Jenny Don’t Change Your Number

Who is being marketed to in your neighborhood or in the communities you are studying? Enter a zip code and the Prizm market segmentation system returns five socio-economic types (out of a total of 67 possible types). What’s really fun are reading about all the different categories that the marketers have dreamt up. Here’s a flash video that provides a glimpse into how the data were generated and organized.

Here’s some PRIZM segments from my life’s geographies–


Where I currently live, Newport News, VA:

  • Blue Chip Blues — a comfortable lifestyle for ethnically-diverse, young, sprawling families with well-paying blue-collar jobs
  • Domestic Duos — a middle-class mix of mainly over-65 singles and married couples living in older suburban homes
  • New Beginnings — households tend to have the modest living standards typical of transient apartment dwellers
  • Park Bench Seniors — With modest educations and incomes, these residents maintain low-key, sedentary lifestyles. Theirs is one of the top-ranked segments for TV viewing, especially daytime soaps and game shows.
  • Suburban Sprawl — they hold decent jobs, own older homes and condos, and pursue conservative versions of the American Dream


The field site for my dissertation, Cherokee, NC:

  • Back Country Folks — residents tend to be poor, over 55 years old, and living in older, modest-sized homes and manufactured housing
  • Bedrock America — With modest educations, sprawling families, and service jobs, many of these residents struggle to make ends meet. One quarter live in mobile homes. One in three haven’t finished high school.
  • Blue Highways — the standout for lower-middle-class residents who live in isolated towns and farmsteads. Here, Boomer men like to hunt and fish; the women enjoy sewing and crafts, and everyone looks forward to going out to a country music concert
  • Crossroads Villagers — a classic rural lifestyle. Residents are high school-educated, with downscale incomes and modest housing; one-quarter live in mobile homes.
  • Shotguns and Pickups — scores near the top of all lifestyles for owning hunting rifles and pickup trucks. These Americans tend to be young, working-class couples with large families


Where my parents live, in a swanky part of Austin, TX:

  • American Dreams — In these multilingual neighborhoods–one in ten speaks a language other than English–middle-aged immigrants and their children live in upper-middle-class comfort.
  • Bohemian Mix — ethnically diverse, progressive mix of young singles, couples, and families ranging from students to professionals. In their funky row houses and apartments, Bohemian Mixers are the early adopters who are quick to check out the latest movie, nightclub, laptop, and microbrew.
  • Money & Brains — high incomes, advanced degrees, and sophisticated tastes to match their credentials. Many of these city dwellers are married couples with few children who live in fashionable homes on small, manicured lots.
  • Urban Achievers — the first stop for up-and-coming immigrants from Asia, South America, and Europe. These young singles, couples, and families are typically college-educated and ethnically diverse
  • Young Digerati — Affluent, highly educated, and ethnically mixed, Young Digerati communities are typically filled with trendy apartments and condos, fitness clubs and clothing boutiques, casual restaurants and all types of bars

I read about this first in the librarian blog/ webcomic, Shelf Check, a must read for anyone who harbors a secret crush on librarying. Posey writes that she saw it first in Lifehacker.

Matt Thompson

Matt Thompson is Project Cataloger at The Mariners’ Museum in Newport News, Virginia, and currently working on a CLIR ‘hidden collections’ grant to describe the museum’s collection of early 20th Century photography. He has a doctorate in anthropology from the University of North Carolina and a Masters in information science from the University of Tennessee.

18 thoughts on “Jenny Don’t Change Your Number

  1. I checked my current zipcode and what’s interesting is what Prizm’s market segmentation misses (and implicitly what a lot of market segmentation misses). My current university is virtually in the middle of my current zip code and yet strangely this fact is missing from Prizm’s marketing. There’s no mention of any of the large mass of college students living in the area who definitely have some characteristics which do not fit with the market segments categories created by Prizm. Yes, some college students might fit in “Bohemian Mix” category but there are a significant number that probably don’t.

    Secondarily, my current zipcode contains both a small former streetcar suburb and a portion of the large Midwestern city it boarders. In many ways these areas are demographically distinct and have their own distinctive social identities. In some cases, this leads to situations in which people in the former streetcar suburb won’t travel to city (even though their border is divided by a small residential street) and vice versa.

    Prizm’s categories in many ways speak to the danger of trying to treat research categories that classify people as “objective” tools. In many ways it seems that prism created an “objective” instrument to classify people. In other words the categories were created before looking at how people actually classified themselves. First, Prizm uses zipcodes which arbitrarily divide up land in ways that don’t necessarily map how people socially interact “on the ground.” Second, Prizm’s market segmentation categories seem to mash together all sorts of different social identities into larger marketing categories. For example some of the categories lump different non-white racial and ethnic groups together in ways that people who might be members of those groups might not agree with (and secondarily, seem to implicitly assume that white and multicultural [i.e. non-white] are two categories that actually provide some social data in of themselves).

    It would be interesting to see what sort of categories Prizm would come up with if they examined how people categorize themselves instead of imposing etic (outsider) categories. I suspect that the market segmentation that they would come up with would be very different with this sort of approach.

  2. Yeah, this definitely is not how people classify themselves but how marketers classify people. And you are right in that the Prizm website is rather opaque when it comes to describing the origins of these categories. Overall it seems more demographic than sociological, much less anthropological.

    However fictional these categories are, one assumes that they must also be functional. That is to say these categories wouldn’t exist if they couldn’t be packaged and sold to businesses who are trying to reach consumers, and businesses wouldn’t buy the package if they weren’t somehow effective. So rather than asking: how does Prizm succeed/ fail in matching the real social world?; I would ask: how does this “knowledge” get implemented and used by businesses to achieve certain goals?

    Getting back to the question of ethnography and the anthropological perspective we might ask the following. Given that the accuracy and knowledge contained within Prizm is partial and that the accuracy and knowledge generated by ethnography is also partial, why would businesses seek ethnographers as experts if Prizm is functional and accomplishes its task in helping you sell widgets, or whatever. In which case, what does it matter that Prizm does not match up with how people classify themselves? Why is knowing how people classify themselves valuable knowledge?

  3. “Yeah, this definitely is not how people classify themselves but how marketers classify people.”

    I briefly worked for a marketing company that utilized categories like this, but they only used 7 in total. It was seemed immediately obvious to me (in my opinion only), that this strategy had a two-fold purpose. One, that they were doing was mainly trying to sell this idea to other company execs. That is, they were trying to convince major corporations that they understood their target audiences, and could leverage this understanding into increased sales.

    Second, they were actively pursuing a strategy that was designed to develop these simple segmented categories in the general public. I.e., they weren’t just using empiric data, but were trying to shape people into distinct groups. This process doesn’t work for everyone, but it works for a lot of people. (I’d be happy to talk about the process of propaganda and the shaping of personhood.)
    The simple fact is that whenever you get into a conversation with someone and they tell you that they are “like this,” they are lying to you and themselves. The reality is that they are who they are depending on who they’re with and the situation. What they describe are idealized versions of themselves, in ideal situations. People are constantly looking for self-definitions, and they will utilize the categories designed by marketers, as much as anything else. Ethnic, or religious, identity is a good example of this and a great way to manipulate people.

    We can shift the conversation to political identities developed by political scientists and others with the simple categories of right and left. When one person expresses any political opinion or points out any particular fact they immediately fall into a political category in the minds of others. These categories are fictions, beyond a few ideologues on the extreme ends of the spectrum, but they are still assumed to be real by most people (at least until you question them, at which point people often admit that they are far too simplistic). I’ve run into this problem on the anth. blogosphere. Once I, or anyone else, states a particular opinion, or points out a certain fact, they are immediately categorized and assumed to hold a static and predefined set of beliefs. There really is no static view of the majority of people in the center with complex and shifting views which rarely line up with the right/left paradigm. When there is no ideational category for someone they simply cease to exist.
    So, just like marketers convince people that they are fags unless they by a particular brand of deodorant, they also use these strategies to convince people that they are either “tree huggers,” or, “tea baggers,” if they don’t vote a certain way or give money to particular causes.

  4. I don’t know if I believe you that people use labels such as these developed by marketers for the purposes of self-identification. Don’t most people chafe at being labeled in this way? Show this list to anyone and tell me if they embrace what it supposedly says about them and where they live. I guess its possible, but its more likely, I think, that many of these categories don’t actually exist.

    Just to run with the political labels (which are not generated by political scientists, but political strategists — essentially glorified marketers) every election cycle, it seems, there’s a new segment of society that must be pitched to. Tree huger and tea partier aren’t demographic groups. There is no tree huger part of town, tea partiers don’t break down by zip code. Soccer moms, Nascar dads. These fictions are geographic and speak more to what Prizm is about.

    Again, to turn this into an anthropological question. I don’t think the issue is so much what are the shortcomings of this methodology. After all our own methodology is replete with shortcomings. But rather, to what ends is this “knowledge” put to use? How do people enact (perform) these ideas? Is it self-identification, or something else? If so what?

    What is the relationship between knowledge (in Prizm) and behavior (how its used)? And what are the ideologies (capitalism) that support, or grant authority to this way of doing?

  5. “Don’t most people chafe at being labeled in this way?”

    This goes along the line of someone saying they are different, just like everyone else. How many people do you think would say they are manipulated on a daily basis, or that a person shouldn’t recycle, or whatever? How many people do you think would say that it’s ok to cheat on a spouse, especially those that have cheated on a spouse? There is a rather wide gulf between what people say, or their self-concept, and what they actually do in any given situation. This is what happens when we fail to make a distinction between the emic and etic. When we as anthropologists fail to distinguish between what people tell us and observable/measurable etic reality. I discussed this on the Islamophobia thread, but Kerim in fascistically controlling and delimiting the debate deleted it. I was able to retrieve part of what I wrote:

    “As anthropologists we must always compare what people tell us to recorded data. Then we can better measure and understand both consistencies and inconsistencies between what people tell us and what they simply believe but is probably not true, and then explain why it is they believe what they do. It is both unanthropological and unscientific to assume we know factual data in relation to personal conceptions or beliefs.” Otherwise, we fall into the fallacy that culture is arbitrary and self-caused.

    A person can say that they are “like this,” all day long, but if it doesn’t translate to observable increased sales goals, then it becomes meaningless. One of the ways that this is done is to reference pre-defined ‘outsider’ groups. For example, Marlboro lights were originally developed as a woman’s cigarette. They weren’t selling until the Marlboro Man was invented, and then sales jumped dramatically. A persona was invented, and it was automatically set against everything that it wasn’t. The persona tied together selective truths with other things that people wanted to hear or believe, which is how propaganda works. Recently, I saw a truck commercial state that a truck was stable with heavy loads and didn’t sway back and forth, “like hippies sitting around a camp fire.”

    A good example of this is a brilliant hybrid car marketing campaign in Northern California years ago, which was developed by the guy that invented the Marlboro Man to repent for his sins. In order to sell the cars to conservative Christians, he developed the line, “What would Jesus drive.” With that line he created cognitive dissonance in the minds of folks, and cleaved away the narratives developed by right-wing propagandists.

    “And what are the ideologies (capitalism) that support, or grant authority to this way of doing? ”

    I think its a conceptual stretch, bound up in a performed “leftist” narrative that would say this is bound up with structures like capitalism. In the world of marketers there isn’t an idea like right or wrong, there is what works and what doesn’t work. One doesn’t sell something by telling someone something they don’t want to hear. I think it’s the totalitarian ideologies historically like fascism or state communism, which have sought to impose ideational constructs upon material reality.
    That being said, I think that it would be a mistake to see this as outside the universal human condition. If culture is patterned behavior by groups of people, then one couldn’t separate this from culture.

  6. This is what happens when we fail to make a distinction between the emic and etic. When we as anthropologists fail to distinguish between what people tell us and observable/measurable etic reality.

    By opposing “what people tell us” with “observable/measurable etic reality” you seem to be suggesting that what people tell us amounts to confabulation and/or lies. That’s a very narrow rendering, though. David Schneider used to say that if an ethnographer from Mars sat all day on any corner with a stop sign in Hyde Park that at the end of the day he/she/it would write up that stop signs mean ‘slow down.’ But we native speakers all know they don’t. The difference is that of model and practice, not false and true.

  7. Rick, in creating a dichotomy between an “emic” and a “observable/measurable etic reality,” aren’t you employing the same rhetorical technique those political scientists employ in their imposition of a “right/left” dichotomy? Should we be as wary of your attempts at this as you seem to (rightly, in my mind ^_~) be of marketers and political scientists?

    And on a similar note, in your dichotomy, how do you fit in attempts to bridge etic and emic*?

    *I don’t know enough of anthropological theory to know of any specific attempts to do so. I’m thinking more of the subtle realist perspective proposed by Nicholas David for ethnoarchaeology, which attempts to bridge interests in ideology and meaning with materialism.

  8. Whoops! Let me amend that previous post. N. David’s subtle realism was not specifically designed to bridge ideational and materialist interests, but provide an epistemological model for ethnoarchaeological investigation that could serve either category of interests. I think, though, it’s a step in the direction of bridging the two, and it was my perception of the similarity of that problem to the emic/etic issue Rick references that led to my question.

  9. “a dichotomy between an “emic” and a “observable/measurable etic reality,”
    Etic is how we see others but also how others see us.
    There is no “reality based” community.
    You can’t separate language from politics.
    Reality is a negotiated agreement.
    [Democracy is a form of government predicated on the acknowledgment of that negotiation.]

    And yes, the Prizm market segmentation is a vulgarization of the vulgarization of the categories of political “science.”

  10. “By opposing “what people tell us” with “observable/measurable etic reality” you seem to be suggesting that what people tell us amounts to confabulation and/or lies.”

    In a way that’s somewhat what I’m saying. Perhaps, more accurate would be to say that it is to be smart enough to know that people lie, and they lie a lot, and they lie to themselves as much as they lie to anyone else. I think I said this before in another thread, but I think too many ethnographers have a kind of ethnocentric blindness, or domain dependency, which assumes that a researcher can do things in other cultures that they would never think about doing in their own. For example, would you take the average American’s personal opinion and explanation for why the average American does something? If you ask a dozen people, you’re likely to get a dozen answers. When you get patterned responses, you are likely recording the emic beliefs and opinions within coherence social networks and structures, but even then that usually isn’t the full story.
    In my work in inner city neighborhoods for example, I’ve found that different groups have access to different pieces of the larger information pie. I have yet to run into any of them that have triangulated the different pieces together, along with spending weeks in the city archives, pouring over historical documents, working with longitudinal data and statistics; all the while comparing everyone’s versions of the truth to each other and comparing where they live, who they are, how long they’ve been there, etc… All of that is my job as an anthropologist. If I just went out and reported what people told me then I’d be a journalist and nothing more.
    If a dozen people tell me that a neighborhood became more violent beginning in the late 1970’s, because of drugs, then it’s my job to figure out why. Drugs existed and were illegal, and were sold in such neighborhoods decades earlier, so why then? And, why did things begin to get better in the mid-1990’s?

    As so, with marketing, you can ask someone why they prefer this food over this other food, but we know biologically everyone tastes food differently and patterned tastes -like why American men prefer steak and potato dinners- are rarely at the conscious level of understanding. It’s been a very long time since people had access to the information they’d need to understand why they prefer this to that, which have their origins bound in hundreds of historical, economic and geographic connections.

  11. “Reality is a negotiated agreement.”

    Somewhat, and only on an idealizational level. Often either something happened or it didn’t. Facts matter, and I think to many of us forget that. Humans need a certain number of calories a day to survive. A new age group can say that they can survive eating only certain foods, and they will survive or they will get sick and perhaps die. The purpose of a theory is to explain phenomena. A theory that explains more with less is a better theory than one that explains less with more.

    A person can tell me all day long that they are like this or like that, but if I can use the social segmentation model and accurately predict their purchasing patterns with one of the segments within an error of 5 times out of a hundred, then the model is sound. More importantly, if I can predict behavior accurately, and the model is in fact sound, then I can predict how a set of stimuli will create a predetermine behavior that I want to influence (like make someone stand in line overnight for a new I-phone). What matters here is prediction, not meaning.

  12. Coming late to this conversation, I note a few points where I have something to add about PRIZM. First, it is no accident that the geographic definitions are in terms of postal codes (zipcodes in the UsA). PRIZM is only one example of what is called geodemographic analysis, and, as the abstract to Graham, I. , 2009-10-28 “Sorting People Out: Geodemographic Analysis and the Performance of Cluster Analysis” puts it,

    Geodemographics is the classification of people on the basis of their location or, more specifically, their home address. The most widely described application of geodemographic analysis is the use of neighbourhood classifications to identify market segments to be targeted in promotional campaigns, but the techniques are also widely used in the identification of social needs. This paper reviews the history of geodemographic segmentation and relates it to Bowker and Starr’s sociology of classification to uncover how the techniques have been shaped by their political and organisational context. The techniques grew out of early pre-war applications of multivariate statistical analysis to social analysis, with early geodemographic applications both influencing and being influenced by the development of cluster analysis methods in statistics. The paper draws on detailed interviews covering the history of geodemographic segmentation with two of the key figures in the development of the techniques, Jonathan Robbin in the United States and Richard Webber in the United Kingdom. On both sides of the Atlantic commercial geodemographics grew out of small-scale progressive initiatives to monitor social exclusion and target social policy which were developed by individuals who combined knowledge of social geography, the programming of early computers and multivariate statistical methods.

    The explosion of commercial interest in geodemographic analysis in the sixties coincided with the growth of marketing as a discipline and the availability of mainframe computers for commercial users. However the large-scale tessellation of countries into patchworks of market segments also required access to the standardised zip code classification and its mapping to data describing each zip-code. In using cluster analysis to identify classes of zip codes developers had to choose which clustering technique to use and how many classes to draw out. The classes derived were influenced by the data that was available and by expectations about how the classification would be used.

    Those expectations included what came to be called direct marketing, in which materials are sent by post to targeted consumer groups. It should also be noted that marketers using these tools were under no illusion that the classifications in question provided complete and accurate descriptions of the people living in the postal code regions in question. No professional marketer expects to sell a product or service to every member of target segment. The classifications were and are seen as useful if campaigns based on theme improve sales by even a few percentage points. In political uses of the same techniques, the goal is to win a majority of the votes of those who turn out on election day. Those who do not turn out or are unlikely to contribute significantly to a winning majority can be ignored.

    Category names are neither emic nor etic descriptions. Their primary purpose is to serve as memorable tags and facilitate communication with the clients who pay for and the creators who produce the materials in question. They are both a shorthand and a way to stimulate the imaginations of everyone involved.

  13. Rick,
    As a wise man once said: “Fact, value! Value, fact!” Which is why it’s important that you be allowed to post your bad arguments and equally important that others criticize them rather than walking away.

  14. “Category names are neither emic nor etic descriptions”

    I think technically they would be defined as etic. A memorable tag to facilitate a operationalized, meaning consistent, description of a social phenomenon would be called etic. Saying a “voiced glottal stop,” is meaningless to anyone but a linguist or anthropologist, and such a description is the basis for the term. I think a very valid criticism is to ask whether the etic description is a valid description, or whether it is a way for slick marketers to make their clients think they know what’s going on in a market. I mean if there was nothing in the physical universe that could accurately be referenced as a voice glottal stop, then the etic term would be meaningless and invalid.

    John, you would know much more than I would as to how often these descriptors are based on empiric data, and how often they are made up in the minds of what someone in an office assumes, or a little of both. My experience with population segmenting is with the military, and there the descriptions are probably closer to the latter two.
    I’ve used zipcode level data, but it’s arbitrarily based on the mail system, so it’s hard to make useful to understand geographic differences. If some data is only available at the city or zipcode level I will overlay it with Census block level data to get a clearer picture of what’s going on.

    “Which is why it’s important that you be allowed to post your bad arguments and equally important that others criticize them rather than walking away.”

    I completely respect that, no argument here. What that means though is that you have to offer a compelling alternative, along with compelling data, which can somehow knock the scientific method off the ladder as the premier and superior way to systematically collect data, compare it to other data, and produce usable theory. Good luck with that.

  15. Rick,

    If “etic” means nothing more than a third-party label, then the PRIZM labels might be called “etic.” Personally, I prefer a stricter sense of both “etic” and “emic,” grounded in their original use in linguistics, where “etic” refers to features that appear across languages and “emic” to a subset of features significant in one particular language. The distended use of “emic” and “etic” to refer to the native and observer’s point of view leads, in my view, to what Gerald Berreman wonderfully labeled “anemic and emetic” analysis.

    My point, which still stands, is that the PRIZM labels are not scientific descriptions in either an emic or etic sense. They are, as I said, a colorful shorthand that makes it easy for clients and creators to envision who they are talking about, even though both know perfectly well that claiming scientific accuracy for them would be ludicrous in the extreme.

    They are, in effect, advertising: Would you, to offer a different example, consider “The ultimate driving machine” a scientific description, either emic or etic, of BMW automobiles?

  16. Interesting to see a discussion about PRIZM on this blog, of all places. As some of the prior commenters noted, it’s mainly a tool used by marketers.

    If I can illuminate the situation in a small way, companies–particularly larger chain retailers–use PRIZM, and similar data tools, to figure out where to locate their stores. Let’s say it’s Aldi, for example, and they own plots of land in Newport News VA and Cherokee NC. Based on the PRIZM results in the post above, they would then probably take an educated guess and build a Trader Joe’s in Newport News and an Aldi in Cherokee. It’s a matter of what they consider appropriate for the market. They also have much finer-tuned tools at their disposal to refine their decision, as basing demographics on an entire zip code is seen as a bit sloppy–though it’s better than a completely blind guess.

    There are a lot of interesting things happening with a data source like this. Not the least of which is that PRIZM becomes a tool to reproduce culture by helping shape the American consumer landscape. If Trader Joe’s catches on in Newport News, it has the potential to change the way that people eat. And food, as we all know, is an integral part of our social world. It even has the ability to symbolize our politics.

  17. “My point, which still stands, is that the PRIZM labels are not scientific descriptions in either an emic or etic sense.”

    I’m not arguing with your general point, as you are the expert in this field, rather I was pointing out the fact that if there was absolutely no basis for such descriptions in the observable world then the categories would simply be a part of an elaborate insider’s game. A social comparison would be the almost religious belief in the infallibility of unregulated, capital markets. In both cases there’s a belief that is disconnected from reality.
    I’m somewhat skeptical that they are made up whole cloth for all such firms. I have zero first-hand knowledge of Prizm, per say, but a lot of anthropologists are employed by firms to gather empiric data used to segment various social groups, which is the first and most important step in the marketing process. I’ve read articles that criticize such models for cross-cultural assumptions, but that doesn’t mean they don’t have some etic reality somewhere. I’m sure that there are few people that fit into any single category in totality, but that doesn’t mean that they don’t like to think of themselves as falling into it, and therefore are easily manipulated by using the themes derived from it. This tendency is universal. For example:

    I was once on a small team for a minor project for Samsung, where I followed around 17 to 25 y.o. as they used their phones for texting. This included PO, interviews, and a few focus groups. I would hate to think the empirically derived data was meaningless. Or, that if I observed a behavior that directly contradicted something that an informant told me, that I would go with what they told me over what they actually did; without consideration.

    As I said earlier, I’ve been trained in segmenting target audiences for the military, and it was drilled into us that reality mattered in such things. As with marketing, properly segmenting target groups and understanding them is the most critical step of any information or influence operation.
    I suspect however, and you can tell me if the assumption is valid, that in both cases (civilian and military) managers who are neither trained in this science, or inclined, often influence and have final say over what’s produced. That is, there’s a tendency for some asshole at the top of the decision making process that puts in his opinions when he shouldn’t be.

Comments are closed.