Persona Development for Information-rich Domains Rashmi Sinha Uzanto Analytics & Design 6486 Benvenue Avenue Oakland, CA 94618 rashmi@rashmisinha.com ABSTRACT
Designing information architecture for complex websites requires understanding user information needs and mental models in that domain. Personas, or user archetypes, created for such domains should also reflect types of information needs, and usage of information set. We have created a statistical technique to identify important underlying groupings of information needs. In a preliminary study, we show how designers can use this information in conjunction with data from interviews and observations to generate and refine personas. Keywords
Personas, information architecture, design, user profiles INTRODUCTION
Personas, or user archetypes, an increasingly popular design technique, are used to personify important user characteristics for product design and marketing. User profiles have a long history in marketing (Moore 1991), while Cooper (1999) heralded the use of personas in product design. Personas help define the product by replacing the abstract, elastic user with the vibrant presence of a specific user who becomes a part of the design process. Cooper suggests that personas should be loosely based on interviews and observation, with little emphasis on identifying representative users. In contrast, Grudin & Pruitt (2002) think that finding representative users is key, and the persona creation process should involve both quantitative and qualitative information, including market segmentation studies, field studies, focus groups etc. Related Research
Market Segmentation: The goal of market segmentation is to produce the maximum appeal to target users, while the goal of product design is to develop a product that best meets the needs and goals of users. Market segmentation often relies on demographic information to classify users. In contrast, our goal is to find common types of information needs, guiding our choice of method and statistics. Copyright is held by the author/owner(s). CHI 2003, April 5–10, 2003, Ft. Lauderdale, Florida, USA. ACM 1-58113-630-7/03/0004.
Study of Individual Differences: Cognitive psychology is characterized by usage of experimental techniques, and the study of average behavior, while the study of individual differences is characterized by a correlational approach, and multivariate statistics. Dillon & Watson (1996) point out that HCI has been influenced by experimental psychology but less by individual differences psychology that offers many lessons for understanding users, including the factoranalytic techniques used in the present study. Personas in Information-rich domains
Understanding user information needs and mental models is important for design in information-rich domains. Information architects use card-sorting and other methods to understand user mental models for better design. We propose that personas for such domains should be similarly informed by user information needs. The cast of personas chosen should reflect the types of information needs. Goals of Study
Current persona development processes emphasize precision (building detailed descriptions), but not accuracy (identifying representative users). The designer makes a subjective judgment regarding what user archetypes to focus on, a judgment that might be difficult for inexperienced designers. Even for experienced designers, personas based on the same user research might vary widely, because there is no tight coupling between user research and persona development. Finally, persona development relies mostly on interviews and observation, techniques that are useful for gaining deeper insight into a few users, but are not economical for gaining a broader understanding of target user groups. Our goal is to create a tighter coupling between user research and persona development by using quantitative methods to identify types of information needs METHOD Participants and Procedure
The persona development exercise was undertaken for an online Bay Area restaurant finder. The project goal was to allow users to find a restaurant to match their tastes and the occasion. The first step was a preliminary exploration of the information domain. We identified 32 dimensions of the restaurant experience by surveying other restaurant finders, and phone interviews with two people.
63 respondents filled out a questionnaire about the restaurant experience, including rating 32 dimensions, e.g., "Good wine selection”, “Is child friendly” and "Good for quiet conversation", on a five point Likert scale (1=not important; 5=very important). Mean ratings suggested what dimensions of the experience were more important (food quality, good service), than others (valet parking, decor), but did not provide insight into individual differences. Identifying the underlying factors
In order to explore individual differences, we used Principal Components Analysis (PCA), an exploratory data analysis technique that can reduce the dimensionality of large datasets, by identifying important underlying factors. Bartlett’s test of spherecity showed that the dimensions were correlated (so PCA was appropriate) [x2 = 804.03, p<.01]. The results (a five-component solution) accounted for 53% of the variance. Equamax rotation was used to divide variance equally between the five components. Table 1 shows the factor loadings for three components. A component can be regarded as an independent cluster of needs. Every respondent also has a score for each component allowing examination of differences between people high on different needs. Table 1: Factor Loadings for 3 factors Factor 1 (10% of variance): Romantic atmosphere, Trendy, Décor, Wine-Selection, Takes reservation, Special meal options, e.g., vegetarian, Well-stocked bar. Factor 2 (10% of variance): Buffet style, Good for groups, No rush in service, Child-friendly, Portion size, Meal options: children’s menu. Factor 3 (13% of variance): Well-stocked bar, Locations, Good for people-watching, Credit card accepted, Price, Outdoor seating, Wine Selection
Creating the Personas
Next we gave all this information (the design task, factor loadings for five components, questionnaire, means etc.) to two information architects - who had previously created and used personas in the course of design work. A brief explanation of factor loadings was provided. The information architects were otherwise left free to use the information as they saw fit. The information architects reported that their main focus was the factor loadings, which helped understand primary and secondary motivation of the persona, and were used as the core of the personas. But they did not always use all the factor loadings for a particular persona, using information that helped formed a coherent picture of a user archetype. Other information such as spending habits, frequency, age, helped flesh out demographic characteristics of a persona. Overall, the information architects felt they had enough information to start with persona development, but needed to do more user research to finish the process. They created four personas from the five factors, reporting that two factors were not sufficiently different to be distinct
personas . Table 2 shows 3 resultant personas. Next, we plan to verify and refine the personas by conducting interviews, targeting users that could be represented by each of the personas. Computing component scores (from PCA analysis) for questionnaire will allow us to identify which persona a particular respondent best represents. Table 2: Personas from Factor Loadings in Table 1 Romantic, relaxed dining experience… Adam (42), a Sales Professional in the Bay Area. He dates frequently and is always on the lookout for a great place to take a date to. A satisfying dining experience (good view, service, wine selection), is more important than price. (based on Factor 1) Buffet-style dining with kids in tow... Susan (32) has two young children, and regrets not going out more often. She likes child-friendly restaurants (menu, seating arrangements & atmosphere), and buffetstyle service, and avoids trendy restaurants that cater to a young crowd. (Based on Factor 2) Meet for drinks, eat some food too... Lin just got out of school and is on her first job. She goes out a few nights a week. Along with friends, she loves to try out new & trendy restaurants. They like places with outdoor seating. (based on Factor 3)
CONCLUSIONS AND FUTURE WORK
The information architects were able to successfully create personas from the components identified by PCA analysis . The personas can now be verified, refined and enriched by conducting targeted interviews and observation. The proposed technique enhances the accuracy of the persona creation, while working in a complementary way with other qualitative methods. Although the technique uses specialized statistical methods, the design team retains ownership of the persona creation process, increasing the chances of the resultant personas being convincing (Blomquist & Arvola, 02). This is the first effort to develop a technique that provides for a direct link from user research and personas. We are also experimenting with other types of data collection and statistical analysis. Currently, we are building a software tool to automate the analysis for persona creation. The tool will allow designers to run the analysis themselves. REFERENCES
1. Blomquist, A & Arvola, M (2002). Personas in Action: Ethnography in an Interaction Design Team. NordiCHI. 2. Cooper, A (1999). The inmates are running the asylum. Macmillan. 3. Dillon, A & Watson, C (1996). User Analysis in HCI: historical lessons from individual differences research. International Journal of Human Computer Studies, 45. 4. Grudin, J & Pruitt, J (2002). Personas, participatory design and product development. PDC 2002. 5. Moore, A. G. (1991). Crossing the chasm. Harper Collins Publishers, New York.