Tuesday, July 21, 2009

Segmenting Multicultural Consumers: Old Dinosaurs Die Slowly

Eventually, any organization that plans a marketing program comes to grips with the realization that their product or service cannot please everyone. Consequently, the need to define a target segment – those customers whose needs are most likely to be satisfied with a product or service – becomes a critical decision for the marketing program. Indeed, it makes little sense to introduce a product and spend significant advertising dollars only to discover later that the customers that you had hoped would buy the product are not interested.

Although market segmentation is a well-established concept in the marketing world, I am often amazed at the many companies who invest considerable financial resources to launch a multicultural campaign with little more than their own personal insights or advice from ethnic employees to guide these campaigns. To such marketers, segmentation research is not a priority because, for all intents and purposes, ethnic consumers tend to think alike, speak the same language, watch similar media, desire the same products, and shop at similar places. Ironically, when the campaign struggles or just fails, consumers are readily blamed for their lack of interest.

Even marketers who understand the need to segment multicultural consumers have been misguided by the industry’s preoccupation with language segmentation or “language buckets,” as I like to call it. Language buckets are created from responses to questions about the language that a person speaks most often at home. In the case of Hispanics, the language buckets typically used are Spanish all the time, Spanish most of the time, Spanish and English equally, English most of the time, and English all of the time. Marketers often use these language buckets to select a segment of Hispanics that they believe will be more responsive to their advertising campaigns, such as Spanish-dominant Hispanics (i.e., speaks Spanish all the time or most of the time). Despite the often biased and self-serving nature of their research, Spanish-language advocates have been successful in convincing Corporate America to spend nearly 90 percent of all Hispanic-targeted media expenditures on Spanish-language media – quite a remarkable achievement given that 60 percent of U.S. Hispanics are native-born and primarily use English-language media. It is not difficult to understand why some industry stakeholders would not want to change how language segmentation is currently practiced.

Nonetheless, choosing potential customers based on their self-reported home language speaking skills is problematic for two important reasons. First, it makes little sense to select a target segment of consumers based on their language abilities without first determining whether a product or service will meet their needs – like putting the cart before the horse – a practice that overlooks other consumer segments that may also find the product appealing. As a case in point, it seems reasonable to assume that most Hispanic homeowners need home improvement supplies, regardless of their language skills. Rather than focus on Hispanic consumers in a specific language bucket, it makes more sense to identify segments of Hispanic homeowners according to their propensity to buy home improvement supplies, and then design the marketing mix that reaches the desired segments effectively – which may include a combination of English and Spanish-language media. The home improvement company could clearly lose sales if their advertising agency recommended a specific linguistic strategy without first understanding the home improvement needs of all Hispanic consumers and then selecting the most desirable target segment.

Secondly, it is disturbing although not surprising to observe how the marketing industry has embraced such a vague and simplistic concept as home language usage to segment Hispanic consumers. The credit goes to The Nielsen Company for popularizing the use of language buckets to segment Hispanics. Nielsen sponsors national telephone surveys each year of U.S. Hispanics to create the needed information for their television ratings, and shares these universe language estimates with other research firms like Arbitron to compare or adjust their language data on Hispanics.

But the language behavior of Hispanics is not as simplistic as The Nielsen Company would have us believe. Based on various studies of U.S. Hispanics, our research shows that the language that one speaks at home varies considerably within any one Hispanic household – it depends on the subject matter under discussion, the age of the individuals engaged, their country of origin, the task at hand, and general proficiency with the language. In a typical Hispanic household, Spanish or English may be used when talking to specific family members but may take a different path when viewing television, completing homework assignments, listening to radio, playing games, talking to friends, and other activities. Because the language that one speaks at home is influenced by many factors, it lacks the stability needed to reliably segment Hispanic consumers and highly questionable when used to adjust television and radio ratings.

Language usage among multicultural consumers can lead to some unexpected surprises. In recruiting consumers for focus groups, we have learned that consumers who speak proficiently in one language do not necessarily know how to read or write in that language. In telephone interviews, we find that respondents often over-state their proficiency in one language as they try to impress the interviewer with their language skills – a social desirability response set. A more valid and practical measure of language ability can be achieved by being more specific about the language skill needed for a particular task or situation. For example:

  • How often do you speak English when talking to friends?
  • How often do you read Spanish when reading a newspaper?
  • How well did you understand the English-language advice provided by your doctor?

Going forward, I believe that multicultural marketers should re-evaluate their campaigns to ensure that they have taken the appropriate steps to identify their target segment(s), and that their marketing strategies are aligned with the needs of the segments that appear the most attractive. Although language segmentation has been a popular way to identify target segments, the quality of the language data used is suspect and distracts attention away from more useful segmentation bases such as product usage, benefits, and lifestyles. Level of acculturation, which is typically defined by language usage and country of origin, has also been used in the past to understand and segment multicultural consumers; however, it has the same potential as language segmentation to mislead marketers by focusing on a specific linguistic strategy rather than one that is based on product consumption.

Thursday, July 16, 2009

Multicultural Research in Need of a Facelift

Over the past 30 years, I’ve conducted my share of multicultural research studies and learned a few things about sound research practices. The elements of sound multicultural research are generally not learned at academic institutions since much of their curriculum and textbooks devote little attention to this topic. Instead, research professionals are more likely to learn through the “school of hard knocks” and may even grasp these elements over time. In my opinion, many research professionals still don’t get it and appear increasingly indifferent to the consequences of their misguided practices. Following is a sampling of some questionable practices that have become rather commonplace in studies of multicultural populations:
  • Incorrect usage of race/ethnic labels to screen respondents
  • The use of monolingual surveys with known bilingual audiences
  • Sample sizes that are too small to detect statistically significant differences
  • Use of online surveys that exclude large proportions of consumers who are not online
  • Consumer segments defined using unreliable language data
  • Over-sampling of foreign-born respondents which leads to biased indicators
  • Interviewers translating questions “on-the-fly”
  • Adjusting survey data with unreliable self-reported language data
  • Use of predictive dialers that lower respondent cooperation rates

One might conclude that these practices are more characteristic of small research organizations with limited resources; large organizations, however, are not immune. J.D. Power & Associates, for example, conducts their U.S. customer satisfaction research in just one language – English – which systematically excludes feedback from many customers who prefer a survey in their native language. Nielsen and Arbitron continue to use self-reported language information to adjust their radio and television ratings, despite evidence that such information is unreliable. Arbitron, in particular, is currently under heavy criticism in regards to their PPM methodology and sampling strategies that allegedly under-estimate Hispanic and African-American radio audiences.

The Census Bureau tells us that by the year 2010, African-Americans, Hispanics and Asians will collectively comprise one-third of the nation’s population – an astonishing 102 million persons. You would think that this seismic demographic transformation would alone motivate industry research organizations to abandon their outdated practices in favor of methods that more accurately capture the experiences of multicultural populations. Unfortunately, that has not been the case.

To get the research industry moving in this direction, we clearly need to expand the dialogue on these issues since academia is moving at a snail’s pace in this arena, too many important deliberations about methodology are taking place behind closed doors, and currently available books on multicultural marketing provide minimal guidance on measurement issues. Thus, I have dedicated this blog as a forum to (a) discuss methodological issues in regards to research with multicultural populations, (b) review selected studies, articles, books, and white papers that address multicultural issues, and (c) create a community of researchers and non-researchers who share similar concerns about the need to improve the quality and transparency of multicultural research. While I do not pretend to know all of the answers to these issues, I will commit to an objective and passionate discussion with members of my blog community.