Contextual user experience research application in Low resource settings

Erick Oduor at UXDX Community: USA Online Series. Video: https://youtu.be/HezrhC6ihTw

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

Background: from Kenya to IBM to market research

[00:00:00] Erick: Today I'll talk about contextual user experience and give an illustration of how you can apply it in a low resource setting. Before we start, I'll give you a little background about my journey. I'm from Kenya, and I grew up there. When I speak about low resource settings, I'm referring to research that I did in Kenya as a graduate student and a PhD student.

[00:00:29] After that I joined IBM, where I conducted human computer interaction research for the first six years. The work that I did revolved around healthcare, financial inclusion, farming and family communication routines. Most of this work focused on capturing the cultural and social practices in these local communities, mostly middle income and low income, and then providing design recommendations on how technology could be designed to take advantage of the pervasive technology that existed, more so with the intention of modifying what already existed as opposed to creating new technology solutions.

[00:01:13] Within IBM, I then switched from Nairobi, where I was based at IBM Research Africa, and worked with teams in Yorktown and Japan on HCI for AI. There we worked on IBM's automated machine learning tools. We worked on visualization of model creation using these tools, and also on the trustworthiness of AI systems. I did that for a while, and then for the last two years at IBM I moved into UX and product, and for that I worked on IBM Cloud products. And then this year I switched into research consulting.

[00:01:56] In market research, I found that there was a different approach to how UX was conducted. At IBM, for instance, we focused on product and worked with product teams. When we were studying social practices in low resource settings, in Kenya, we focused on understanding communities to inform the design of technology. But now in market research we are focusing on creating business opportunities, while staying there to keep working. By keeping working, I mean taking the design processes that I'll talk about here from one stage to the other, as opposed to just conducting UX for a specific product and stopping there.

Defining contextual UX

[00:02:43] How did we come to this term, contextual UX? As I mentioned up here, it's an evolving definition. I started thinking about it when I switched from product research at IBM into market research. The first time I came into contact with this term was when I was writing my dissertation and reading Karen Holtzblatt's work, where she defined contextual design as a user centered approach that uses in-depth ethnographic research rooted in contextual inquiry. The data collected would be analyzed, and this would inform how the process of design would pick up.

[00:03:26] Based on that understanding of context, and also having worked in different contexts, the rural communities, the product context at IBM, and now market research in industry, I thought of another way of expressing a UX process that captured all these aspects of my experience. The Interaction Design Foundation defines UX research as the systematic study of target users and their requirements, to add context and insights to the design process through various means, and to uncover problems and design opportunities.

[00:04:06] For us, the starting point of our definition is that I see contextual UX as a framework that seeks to understand the social practices of a user based on their context, and uses that to inform the design of personalized and immersive experiences. In the case of my experience, it's more around digital solutions.

[00:04:32] Overall, the UX approach focuses on improving the digital experience of users, but more so, you work with different sets of teams. You work with stakeholders, and within stakeholders you have users, you also have people on the product, and then you have people who build the product. As you capture a process or a problem that you want to understand, you get input from these three sets of people on what approach or what goals they need met, and then you conduct a research method. Some activities that you conduct to help answer these questions are desktop research, design thinking and contextual interviews. And for products that are already in market, where you want to monitor their performance, you conduct heuristic and maturity assessments.

[00:05:27] Once you do that, you also provide deliverables back to the stakeholders. These could be prototypes, they could be reports, and they could also be recommendations on which ways to improve the product. That is the standard way of applying UX to solve user problems.

A circular process

[00:05:47] What we have here is nothing different from what you've seen before. We have five areas where we perceive UX could help improve. But the reason why we are drawing this in a circular way is because, as I mentioned, when I joined market research and started conducting UX, the first goal is to democratize user experience research. And the other, depending on what part of this circle you engage with your clients on, is to also show them other aspects of how they can improve their solutions.

[00:06:32] We'll take an example. At IBM, we worked on specific parts of this circle that you see here. For instance, we worked on products. We would test a product, conduct user interviews, solve a user problem, work with stakeholders and then provide feedback. But in industry research, when you get a concept from a client, you address a problem, but the other intention is to show them how you can move this work to the next step, which is design. It doesn't have to go through the whole circle; it depends on the budget and the interest of your client. The idea is to show them how this all fits together. That's why we have the circular part.

[00:07:16] Generally, to the right, when you're conceiving the concept and doing the design, it's more interaction, and you want to show what could be provided there, methods that are particular to helping solve that problem. And then to the right is more existing products that need to be improved. In terms of personalization, what we mean here is that several people can come to a website, for instance, but depending on their role, all of them would have different goals. That also calls into practice the need to understand the users and define personas. Those are standard UX methods. But what is interesting is that as you build that, sometimes you also need concepts to improve the performance of personalization. That is why we have a circular diagram.

Case study: hypertension management in rural Kenya

[00:08:09] Now we want to take that concept that I just shared with you and show how it could be applied in a low resource setting. To do that, I'm going to talk about one of the projects that I conducted while I was at IBM.

[00:08:29] First, how do you even identify projects for scaling in a low resource setting? There are several ways that this is done, and most of this is in research. You can look for unsolved problems that are in the market. A lot of work has been done in fintech, especially in Africa. You can also look at other solutions that were implemented earlier. In the area of healthcare, before the advent of video communication, we had telephone systems that hospitals could use to communicate with patients. For instance, there was a system in India where messages could be sent to women who were pregnant. Those technologies that were designed before the wide use of video could be iterated. And you could also pick up research recommendations from published work, such as in the area of healthcare.

[00:09:26] What we're going to focus on is work that we did in healthcare, where we moved it from research into a business product. What you see here is a rural health facility in Kenya, and they were interested in understanding how technology could be used to solve challenges in the area of hypertension management. This is the title of a research publication that we did, and you can see that we looked to explore rural community practices in HIV management for the design of technology for hypertensive patients living with HIV.

[00:10:10] What this meant was that this facility had undergone a very successful process where they used technology to monitor HIV within the community, and the country as a whole also had an electronic HIV management system that we learned from. But we focused on this local community because, in addition to using the technology, they were interested in following up with the patients once they went out into the village. That is where we started, and we conducted this study.

[00:10:42] But it stopped with the institutions, that was IBM and the health facility. Once we finished the research and presented the results, the project never moved into actual implementation. That is one challenge that we have in low income rural settings, or developing countries, as they're called. The problem is that when these research projects are initiated, we go to the community and tell them: we are here to learn about the challenges that you're experiencing, we're going to design technology and come back. But this facility was in my village, so even when it stopped, I still went back there. I had a connection with it, so we pushed it forward.

[00:11:27] I've skipped several steps here, but I'm showing that we iterated it into technology solutions. This is the interface of our MVP, and what you can see, in the middle of this diagram, is that you can monitor visits of patients and capture trends. Then to the right, at the bottom, you can see that we can also sync some devices. What we're doing here is digitizing reports for patients, and then these reports can be shared and trends can be captured. That solved one of the problems that we found in our research, where most of the treatments in hypertension were conducted on paper records, and when people moved from one facility to another, these records would disappear. We also integrated other solutions into this MVP of ours. Another one is that patients are able today to make appointments remotely via a chat interface, among other things.

[00:12:31] I'm just giving a highlight of how the project moved from research into a business product. But the lessons that we learned were, one, we needed to support the role of community health volunteers. These are people who are more connected to patients, because they go to the village, than the facility, where you go and see the doctor for about 30 minutes, they give you instructions, and you go back to the village. Two, patient data was addressed, in that we are digitizing records that would otherwise disappear, and over time people who have chronic illnesses face challenges when they can't access these records. And third, appropriation of technology. We wanted to have a solution that people could access using cell phones, which were already pervasive in these communities.

When research stops at the paper

[00:13:20] The key point, and one challenge that is faced when you work on such projects: this is just one, but as I mentioned, we've done several projects in healthcare, in farming, in water, and we face the same challenge. One of the challenges that I think should be highlighted is that with these projects, eventually we write research papers that people don't end up reading in the village.

[00:13:45] Changes in strategy happen. Institutions will set up these projects, and then once focus shifts, they drop them. This shift in strategy, and some recommendations not ending up being implemented because they're expensive, is a big challenge. Because when you go back to the community, it's usually a big problem when you start new research projects. When people have done research projects before you, the communities always look at you as if you are there just to meet your selfish goals.

[00:14:23] That is a big challenge that I mentioned, but we'll switch back to the project which I mentioned. We've moved it from a research recommendation to an MVP, and an MVP is already a product that you can share. I think one of the ways we can approach addressing the challenge of research that is incomplete is to think about ways through which we can access these projects, build MVPs, engage communities, and share information that helps people get to building technologies, or picking up projects, through opportunities that do exist.

Opportunities: training, new learning avenues and AI tools

[00:15:06] The opportunities are what I'm going to talk about now. The first one is that we can leverage technology training and focus on current trends to see how we can move projects that I think are stuck at a point where we conduct a lot of research, we provide a lot of recommendations, and they remain sitting there. One option would be to offer training based on available free learning material. What comes to mind here is Y Combinator training. They provide free learning materials on business models, pricing, startup funding, or even launching MVPs. In addition, you might also be lucky enough to get funding. But the information that they provide, I think, is very valuable. So if this information could be curated in some way and shared with people in low resource settings, especially people who are interested in those topics, that would be a very good way of sharing knowledge with people who can use it.

[00:16:02] We also want to capitalize on new learning avenues, especially after the experience that we had with Covid. There's a lot of information that is available on online forums. For example, now, once you have a product such as an MVP, there are trainings that can teach you how to design web pages highlighting products, how to build brands, how to spot opportunities in social media such as Facebook groups and competitor challenges, and how to build negotiation skills. What I'm saying is there's a lot of information that is available, but there's a need to curate it in a way that is meaningful, especially for people in low resource settings, where this technology might be moving so fast that they don't even see it or don't have access to it.

[00:16:56] The third one is to apply AI tools to automate activities. This will not solve all problems, but it helps you understand what is going on in the field. You can see there are tools such as SimilarWeb. This is software that you can install in your Chrome browser. It's an add-on, and you can monitor people who visit your competitors' websites, and you can even see the comments they're making, follow up, and comment in these groups. That way you also get traction, because you can just go to a group where there's a problem and mention: oh, try this other product, it's also solving that problem. There are simple things like those. Still, we won't underestimate the need to have some knowledge to be able to implement this. But the point here is that there are opportunities that could be explored to bridge the gap that has been created over time by the digital divide.

[00:18:02] On the topic of the digital divide, the way we think about it is that people in low resource settings need to catch up with people in the developed side of the world. I think, once you have an MVP, the rest of it is a bit of marketing and understanding other aspects of product development. That information is available somewhere, but I think what is missing is a way to curate it, so that the right people get this information and learn what they need to learn, so that they're able to scale their products.

What we are proposing

[00:18:36] So what are we proposing? We are highlighting three areas where I think we could start. First, we're proposing that we provide a basis of research that addresses key areas such as healthcare, water, natural resources and financial inclusion. These are topics that are very hot in developing areas, and I think we have a lot of publications that already exist in this area. If there was a way we could look at work that is sitting there, whose recommendations have not been implemented, it would be good to bring it up in some way. Again, it's training, or looking for ways to take this information and make it usable to the local communities. As I said, we don't have a strong culture in low resource settings of communities reading research papers for the findings.

[00:19:29] The other part is UX research, which I also refer to as human computer interaction, because the recommendations that we provided in most of the work that I mentioned were design guidelines. Those design guidelines, without interpretation or implementation, will just remain there. So I think we also want to look at today: as I mentioned, there are a lot of tools coming up, AI is advancing. Over time, we can go there and teach people how they can use user testing to conduct user research, but we also need products that solve our problems, and then conduct research on those ones. But there are ready-made products that exist. There are products that can help you with capturing clients who come to your website, monitoring what they do, and even collecting payments. Figuring out how to connect what we have with what already exists can solve a huge problem. Then you don't have to employ a lot of people to do these other parts to build a complete product on your own.

[00:20:36] And then finally, integrating UX solutions to have a complete flow. I think this is also a very good opportunity, because we have social media that people are now engaging in. It's easier to market your products this way. It's not easy in terms of doing, but I mean in terms of the opportunities that are available. Traditionally, some of these processes required large teams for implementation. One thing that I've liked over the last two years is that I've been following Shopify shop creation. I think there are a lot of automated parts of it that show there's big potential if you can move an MVP product, or a solution that someone has. There are a lot of integrations that you can add: to help you with payments, help you put your website online, help you with marketing and all these things. You can learn them. But I think the challenge that still remains is how this can be provided so that it's useful to people who are in low resource settings. And that brings us to the end of the talk.

Q&A

[00:21:52] Host: Great, thank you very much, Erick. I thought that was a really interesting talk, and it hit on some of the challenges that I guess are a bit unique to the low resource setting, but also very familiar. That brings me to my first question. You hit on the point that all the research gets done, but then often it just languishes there and doesn't get implemented. When I was looking at your solutions, it was: well, how can people still get access to that research, all of those research papers that have been written and all that information? But how do we solve the problem that's still there, that gap of getting it implemented, without relying on just having the good luck to have you in the village to build the MVP?

[00:22:43] Erick: My view of how this could be done is that someone has to do the heavy lifting in the beginning. What I mean by heavy lifting is: I was part of a project. I have a better understanding of the overall view. I initiated the project. I've worked with these people in different countries. Saying that you have to get that information just the way that I did... I think I have a bigger role now, to simplify this information in a way that people can use it. It's not work that somebody else will do. It's work where we will set up communities, just the same way UXDX is set up. Someone had to do the work to bring a community together. So we need to do that.

[00:23:41] I think we need to collaborate with people who are building and doing research, and we need a different way of presenting research to people in low resource settings. They still need to know how they can play a part. If we don't do that, I don't have a solution, but I think we need a community to think about ways through which we can present this research. Otherwise, it looks like you came, conducted studies and left, and we'll continue with our work. So it's a process, but the first part of it is to form a community that will approach this solution.

[00:24:19] Host: Great. And I love it; we had a quick chat beforehand. I'm quite bullish on AI and where it's going to go over the next few years, just with what we've seen with GitHub Copilot and others this year. I'm hopeful that over time it'll actually become a lot easier to build the software, but the challenge is actually going to be in the research, and making sure that we ask the computers to build the right thing. So it's really interesting. I think your way of trying to bring all of that research up is going to be incredibly valuable, then.

[00:24:57] Erick: Yeah. And as I said earlier, I think that's a big opportunity. AI provides a shortcut, but it's also a challenge. A shortcut in terms of the parts of automation. But also: are we going to be able to bring our communities to a level where they can input information to support this automation? Because, as you understand, AI also depends on the data that it's fed. So I think that means that people like me, or people who are interested in advancing this work in developing settings, need to quickly think of ways of coming together and thinking it through. Nobody else is going to do it. So I think it's just us, and then the communities that we build, people who are interested. And then we'll make some steps. We might not solve all the problems, but we can make the first two steps towards solving the problem. So I agree with you.

[00:26:09] Host: Excellent. Well, I think that brings us to time. I think it was a really interesting talk. As I said, I think people are seeing that same problem, whether it's in a low resource setting or in high resource settings, where all this research gets done and yet it isn't being implemented enough. So I think there are some great ideas there on how to solve it and bubble up the research in the right way. I hope that's useful for everybody out there who's watching. I just want to thank you, Erick, for sharing today.

[00:26:40] Erick: Yeah, thank you.

[00:26:41] Host: Thank you.