¿Habla Español? If you are a usability practitioner in the United States, you may want to learn. Spanish-speaking immigrants from Latin America represent a large and growing market. Though there is much literature on cross-cultural usability in general, there is very little on this particular group. This article discusses a series of studies we ran with users from this group, looking in particular at their special characteristics and challenges.
The Latino Market
Latinos (Latin Americans living in the USA) make up a very large market, representing the second largest Spanish-speaking population in the world, a larger population in the U.S. than that of African-Americans, and with $700 billion in buying power.
According to Latino Boom by Chiqui Cartagena, this market is also growing. With a current growth rate of 60 percent, Latinos will represent $2.5 trillion in buying power in 2020 and 25 percent of the U.S. population in 2050.
At the same time, however, this market is grossly underserved. For example, Cartagena reports that U.S. companies spend only 5 percent of their marketing and advertising budgets on Latinos, a third of what they should spend based on population figures. And in 2004, only 20 percent of Fortune 100 firms offered Spanish-language content on their websites.
A related problem exists in the usability field. Though there is no shortage of material on multi-cultural usability in general, there is very little on this particular audience. For example, of the forty-seven offerings at UPA’s 2005 conference—the theme of which was “bridging cultures”—none focused on Latinos.
The profile of Latinos in the U.S. is much more complex than the average American might imagine. On the one hand, Latinos do tend to conform to certain stereotypes. They tend to be less assimilated than other immigrants, have lower incomes and education levels, and are younger than other groups.
On the other hand, there are many stereotypes to which this group does not conform. Most are born in the U.S., not elsewhere. Most are English-dominant or truly bilingual. Most are U.S. citizens or documented aliens. They also typically identify themselves with their ancestral countries of origin, each of which has its own unique culture and dialect.
This contravention of stereotypes is especially the case relative to Internet and technology use. In fact, over half use the Internet, are more likely to be online than the average European, and spend more time online than non-Latino Americans.
My company is a large U.S. bank. As such, it faces significant challenges—and opportunities—in addressing its Latino customers and potential customers.
Perhaps, most significantly, is that Latinos use financial services much less than other groups, with 35 percent being “unbanked,” that is, not using any bank services at all. These behaviors result largely from negative experiences with banks in their ancestral homeland.
This phenomenon, however, represents a huge, untapped market, and U.S. financial institutions have engaged in a recent “rush to market” to capture this significant market without fully understanding it.
My own company’s foray into this field involved the launch of a Spanish-language subset of its much larger English-language website. Though this effort was informed by several focus groups, the sub-site went live before being usability-tested. The project team approached me later to get some post-launch feedback.
One of the first challenges I had to deal with was the typical marketing segmentation for Latinos:
Acculturated–10-20 percent; have been in the U.S. the longest (over ten years). Their Spanish skills have become poor, but they are culturally proud.
Assimilated – 30 – 60 percent; in the U.S. for five to ten years; the fastest growing segment. They are younger, urban, have more education, and prefer English. They are more thoroughly bicultural.
Isolated – 35 percent; speak Spanish exclusively; they are older and represent the most recent immigrants.
My team wanted to focus solely on the first two groups, arguing that the last one had neither the technological nor financial resources to be real prospects.
The team also wanted to test in different locations, explaining to me that Latinos are not a homogenous group. As a result, we decided on testing in Houston (for Mexicans), New York (for Dominicans and Puerto Ricans), and Miami (for Cubans and other nationalities). We ended up testing a total of thirteen users.
We identified some basic banking tasks (such as finding a branch) and some focused tasks (for particular issues, like switching between languages). We also used one self-directed task.
A final challenge was how to plan and execute the overall effort. Though I had been doing usability work for a number of years, I do not speak Spanish. At the same time, however, we had a number of Spanish speakers on our team, none of whom had ever done usability work. Our solution was to contract with a company that specialized in IT globalization and had done work with both Latinos and usability. This company engaged a facilitator for us, as well as labs in the cities we were interested in.
We found a number of issues with the bank’s Spanish-language website that were not unique to our Latino users, but that we had uncovered previously with others:
Search – Like other users, about half of the Latino users were search-dominant
Online Help – Like others, our users did not like online help (because of poor previous experiences) and were unlikely to click on it
Lengthy lists – Users became impatient and stopped reading
FAQs – The users were divided on the usefulness of FAQs, in contrast with the view of the team, which thought FAQs would be particularly appealing
Privacy/security – Users had genuine concerns, a common finding on any financial services website
Human contact – About half of the users wanted to complete their tasks online; the other half wanted to talk to a human being at some point
We also found some issues that were unique to this audience:
Search – Users were as likely to search in Spanish as English, which caused particular problems when users searched in Spanish on an English page that returned no results. We also found that the search engine required users to enter accent marks, frustrating those users whose Spanish was less than perfect.
Acceptance – Users felt very strongly about the limited nature of the Latino sub-site. This finding clashed with the team’s hope that the users would at least appreciate our effort. Users spoke out quite passionately even when they themselves preferred English. They did so as self-conscious ambassadors of their community, typically citing friends, co-workers, and family who did not share their own strong English skills. We connected this behavior with the traditionally communal, rather than individualistic, nature of Latinos.
Switching languages – Users were much more likely to switch to English from Spanish than vice versa. This was largely due to their comfort with English and their exposure to many more English sites on the Internet.
Educational material – Our Latino users were more interested in this material (for instance, budgeting) than other users. We saw this as representing their overall desire for self-improvement, a common characteristic of immigrant groups. Our suggestion was to keep this material, which we had considered deleting.
Chat – Users were very interested in online chat but didn’t fully understand what it would entail (confusing it, for example, with instant messaging). Our suggestion was to offer it in appropriate places, but to make sure it was explained in some detail.
Access – Users liked the presence of an en español link, even if they never felt tempted to use it. At the same time, however, our link was rather buried. Our suggestion was to move it up to a more standard location and to make it stand out visually.
In addition to what we learned about the site, we also learned a number of things about the testing process and about working with these users:
Language preference – As noted, these users were just as comfortable in English as in Spanish. It was, in fact, quite surprising to see these users try to engage the facilitator in English when greeted in Spanish. They also struggled with Spanish or switched to English while thinking out loud. To get feedback on our Spanish-language site, it was necessary at times to force users over to it.
Sociability – Latino culture is traditionally thought of as warmer, friendlier, and more relaxed. This was definitely the case for the users we observed. We needed to spend more time in social pleasantries at the beginning and end of each test, and we wondered if remote facilitation would be inappropriate for this audience.
Forthrightness – Our users seemed to have no trouble speaking their minds. They can be contrasted with, for example, Asian users, who traditionally are much less likely to open up or be critical.
Formality – Although this group may be both more sociable and forthright, they also tend to be more formal. In testing, this trait became obvious in dress, greetings, and so on.
Speaking for others – As mentioned, Latino users readily speak for others. When other users have done this, we often disregard these comments or ask users what they would do. Latino users, however, seem to be genuine sources of feedback for other possible users, as well as guides to how our effort may be perceived within the community.
Another important lesson was that, although we definitely needed help on our first effort, we now had the resources to manage and run one of these tests ourselves.
An opportunity to do our own study occurred when the project team came to me with an interesting discrepancy between the completion rates on the Spanish and English versions of our online account applications.
For this test, however, the team did not have the budget it had previously. So we elected to test in our own lab, recruit from the local area, and train one of the Spanish speakers on the team to facilitate. Realizing the unique and valuable skills of a simultaneous interpreter from the previous tests, we did decide to contract with one.
We again saw a number of the same issues that our first test revealed, many having to do with language. Users were just as likely to use English as Spanish, but they still felt strongly about having all content in Spanish. We also found some new issues:
Mixing – Users objected to the mixing of English and Spanish on the same page, dismissing it as “Spanglish.”
Dialect – Users found many issues with dialect. We discovered that the application had been translated by speakers from a single country, using only their country’s dialect.
Literalness – Users had problems with terms that seemed to be literal translations of computer jargon (for example, deseleccione for deselection).
Slang – As noted before, our Latino users tended to be more formal. This trait was reflected again in their response to the application’s interface. For example, users frowned upon the Spanish equivalent of “make a mark” (apunta una marca) for selecting a checkbox.
For each of these problems, we recommended engaging a seasoned translator. To deal with dialect, in particular, this individual should use Castilian or Mexican Spanish or, better yet, dialect-neutral language.
Some other issues, unrelated to language, included:
Error checking – Users entered data items like phone numbers and dates in a number of different formats, each typical of the respective native country. We recommended the system use much less strict error checking, allowing different variants and processing them on the backend.
Cultural issues – Users discovered several data items that were particular to the U.S.: 4-digit Zip Code extensions, P.O. boxes, etc. We recommended identifying which of these items were truly essential to the application, deleting any that were not, and including field-level online help for those that were.
Some new issues surfaced. The marking of required fields and impatience with the length of the application, for example, were issues with other, non-Latino users as well. Finally, we were also able to test our new access point to the Spanish-language site, which performed quite well for the Latino users.
We also learned some important things about the testing process and about working with our users, the most important of which is that we can definitely do a bilingual test ourselves. And with a little help from your amigos, you can, too.