Predicting the future isn’t just for mystics. With big data, especially applied to customer segmentation and persona development, we now have a much stronger predictive capacity. We can collect little pieces of information that define our customers’ lives. We can start small, at their email address, and grow from there to find out what their favorite movie is or whether they are pregnant. We can use these little pieces of data to understand the bigger picture of who our customers are. We can even use this data to give our customers a seat at the table, as Jeff Bezos insists on at Amazon, to put the customer at the heart of our decision-making.
By gathering data, we can see patterns over time that take the guesswork out of predicting the future — for example, movie buffs buy streaming services and pregnant women buy bibs. This type of categorization, scaled, is customer segmentation. We can group people based on age, location, or race. We can also group them by personality traits like opinion, interests, and lifestyle. We can group them by the goals they’re most interested in pursuing, or their underlying beliefs about which way the toilet paper roll should face.
Segmentation Based on Demographics & Socioeconomics
The bread and butter of segmentation is the quantitative method, which is quick and often effective. Grouping people based on their location, their age and stage of life, gender, race, nationality and religion creates finite groups that can be marketed to with precision. Of course, the most relevant metric will depend on the product. Two 30-year-old males might be leading completely different lives — one a father with a mortgage and the other a kid at heart living with his mom. Their buying habits will be as different as night and day.
On the flip side, a company can identify its least-likely consumers based on some of these demographics. For example, a water skiing company would underperform in a location without bodies of water and would perform better by targeting coastal or lakeside locations.
Similarly, analyzing socioeconomic factors like income, education, and occupation can give insights into what your customers value. If your company sells an upscale, luxury item, it should be targeting high earners, and maybe targeting those high earners with prestigious educational backgrounds on the assumption that they care about reputation.
Segmentation Based On Behavior
The most valuable segments, though, are deeper than demographics. For some companies, grouping based on age is appropriate, like a clothing brand marketed to teens. For others, age segmentation is worthless. Grouping customers based on purchasing behavior — which will vary by product — is more effective. For example, imagine that Verizon’s most valuable segmentation technique is based on data usage habits. That would put Stephanie, 28 — whose daughter streams a lot of Caillou episodes from Stephanie’s phone — in the same segment as David, 45, who binges Breaking Bad and Making a Murderer. They’re in the same segment because they both use lots of data by streaming videos. If Verizon relied on age segmentation like the clothing brand did, they couldn’t target both Stephanie and David in a campaign that gives a discount on video streaming.
Segmentation Based on Goal
Segmentation based on company goals, like the Verizon example, is common and effective. Android used this technique, grouping people who they believed were most likely to buy an Android smartphone. They predicted that this segment was filled with technophile, masculine men.
Consider an example that was competing against that company goal-centric Android ad: the original iPhone ad, which focused on segmenting users based on their goals instead of their traits. That segmentation technique revealed the value of the iPhone to vastly different demographics, as different as a stereotypical grandma and a computer engineer, because it showed an any-user ordering food, watching a movie, and using a calculator. This campaign wasn’t directed at grandmas or computer engineers, but instead, it was directed at people who wanted to use their phone to order dinner or watch a movie in a cab. This campaign cut across typical segments.
Segmentation Based On Mental Model
Jerry Olson argues that even segmenting based on behavior or goals isn’t enough because you aren’t getting at what “fundamentally motivates their behavior.” For example, Olson interviewed owners of differing car models in six countries and found just one integral mental model, a belief that motivated purchasing decisions about car oil. “Virtually every customer expressed, metaphorically, the same deep orientations: ‘My vehicle is “alive” and I am responsible for its well-being. I “feed” my vehicle good motor oil (nutrition) to keep it healthy. In return, my vehicle will take care of me.’” This kind of underlying mental model cuts across the other groups that a segmenting effort might fall prey to, like car model, driving experience, etc. By segmenting based on a small handful of mental models (just one in the oil example) instead of age or behavior-based segments, the marketing team can develop fewer messaging strategies that appeal to a customer’s mindset, which packs a lot of predictive punch.
Getting the data
Obviously, segmentation comes in a myriad of forms. But once you decide what information you want about your customers, how do you go about getting it? A few suggestions from ConversionXL include:
- Sending out surveys: They recommend between 7-10 questions like, “When did you realize you needed a product/service like ours?”
- Phone and in-person interviews: Sometimes talking one-on-one reveals those underlying beliefs that a survey just can’t get at.
- Web and exit surveys: From a user experience perspective, I can’t get behind these as they disrupt the flow. But surprisingly, they garner high conversion rates, so they’re effective. It might be a single question like, “Did you find what you were looking for? YES/NO.”
- Google Analytics: Dig into the average revenue per user and whether similarities emerge from how much they spend, how new vs repeat customers interact with the site, etc.
But segmentation, all-powerful in predicting buyer behavior, can create problems when used to define the product, according to Cooper, a company that employs a different method of understanding customers: Personas. Cooper grappled with turning segmentation data into insights to develop user-centered products. “For example, market segmentation information might suggest that a particular e-commerce site appeals to the needs of consumers on a tight budget. But then what? How do you design the site to meet the needs of those consumers?”
The rich data that you curate for segments can be doubly effective by also developing personas. Cooper’s definition: “Personas are a set of fictional, representative user archetypes based on the behaviors, attitudes, and goals of the people we interview in our research phase. Personas have names, personalities, pictures, personal backgrounds, families, and, most importantly, goals; they are not “average” users but specific characters.” While personas are often created from the same research that forms segments, Cooper goes on to discuss that a persona is not a summary of a segment with a name attached (example of slapping a name on a demographic segment: Mary the mom is 25-35 years old with 1.5 children and is in the middle class). Instead, a valuable persona gives life to a goal-based segment — the segment that reveals motivations and potential usage patterns. For example, a segment for a rear-seat TV customer might include middle-class parents with children aged 4-16. From this segment, a marketing venture might include any number of campaigns that target busy parents as well as their children. A persona, though, allows marketers to better target potential use cases and motivations. An example persona from Cooper:
“Kathleen is 33yrs old and lives in Seattle. She’s a stay-at-home mom with two children: Katie, 7, and Andrew, 4. She drives the kids to school (usually carpooling with 2-3 other kids) in her Volvo wagon. Kathleen is thinking about buying the Sony rear-seat entertainment system she saw last weekend at Best Buy to keep the children occupied on the upcoming trip to see family in Canada.
She doesn’t want to be distracted by the noise from the videos or games so wants to make sure she can set the sound to be heard only in the back seat. Kathleen also wants to make sure her kids are watching appropriate programs; therefore she wants some channel controls close at hand, but she thinks Katie should be able to control the system most of the time so she won’t be distracted.
From this example, the designers can ascertain that Kathleen does not necessarily want her kids to be wearing headphones for an entire journey, as she likes to talk with them on their trips, and that she may want Katie to have some control of the entertainment system from the back seat.”
Personas are excellent tools to inspire and lead design decisions because they inspire empathy. A persona that doesn’t inspire empathy is one that might as well be a data-driven segment. Of course, easier said than done — it’s too common for personas to be hung up on a wall, raved about early in the design process, and forgotten as features are added and as developers are trying to find ways to make the design work in the CMS.
But to make a persona that continues to inspire empathy? That’s a feat. A few recommendations from UXers and marketers alike:
- Find a real-life representation of your persona. Encourage your stakeholders and designers to think of someone in their own lives who they could think, “Wow, my uncle Jim is just like our persona, Harold!” Then, instead of forgetting about the personas, stakeholders will think of them when they make product decisions. Harold as a persona might be easy to see as a bunch of data and a stock photo, but would my uncle Jim like this feature?
- Do an in-house activity to get to know your personas. It’ll switch the mindset to give your internal team some feeling of ownership over the personas.
- Don’t pick gimmicky persona names. Sally the sales woman from Sacramento? How easy is it to relegate her to the “fiction” category of your mind, instead of thinking of her as a real person who will interact with your decisions through your product?
- Avoid stereotyping. An example is designing a site “for women,” when women as a group have vastly different goals and motivations. A quote from Sam Ladner: “Don’t treat me, a childless woman of 38, as a ‘mom on the go,’ simply because YOUR data tell you I should have children. Instead, empathize with me.” A few more stereotypes that not only kill your internal stakeholders’ empathy but also inspire poor design decisions that might offend your customers: the doofus dad, the frat boy, and the girl-next-door supermodel.
When it comes down to it, just the desire to understand your customers is the first step toward creating a customer-centric workflow and end product.