Barriers and bridges in the adoption of today’s mobile phone contextual services

Editor's Notes

1. Good afternoon. This is joint work with my colleagues Rodrigo de Oliveira (who is here in the audience), Anna Hiltunen, and Nuria Oliver at Telefonica Research, in Barcelona.\n
2. Let me jump right in: Applications on mobile phones are becoming extremely popular. They allow the user to use a mobile in an unprecedented number of ways. A Mobile is a communication device but it is also becoming rapidly an entertainment center, or a navigation device for instance. Applications add functionalities to the device and extend the number of situations in which it is used. Yet, a large portion of the market is dominated by feature phones. These are devices that offer a limited connectivity to the internet and they do not allow to install 3rd party applications.\n
3. One of the reasons of the success of smartphones is the abundance of contextual applications. These are applications that sense and model the user’s context to operate. \n\nSo, in this diagram we list the five most common dimensions of context and examples of contextual applications that model a particular dimension. [talk about them]. Usually location and time are intertwined dimensions.\n
4. We were interested in understanding the following: \n(1) identifying the human needs that support the adoption of contextual services (we call these contextual needs) and \n(2) understanding how these needs relate to more general human needs. \nMoreover, we analyze potential barriers for adoption of contextual services (3)\n
5. To reflect on these issues we have carried out two user studies: the first is a qualitative ethnographic observational study of the contextual needs that smartphone users experience in their lives. It involved 8 participants. \n\nThe second study is a quantitative large-scale questionnaire where we aimed at generalizing the findings of the qualitative study to a larger population. Almost 400 people answered this questionnaire.\n\nWe deployed both studies in Spain.\n
6. \n
7. The study comprised three parts: the first was an initial interview where we gathered demographic information and explained the different aspects of the study. the participants were given a diary kit that I will show in the next slide. The second activity was the in-situ interview and observation. \n\nFinally, we conducted an interview in the participant's home to collect the diary and clarified some of the entries. \n\nAfter each interview we had a debriefing session with the team to highlight important parts of the interview.\n
8. We complemented the contextual interview with a diary study that was often used in related studies to capture impromptu needs of contextual services and applications. Diary studies have the advantage that they help participants reflect and document their experiences when the researcher is not there. They also provide a more longitudinal depth to the study. In our case they provided accounts for needs that spanned over three weeks.\n\n> in this picture you can see the kit that we gave to participants. They could use both a paper diary and a video diary because we felt that these two modalities could complement each other (one is faster, the other is more discreet). The other elements of the kit are ...\n
9. 10. So, we used an external recruitment agency to enroll 10 participants. Two dropped out in the middle of the study for personal reasons. We gave the agency a pretty detailed screener. We were looking for people that were both power users of smartphones and beginners. We looked also at whether they had previous experience with any AR application and location-based services. Again, we tried to have a good mix of both. \n\nWe balanced the sample in terms of male and female because we felt that gender could play a role especially for location-based services. \n
10. 12. Due to time constraints we had to create two teams for the field work: we asked help of two local facilitators that conducted the interviews and that were native in Spanish. We prepared a moderator script to make sure we were not forgetting any important point and to make the sessions run more smoothly.\n\n> here is a picture taken in the field. This is the subject that was going shopping in this mall and behind the facilitator.\n
11. The participants produced 116 diary entries (median 15). They also produced about 6 hours of video diary entries. We also collected several hours of interview recordings. \nAll this multimedia material was analyzed and categorized. Interview notes were transcribed and annotated with relevant pictures and extra information that was extrapolated from the debriefing sessions. \n\nThe different contextual needs participants thought about during the field study were organized into recurring contextual needs. We developed the categories of the code by looking at the diary entries, and the interview notes. Overlaps between the categories was clarified to make sure entries fell uniquely into one category. \n
12. 22. I joined effort with another colleague UX Res, who helped me develop a coding scheme for the different contextual needs participants thought about during the field study. We developed the categories of the code by looking at the diary entries, and the interview notes. Overlaps between the categories was clarified to make sure entries fell uniquely into one category. \n\n Using the same technique, we could identify 9 barriers for adoptions of contextual services.\n\n> here you can see an excerpt of the coding scheme.\n
13. 24 contextual needs emerged from this exercise. We organized them according to the 4 contextual dimension that was suggested by the related literature.\nThe majority of the needs that we could observe in our dataset belonged to the spatio-temporal dimension (8 entries) and to the social and personal dimension (6 entries), while we could observe only 4 needs in the data that belonged to the activity dimension.\n\nParticularly [click] these contextual needs were not identified by previous work on mobile information needs.\n
14. Example. We named this category of contextual needs “social awareness” because participants expressed the need of monitoring what their friends were doing. \n
15. Using the same technique, we could identify 9 barriers for adoptions of contextual services. \nHere, some reasons dealt with the quality/quantity of the data provided by the service (B1, and B6, table 2), or the way private information was handled (B2). Other reasons focused on the interaction design of the service (B4, B5, and B8) that did not correspond to the users’ needs. Finally, some reasons focused on the lack of inclusion of the service in the ecology of routines and technological artifacts already in use in the users’ daily lives.\n\n
16. Example. Many participants reported the fact that they did not trust either the information provided by the service or the company that was operating the service. \n\nIn the quote one of the participants was explaining that many of the recommendations left in Around Me about Barcelona are not useful for citizens of the city because people living in Barcelona usually avoid touristic places.\n
17. To summarize, this is what we found. We observed 24 contextual needs. We do not argue that this list is complete as our study was exploratory with a limited sample and a short observation period. \n\nWhile for many needs the goal is that of increasing personal knowledge about things (e.g., weather forecast) –that is the informational nature Church and Smyth talk about [3]– other needs have different goals, for instance supporting a physiological function.\n\nThis scheme helped us understand that not all the contextual needs that we identified were covered by previous studies. Finally, we observed 9 barriers for adoption.\n
18. \n
19. Findings from this study extend those obtained in study 1. Particularly, we found that participants considered contextual needs derived from basic human needs as more important than those derived from high-level human needs. \n\nMoreover, the frequency with which participants currently address these needs is inversely proportional to their level of importance. These results might be related to the fact that participants lived in a developed country and therefore had a somewhat privileged socio-economical status (e.g., only 11% unemployed, 98% finished at least secondary school, 99% own at least one mobile phone, 95% use computers every day, etc.).\n
20. Also, we were interested in understanding how these contextual needs mapped to the more general human needs. To do that we referred to psychological theories of human motivation. Particularly, we referred to Maslow's theory of human needs. [explain] \n\nMaslow’s theory is often represented with a pyramid. It says that human needs can be organized in 5 hierarchical levels. For an individual to express needs in the upper part of the pyramid, the most basic needs have to be satisfied.\n\nWithout going into much details, [click] Maslow's theory helped us understand that most of the needs participants referred to during the field work, were associated to high-level human needs. \n
21. This result could be red in two different ways: [click] on one hand this means that if you want a contextual app to be successful in developed countries then you have to focus on needs in the upper part of the pyramid. \n\n[click] On the other hand, if you are developing an contextual app for an emerging market, then you have also the opportunity to focus on the lower part of the pyramid.\n
22. This result could be red in two different ways: [click] on one hand this means that if you want a contextual app to be successful in developed countries then you have to focus on needs in the upper part of the pyramid. \n\n[click] On the other hand, if you are developing an contextual app for an emerging market, then you have also the opportunity to focus on the lower part of the pyramid.\n
23. I do not have time to explain the details the design of the questionnaire and the rest of the findings. However, I want to spend one minute to talk about one of the most important outcomes: a quantitative relationship between the barriers for adoptions. Using this information designers and developers of contextual services can prioritize budget and resources to increase the popularity of a service and the users' satisfaction. \n
24. If you are part of a development team you can use this information to prioritize results. Using this information, a Product Manager could see that, privacy issues are generally considered more important than the difficulty of using the system, for instance and s/he could decide to give special attention to the privacy treats entailed by the services under development.\n
25. I would like to conclude this presentation with three implications for design.\n
26. When we looked at the percentage of time that participants satisfied their contextual needs by means of their mobile phones, we found activity and personal-related needs to be the least satisfied. However, the contextual needs in these two dimensions were pervasive in the lives of our first study participants. \n\nIn 20% of the cases, participants reported not liking existing applications in these two dimensions. Hence, our findings suggest that there is a need – and an opportunity – for novel mobile contextual applications that fulfill needs in the activity and personal dimensions. For example, a mobile application that monitors their users’ physiological state while exercising and helps them achieve pre-defined exercise goals [16].\n
27. Just to expand the information presented in the previous slide, if you look at the current panorama of contextual applications, these two areas are usually under-represented. \n\nHowever, we found evidences that support the development of applications that use contextual information from these two dimensions.\n
28. Second implication. \n\nThe reality today is that there are limited tools for feature phone application development, there is no application store and feature phone typically have more limited computation and sensing capabilities when compared to smartphones. At the same time, the vast majority (about 90% [20]) of mobile users in the world – and particularly in developing countries – use feature phones. \n\nHence, there is a need and an opportunity to develop intelligent contextual applications for feature phone users. One idea would be to move the applications’ intelligence to the cloud, thus bypassing the phones’ limited computational power. For example, simple location-based services could compute the approximate location of the users by looking at the closest BTS that (s)he is connected to.\n
29. Third implication. \n\nIn developing countries, a big part of the population is still fighting to satisfy their basic needs (e.g., safe food, basic health, safety, etc.). Given that most of them do have a feature phone, the challenge would be to not only build a smart device whose cost should be comparable with that of a feature phone, but also to develop contextual applications that could address more basic human needs. For instance, we can think about a service to map contaminated wells through user-generated content, or an application to share recipes to prepare well balanced baby-food from nutrients available in local markets. \n\nWe plan to focus on these challenges as part of our future research agenda.\n
30. Let me conclude by acknowledging my colleagues at Barcelona Media and at Telefonica Research that were instrumental to the development of this research.\n
31. Thanks.\n
32. \n