises, Grand Canyon Railway and Hotel, Yosemite National Park Lodges. “By applying those persona-based segments to email campaigns, Xanterra was able to do 45 levels of personalization per email,” Tripp says. As a result, Xanterra increased its email-generated revenue overall; in select email campaigns the increase was a more than 1,000% jump in revenue by using segmentation, Tripp explains. “It helped increase revenue per email from 20 to 30 cents to over a dollar,” he adds. “That’s just scratching the surface of what’s possible.”



Is there any scenario when segmentation isn’t valuable?

Most marketers agree that segmentation is indispensable. “You always will have subgroups of customers, so segmentation will always be valuable,” Thunder Experience Cloud’s Wong says.


Some can’t emphasize its importance enough. “It’s irresponsible not to segment your target customers, even for really big brands like a Coca-Cola,” Richard Black, CMO of Momentum Worldwide. “There are segments that brands need to reach, whether that’s millennials, or Hispanics, or a different slice of an audience. Segmentation exists in all businesses.”


In some cases, however, segmentation may not be useful or effective. The most obvious is “if the goal is to reach everyone with the same message,” says Sabio Mobile’s Camacho. One example is Starbucks’ recent email to all of its opt-in customers about its afternoon store closures for sensitivity training and the reasoning behind it. “Something where the desired action isn’t dependent on market demand,” Camacho adds, “but is needed by everyone despite their immediate intent to purchase.”


Sometimes marketers might skip segmentation when mass marketing is relatively inexpensive. “It’s less valuable when there’s very little cost to promote broadly,” says Alec Casey, CMO of Trusted Media Brands (TMB), which publishes Reader’s Digest, The Family Handyman, Taste of Home, and numerous other category-leading publications. “For instance, the cost to send an email is very low, so we don’t get into granular segmentation for those campaigns; we just send them to the people most likely to respond. Now, direct mail, where it’s a lot more expensive? We spend a lot of time building complicated models to segment finely.”


Congnitiv’s Fain asserts that retargeting is the only time segments are absolutely essential; specifically when retargeting people who have already shown interest in a company’s website. “That’s not a demographic, that’s not a behavior except that, ‘These people have shown interest in my site,’ so they’re likely to drive my revenue forward,” Fain says. “It’s a no-brainer.”


Today’s privacy-centric sentiment is a reason to be cautious with segmentation, warns RedPoint Global’s Tripp. Targeted marketing shouldn’t take a consumer by surprise, he says. Marketers don’t want customers thinking, for example, “How did they know this about me? How did they know I was a gluten-free customer?”


“Capturing that data is clever, but then using it can actually be quite disruptive and create friction in the experience,” Tripp says. “It’s a delicate balance not to overstep your bounds as a brand. That line is gray and varies by industry.”


Tripp’s advice to marketers is to always be explicit in the way they collect and use data, and then be implicit with personalization. “There needs to be some art,” he says, “in terms of subtle recommendations, subtle hints that we’re guiding a consumer along the right path without hitting him over the head with the fact that, ‘Hey, we know this about you, isn’t that great?’”



Are there ways to segment customers that are better than other ways?

Some marketers stick with demographics for their segments; other marketers prefer more behaviorally based segments. Many take a more blended approach, like Xanterra’s persona-based segments. “There aren’t better ways to segment, just different approaches,” says Momentum Worldwide’s Black. The “best” way to segment customers is, of course, whatever way works best for your company

and customers.


“The ideal way is to segment based on a customer’s explicit propensity to do something,” says Thunder Experience Cloud’s Wong. “When I say explicit, I mean a revealed preference through action. For example, how often are they buying from you already, how often are they visiting you, are they explicitly showing interest in a particular thing?”


When marketers don’t have access to that type of data, they should use the next most observable data, Wong advises. That next-best data tends to be demographics such age groups or location or other proxies for interests that consumers might be expressing, he adds.


H Code’s Morse also emphasizes the important of explicit triggers as ways to create segments that will have the greatest business impact. He recommends that marketers look at the context of the content a user is consuming to help understand that person’s mind-set. “With context like location data, you can be much more effective in reaching the user at the right moment than just reaching the right user,” he says.


Understanding behaviors, context, and interests and using them for segmentation is especially important for multichannel campaigns, notes eRelevance’s Cohen. Marketers should be “listening” and building profiles for their target audiences that include channel preferences, he says. “If you’re doing email only and getting a 20 percent open rate, that means 80 percent of your customers didn’t see any message [beyond your subject line]. You’ve got to include multiple channels in your segmenting, so your reach is extended.”


As a publisher and subscription marketing company, TMB is all about direct response marketing, Casey points out. “When we’re talking about segmentation, we’re not using conventional demographics or even psychographics,” he says. “We look for transactional information to drive our segmentation.”


That means the most important piece of information Casey and his team need is what type of marketing each individual customer most recently responded to. “Is a customer direct mail responsive, email responsive, insert card responsive; what spurred a transaction? That’s the most important segmentation I have,” he says. “Next is knowing where you are in your relationship with me. In other words, if you’re renewing, we’re going to treat you differently than somebody who’s never responded to an email from us in the past.”


TMB still use attributes such as demographics and affinities when it’s appropriate and its marketers don’t have better transactional information, Casey notes. “Transactional data is so useful is because it’s behavioral,” he adds. “We find that what somebody says they want to do, or says they’re interested in, or because of who they are could be interested in is much less powerful than knowing what they actually did. That’s where the transactional data gets to be so powerful.”



How do you know if you have the right segments?

It’s not uncommon for marketers to uncover a new or an unexpected segment of customers when tracking campaign performance. And it’s all-too-common today

to have rivers of new data flooding in that prompts the need to rethink existing segments. Yet, some marketers “create segments and use them forever,” says eRelevance’s Cohen. “Segments can change for each campaign, and that’s OK. They’re not a persistent object.”


The key, of course, is knowing when to assess whether your current segments

are stale.


“We know that when we build a model, which is typically the way we segment, it usually has a lifecycle of about two years if it’s a good model,” TMB’s Casey says. “And what that means is as information evolves, as the sources of information evolve, or even as what we’re doing in our own marketing evolves, the factors that are important now will probably stay important, but they may become less important than another factor. So, as your ways of speaking to the customer evolve, your models have to evolve, too.” In other words, Casey notes, when you start seeing your segments’ response fall off over time, it’s time to rethink them.


Sabio Mobile’s Camacho concurs that campaign performance is a key indicator of a segment’s viability. “If your campaign is performing above industry benchmarks and the data supports it, and the campaign performance affirms it, then you know you’re right on target,” he says.


One recommendation is to use lift metrics to determine whether it’s time to update customer segments. Dave Huffman, SVP of commercial operations for Claritas, advises marketers to ask, how much better performing is the campaign, the direct-mail piece, or the digital targeting with segmentation versus being purely random? Huffman also recommends using response models and attribution modeling where they can help marketers attribute the ROI and responses. “You know those segments are working because you see the purchasing,” he says.

“You see the action you’re looking for from the offer based on transactional data you’re getting.”


Like much of marketing, segmentation should have an element of test and learn. “It’s about trial,” says Momentum Worldwide’s Black. “Do it in a way where you can get a quick read before investing a lot of time. Marketing today is a lot about being fearless, fast, and agile, and then running with an initiative when it works.”


Many marketers we spoke with recommend having a budget for prospecting new and undiscovered audiences, so you’re prepared if you starting to see growth slow down or decrease from your current segments. “That requires a mind-set of, ‘OK, I’m not going to be as efficient with every single dollar, but I’m going to learn so much more,’” says H Code’s Morse. “New audiences [continually] develop that are doing and buying different things.”


For some marketers, segmentation is a continual process. “We segment by campaign because there are a lot of elements that are dynamic and we’re learning constantly about customers,” says eRelevance’s Cohen. “The next campaign can benefit from the learning from each previous campaign.”


Will marketers ever uncover the absolutely ideal segmentation? It’s unlikely, says Thunder Experience Cloud’s Wong, because new data is always being created, customers move in and out of segments, and business strategies change. “As new sources of data become available that you can overlay on your customer data, it can reveal insights and new discovery,” he says. “And then, obviously, once you’ve cycled through the sources of data, you can pause until there’s new data or new ways to look at the data.”


How consumers make purchase decisions is complex, so the challenge to marketers is to keep segmentation from becoming as complicated as their customers are. “Shouldn’t your target audience be anybody who is probably interested in buying your product? asks Cognitiv’s Fain. His advice: “Build segments that would be more easily described as look-alikes. Create a heuristic, a simple descriptor of these people. Then [find] more of those people — because you should always be looking for new customers.”


RedPoint Global’s Tripp agrees. “We should always be evaluating, always segmenting — what we call zero-segment marketing,” he says. Ideally, he adds, marketers would automate the segmentation process and apply machine learning to it, so it’s ongoing and manageable at scale. “What we’re talking about is, how can we create an environment where we could always evaluate all customers at all times to determine: What is their current state? What is their engagement? What is the right context, the right offer, the right message?” Tripp asserts that this would enable marketers to score and evaluate all customers in real time to determine whether they should receive a specific offer, content, or sequence of events.


Is there an ideal size for a segment?

Taking an approach like the ones Cohen, Fain, and Tripp recommend suggests that the ideal size of a segment will ebb and flow with each campaign or each changing customer attribute. That’s not far off from what many other marketers believe.


“It’s Goldilocks for sure. You don’t want to be too small and you don’t want to be too big,” says Thunder Experience Cloud’s Wong. “You can keep subdividing groups into smaller and smaller segments, and there are diminishing returns at some point.” Wong recommends using segments large enough relative to your conversion rate, or other success metrics, so you can see a difference in performance between the segments. Or, if you’re trying to understand what works best within a segment, you need to have a large enough segment size that the success rate, when you multiply that against the segment size, shows a meaningful difference, he adds.


“It’s striking that balance between scale and hyper-marketing,” says Sabio Mobile’s Camacho. “Sometimes [a company] wants to target really specific groups. They may have the technical ability to do it, but if the goal is to drive sales and they’re only reaching a hyper-targeted group, it’s unlikely that it will drive the kind of sales growth they require.”


Determining segment size goes back to cost, TMB’s Casey points out, adding that cost can be two different things. One is the real cost — how much it costs to promote somebody; for example, sending email versus direct mail. The other is complexity — how much time do your marketers have to put together a campaign, for example. “If it gets really complicated, it’s not worth it with a small segment,” he says. A campaign may need, say, thousands of people to make it worth the time that the marketers are spending to create it, even if the cost is low to create the smaller segment, he adds.


Sometimes the size of a segment simply depends on what message you’re communicating, Momentum Worldwide’s Black points out. “If you do a birthday campaign, for instance, the segment size is irrelevant — it’s customers with a birthday that month — but the campaign is very effective,” he says.



Can you have too many segments?

The general consensus is that you can indeed have too many customer segments. There are, however, multiple reasons why less is more.


The biggest: cost. If you have a large number of segments, you’ll likely use at least some unique creative elements and perhaps even KPIs for each one — both of which may require additional resources. Sabio Mobile’s Camacho recommends using cost/benefit analysis to determine the right number of segments.


Thunder Experience Cloud’s Wong echoes that sentiment. “The number of segments that’s ideal for any given client depends on their ability to take action against it, whether that’s changing a creative message or offer or pricing treatment for each of those groups,” he says.


“You might want to have a formula for every person you’re prospecting to, but that just doesn’t make sense because there’s not enough upside to it,” adds TMB’s Casey. “You can get too carried away with segments and build such complexity that you can never execute it.”


Too many segments can create an unmanageable environment, especially if there are multiple versions of various segments and duplicates that were “saved as,” notes RedPoint Global’s Tripp. “If you don’t have a system in place to reconcile that, you’ll have duplications in campaigns across your different touchpoints, which introduces a fatigue problem.”


Another drawback to an overabundance of segments? Overlooked customers.

Too many segments not only become hard to keep track of, they also create “orphans”—outliers who don’t fit in any of the small or specific segments, notes eRelevance’s Cohen.


Additionally, the more segments you use, the less scale you’ll have. “You’ll miss out on a percentage of the population that you might not be targeting that is your audience,” says H Code’s Morse. “And you can optimize yourself out very quickly when you start overlapping segments. You lose a large percentage of the available users to be able to target, which means you start losing ROI quickly.”


How would you define a segment of one?

One-to-one marketing has been woven into marketing parlance since the early 1990s. Even so, the debate continues on whether you can “segment” individual customers at scale. Indeed, while some marketers define “a segment as one” as an actual individual, others think of it as a cohort based on the attributes of a typical high-value customer.


“It’s understanding the person you want to reach on a one-to-one basis, his interests and likes, and other attributes about him that are important,” Morse says. “And then saying, ‘[We have] an audience size of 50,000 who are similar to that person,’ so you can have more scale and efficiency with what you’re trying to do.”


Karthick Iyer, COO of Claritas, takes a similar stance. Iyer views a segment of one as “mass customization intersecting with mass personalization intersecting with one-to-one.” There isn’t a 100 percent unique marketing message for each individual customer, he says; instead, the personalization is created from what customers are likely to respond to, based on what a marketer has learned about them. “The personalization is how you communicate, where you communicate, when you communicate, and how you make the person feel,” Iyer says. “It’s, ‘Hey! I’m talking to you!’ Rather than, ‘I’m sending a general message.’”


One of the challenges of true one-to-one marketing is today’s omnichannel environment. It can be difficult to know who a specific customer is in various channels and harder still to maintain a seamless “conversation” with that customer across those channels.


“I think of a segment of one as a single person that you’re able to track continuously across different touchpoints,” says Thunder Experience Cloud’s Wong. “That’s important because if you’re going to market to a person individually, you need a persistent ID that allows you to target and test over time. Otherwise, you won’t be able to learn anything because you can’t “remember” that this is the same person or show different treatments and know which led to what results.”


Echoing Morse and Iyer, Wong adds that having a persistent ID would allow marketers to model for look-a-likes to create meaningful larger segments. “A segment of one is possible, but you have to think about it as it’s not just one person that you’re marketing to and testing to,” Wong says. “You’re actually trying to use information about others like that one person to better personalize and better optimize the experience for that one customer, as well as [the look-a-likes], over time and across devices.”


Personalization at the individual level, while technically possible, is often impractical due to the related costs, Sabio Mobile’s Camacho points out. Even marketers with access to technologies that can change creative dynamically may not have resources required to create all the necessary content.


“There are already marketers out there doing dynamic creative optimization, so the tool set exists right now for companies to deliver that right message, right person, right place, right time,” says Cognitiv’s Fain. “But the practical reality usually doesn’t end up as one-to-one marketing. It ends up as, say, 15 different groups, [each] with some demographic descriptor that could help with messaging and the like.”


The way TMB approaches “segments of one” is more about the tone of its communications than delivering wholly unique communications to each individual customer. “There are two sides to one-to-one,” Casey says. “One is how you communicate with your customer; the creative side. It’s absolutely possible to talk as if you’re talking to an individual customer in how you do your creative. And, in fact, it’s important that your message is credible to the consumer; that they feel like that they’ve been selected to receive a particular item or offer because we think they’d like it.


“The actual act of talking to somebody individually is more difficult, obviously,” he adds. “We’re getting to a place where we can do it on the web. That’s where the costs are low enough that you can create an algorithm or real-time data links that will get the right offer to the right person. It’s not perfect yet, but we’re getting much better at it.”


Even with all the progress TMB has made toward individual personalization, testing and learning has shown that it’s not always necessary. “We did a lot of experimentation on [using a one-to-one approach for] next-best offers,” Casey says. “We found that one or two products made sense as the next offer almost no matter what. So while we might miss some outliers, we’ve coalesced around more mass personalization than truly marketing to each person individually.”


Ultimately, segmentation is about getting the most from what you know about customers to speak to them with relevance. “From a marketer’s perspective, rather than try to understand each individual consumer, which is a very challenging thing to do, you have to be able to put them in buckets — that is, segments — so you can say, “This bucket is the ideal consumer for us,” says Claritas’s Iyer. “Even mass customization has its limits, so it’s unlikely that there will be one person in a bucket, if you will.”


Iyer and others echo the sentiment that, even as technology helps marketers embrace a more one-to-one approach, segmentation will remain a marketing staple for the foreseeable future. “In a world where the nature of how information flows is ever-changing, where privacy laws will continue to increase, and where using what we know about customers allows us to provide a better customer experience, segmentation remains a critical element, of marketing” Iyer says. “Segmentation is here to stay.”


About the Author

Ginger Conlon, chief editor and marketing alchemist at MKTGinsight, catalyzes change in marketing organizations. She is a frequent speaker on marketing and customer experience, and serves in advisory or leadership roles for several industry organizations. Ginger was honored with a Silver Apple lifetime achievement award for her contributions to the marketing industry.

Find her at @customeralchemy and on LinkedIn.