Editor’s note: Continuing our exploration of how technology has been used positively to improves people’s health, we find in Debbie Rogers a story that somewhat mirrors Jonathan’s from the previous episode. Not just because both were colleagues at Praekelt, but because both started with a creative background and later ventured into a social enterprise realizing the kind of impact and meaning they wanted to create. Reach Digital Health is an established organization in South Africa with a long track record of impactful solutions. We’ve seen a similar mentality in Season 1, with Sanjay Purohit of Societal Platform: if you want to achieve massive scale, you need to 1) leverage digital technology, 2) understand the key moments in a specific scenario on which that technology will be built, 3) focus on one small, simple user action at a time. That is what Debbie’s team was able to accomplish, leveraging various messaging technologies time and again to meet the needs of national health programs, and provide health support across South Africa.Â
Transcript
Jim Fruchterman
Welcome to Tech Matters, a biweekly podcast about digital technology and social entrepreneurship. I’m your host, Jim Fruchterman . Over the course of this series, I’ll be talking to some amazing social change leaders about how they’re using tech to help tackle the wicked problems of the world. We’ll also learn from them about what it means to be a tech social entrepreneur, how to build a great tech team, exit strategies, the ethical use of data, finding money, of course, and finally, making sure that when you’re designing software, you’re putting people first.
Today, my guest is Debbie Rogers of Reach Digital Health, one of the coolest, I think, large-scale implementations of digital technology for health that I’ve run into in the entire world. So Debbie, thanks for spending some time with us on the podcast.
Debbie Rogers [0:54]
It’s an absolute pleasure.
Jim [0:57]
Well, great. Well, our sort of routine is that we want to get into, you know, why is a technology person like you in a nonprofit? So tell us a little bit about the story, get there, and then we’ll get into the wonders of what Reach Digital Health actually does. So what’s your story?
Debbie [1:14]
So when I left school and went to university, I really wanted to do something practical. I wanted to do something that I could almost see, touch, feel people could interact with. I came from a family of academics and I wanted to go against the grain to an extent. And so I chose electronic and software engineering as my degree. And I really enjoyed that, but to be honest, I found the bro culture really difficult. I was one of eight women in a class of 180 men and I wasn’t feeling confident enough in myself at the time to feel like I could belong in that space.
So I actually sort of turned away from it for a while. And I went in completely the opposite direction because swimming against the grain is a little bit of a theme with me. But I then did my master’s in digital art. So in digital art, basically what I was doing was using digital installations to make a comment about something in the world. And a lot of what I talked about was the intersection of arts and science. A lot of it was around feminism and my place in the space, particularly in the electrical engineering space. And I felt like I was saying something important. I had this ability to talk about the world that I hadn’t had in electronic and software engineering. But at the same time, I didn’t feel like I was changing anything. So I was saying a lot, but I wasn’t actually enacting any kind of change. And so I became a little bit disillusioned to be perfectly honest. I didn’t really feel like I fit in the electrical engineering field. I didn’t really feel like I fit in the art field. I was a programmer for a split second. I was an artist for a split second, but neither of those really fit for me. And just perfectly by chance, I ended up being offered a position as a multimedia design lecturer at a branding and marketing school in South Africa.
Jim [3:26]
So you went back to academia.
Debbie [3:28]
Yes, I went to academia, even though I saw that was not what I was going to do. I was a lecturer for a few years and I learned so much about art and advertising, marketing, creativity in now a third different way. At the end of my tenure there, I was starting to feel a little bit antsy. I wanted to get out of the academic space because I promised myself I wouldn’t be in there.
And I was asked by a friend to come and interview for a position at an organization called Praekelt Consulting. And this organization had been started by Gustav Praekelt and it was a digital media marketing and advertising organization. When there were very few of those around, it was quite pioneering in its time. And so I interviewed with him and I went to the position. And one of the amazing things about this was that he had recently started a nonprofit business and that was really where I found my happy place. I felt like not only was I saying something about the world, trying to make a difference in the world, but I was actually enacting it as well. And I started just exploring service design in that space and set up the service design department, ran the service design department for a while. And long story short, ended up a few years ago taking over from this as the CEO and rebranding the organization as Reach Digital Health.
Jim [4:57]
Wow. So, the tech industry does have a very strong pro culture. And I think that one of the attractions to tech for good, especially nonprofit tech for good, is that culture is a lot less than evidence. And of course, part of that is that an even majority of the CEOs in our field are women, which have less of a tendency to tolerate such a culture. And I think that’s one of the reasons people love working in it.
And you found it. You found that. So great. So now, all right. And of course, my understanding is Praekelt was a tech, for-profit tech leader. And like many people who’ve been successful, wanted to do something to give back. So let’s talk about what did this Praekelt Foundation turning into Reach Digital Health, what were the key products or services that you guys delivered that got you going?
Debbie [5:55]
Yeah, absolutely. So I think, you know, really, the idea behind Praekelt Foundation was that Gustav was seeing so many people have access to mobile technology. So at the time, when he founded it, mobile phone penetration had eclipsed radio penetration. And so he saw this potential, but also saw living in South Africa, a lot of challenges, particularly around HIV, sort of the height of the epidemic in the country. And it was affecting so many parts of our society. It was people’s health, but also our economy and culture and various and politics, exactly. And so he thought, okay, there’s these really big challenges. There’s also this huge opportunity that only Africa has at the moment.
How could we use this opportunity to try and solve some of these challenges? So we started off very much in health and in HIV. This was sort of the premise under which we had been founded. And so we started off one of the very simplest initial programs that we did was we went to a clinic where people were receiving antiretroviral medication. And they had a huge problem with loss to follow up. In other words, people were coming to their appointment, they were getting their antiretroviral medication, and then they were disappearing. And nobody knew what had happened. Nobody knew if they had just moved to another clinic, or if they had perhaps passed away, or if they were adherent or not adherent, or what was the problem. And 30% of people doing this was massive. It’s a huge number of people. And so they wanted to try and address loss to follow up, thought that mobile might be useful. And we found a lot about this was really just every sort of three to six months, you would be coming back for another appointment, you might forget your appointment when the appointment was given to you, you might not have known that on that day, you needed to work for some reason, you might not have money to pay for transport on the day you’re supposed to go to your appointment. And there was just no way for anybody to be able to change their appointment to be reminded about their appointment. And so the appointment would come and go and they wouldn’t know, can I even go back to this clinic and would disappear off the map. And so we implemented super simple SMS technology, which was reminding somebody about the appointment two weeks before, around reminding them two days before. And then if they didn’t attend their appointment, encouraging them to reschedule their appointment. And the reschedule was done in quite a cool way.
Basically, there’s this thing in South Africa called the Please Call Me. And what it is is you can send a free message via the mobile network operator to somebody else to say, please call me on this number. So it’s completely free of charge. And we use that technology to allow people to send a please call me to the clinic. So the clinic would call them back and reschedule their appointment for them. So there was no cost to the user as well, which I think was super cool. And we helped to reduce loss to follow up from 30% to 6%.
Jim [9:18]
Wow. And of course, in terms of funds spent, I mean, when you have a disease that is so often fatal, if you don’t take your medications, reducing non-adherence by 25% saves an awful lot of lives. And I’m guessing for relatively small amounts of money.
Debbie [9:38]
Tiny amounts relatively. Yes.
Jim [9:40]
There’s no exaggeration that Africa led in so many of these mobile areas, mobile money, mobile health. I think my doctors in the US got texting me to remind me of my appointments like 10 years after you guys were doing that, so I often use this as an example. The first thing you’re doing is you’re doing this in partnership with the government?
Debbie [10:04]
This was actually initially a private, well, a clinic run by another NGO, so it was being run at a government hospital, but by another NGO. So we’re working mainly with the NGO.
We had this advertising and marketing background, so we were doing work for organizations like MTN, a mobile network operator, or Vodacom, also mobile network operators, South Africa, and we were just seeing opportunities like, oh, MTN has free advertising that they’re not using. Maybe we could use that to direct people to the HIV helpline, which already exists. And that project was, we drove, we sent 2.1 billion please call me messages advertising, please call me messages for no cost, and drove 1.9 million calls to the HIV helpline, just with that simple idea.
Jim [11:00]
Well, I think it goes without saying, with a data-oriented culture where you’re actually measuring and experimenting and using the data to inform, oh, this work, this didn’t, let’s do more of the thing at work. And that’s not necessarily, when it came to digital communications, a natural for other NGOs or the government.
Debbie [11:19]
No, definitely. I think that one of our big advantages was where we came from, because I saw a lot of the, at the time, a lot of the work that was being done in mobile technology and in health, I think it was very cold at the time, was being driven by researchers. And so researchers were really coming at it from like, this is the public health problem. I’m going to create a mobile technology to solve this and very focused on what is the, what is the outcome, which is not in itself bad, but what was happening was that there were lots of little experiments happening all over. “Pilot-itis.” Exactly. And there was not a thought about scale. The thought was, how do I get my research project done? How do I show outcomes? And then that was the end of that, whereas I think we came from the marketing side, we were like, go big or go home.
That’s kind of how you think about it. You know, we were used to driving millions and millions of people to buy new mobile phones. We didn’t see a problem in doing that for, for mHealth. And so I think that really shaped the way we were thought about problems and set us apart from where a lot of the other innovations in the field was happening. I mean, maybe, that wasn’t necessarily perfect either. We probably had less of a focus on impact than we should have had initially. We had a lot of focus on scale and a lot of focus on can we do it and how do we implement it? And what is the path to scale? So we had a lot of focus on that, didn’t necessarily have robust data to say, yes, it’s worthwhile doing this at scale. We just sort of, you know, used some of our intuition around whether or not it was useful. Well, it’s harder.
Jim [13:10]
It’s harder to measure the ultimate impact in society on, let’s say, something like health outcomes than it is to measure the impact of an advertising campaign in terms of getting responses, right? Absolutely.
So how did you get from COVID to maternal?
Debbie [13:24]
From HIV to my total health, yeah.
Jim 13:28
Oh, sorry, sorry, sorry. [laugh] I’m leaping ahead to COVID. Okay, here we go. HIV first. Yeah. Maternal next. And we’ll get the other one later. Okay. So tell me about maternal.
Debbie [13:38]
Yeah, there was this wonderful program called MAMA. It was an initiative by the UN Foundation, Johnson & Johnson Baby Center in USAID. And essentially what they did was Baby Center was a US company that had been doing stages and stages based messaging to mothers in the US via things like email for a number of years. And they’d seen this and they’d also come from all the marketing background. They had this incredible uptake from users in this way of using ages and stages based messaging. Ages and stages being if you’re 10 weeks pregnant, you need certain information. If your baby is 14 weeks old, you need other information. So targeting the information based on the age of the stage of the pregnancy or the age of the baby. And they thought, okay, well, how can we do this in the global good space? How can we use this in the public health space? There was immediately a recognition that email and websites, which is what they were using, was not going to be the way to do it, thank goodness. So they focused on SMS messaging and packaging to ages and stages messaging in sort of an SMS type format. And then what the organization did, they identified four countries where they felt that there would be the biggest need, but also the biggest impact and the best mobile penetration. So really set up the program for success.
And they chose Bangladesh, India, South Africa, and Nigeria. And what they wanted to do was to run an ages and stages based mobile technology platform in each of these countries. And they went out and they looked for organizations that were working in the space. And we were lucky enough in South Africa to be one of the lead organizations of a consortium to run this program. Now, when we started, this was in 2011. We didn’t really know, is SMS going to be best? Is voice going to be best? Is a mobile site going to be best? Even back in the day, there was this crazy South African social media site called Mxit, which had millions and millions of users. It was kind of like a really basic WhatsApp. And we wanted to experiment with that. So what we did is we experimented with all types of different channels. And we started getting really large scale on certain channels, really good results from an impact perspective on other channels, kind of like trying to see which of channels was both most cost effective and most effective. So we were trying to experiment around that. And so we built up about I think 750,000 users on the platform. And we had learned a lot, a lot of what not to do, by the way, as well as what to do, which I think is sometimes more important than what to do, a lot of what not to do. And the Department of Health in South Africa had just published a normative standards framework document. So now the normative standards framework document, very exciting title, was basically just about how from a technical and security perspective, how do you pass data between different systems? So it talked about, like from a mobile technology system to an eHealth system, how would we pass data from a mobile phone to DHIS2, for example, it even spoke about…
Jim [17:12]
DHIS2 is a very famous district health, basically running clinics and regions for national health ministries.
Debbie [17:20]
Absolutely, absolutely. And we had DHIS2 to set up in South Africa. So it was an important part of our e-health strategy at the time. And the Minister of Health at the time, Dr. Aaron Motsoaledi, wanted a use case for the normative standards framework. He didn’t want it just to be a PDF that lies somewhere. He wanted us to, wanted to show that it can be used. And given what was really a terrible state of maternal health in the country at the time, he wanted a use case around maternal and child health.
And what the National Department of Health did was they looked around and they saw people who were doing work in maternal health. And we were one of the organizations who was doing work in maternal health. And they brought us all together. And they said, okay, let’s have a hackathon and let’s see if we can connect your system to DHIS2 using an Open HIE framework, which was the interconnectivity framework that they wanted to use. And we were able to do it within about two hours, I think, because it’s technically quite a simple system, right?
Jim [18:29]
So basically, you’re trying to connect SMSing mothers, in this case, with the health information system of the National Health Service, which is basically making that connection from health promotion or answering questions, maybe, to actually getting into the system of the government health system. Exactly.
Debbie [18:50]
And we were able to do it technically really quickly. And sometimes when you can do something. As it often turns out, it’s much harder to get people to use.
Jim [18:58]
Than it is to get the technology to work, yes.
Debbie [19:01]
Yes. And as it’s often one of our problems as technical people, if we do something really quickly, we set up an expectation that everything can be done quickly. And so the National Department of Health said, right, well, we have an election in three months time. And we would like to set up a national pregnancy registry where we use mobile technology to gather the details of every pregnant woman in the country, feed that through to DHIS2. And we’d like it to launch in six months, and we’d like it to be national scale immediately. And they looked around the room and said, who’s in for it? And pretty much everyone else left, saying, you’re crazy, this can’t be done.
We may have been able to do it technically, but there’s no way we can do it from an implementation perspective. But because we have been spending so much time learning the hard lessons around the mobile messaging, and also because we really had great user insights around the value that we provided mothers. So interesting side note, I’d just come back from maternity leave when this happened. And when we were in these meetings, literally, I think the week that I came back, I started in these meetings. And I was sitting there and thinking, I don’t think I’d give my information to the Ministry of Health for no reason. I’ve got a small baby, and I just don’t see myself doing this. And then we thought about it, and we thought, but what we have in terms of messaging, people love. Like 98% of mothers would recommend the system. So this is a really big value to the mothers. What if we combine the ability to send messaging to the mothers with the ability to register mothers? And so the value to the mother was that they get the messaging. And in order to do that, they give their information. And so that’s how the pregnancy registry turned into a system that also did health promotion messaging. And so Mama evolved through this process into what we now call MomConnect. Luckily, we didn’t launch it six weeks later. We launched it six months later, which I still think is pretty impressive at national scale.
Jim [21:17]
Yeah, pretty darn good, but I just did.
Debbie [21:20]
Yes. But, you know, six months later, August, we launched the program nationally, or the Minister of Health in South Africa launched the program nationally.
And, you know, I guess, in hindsight, it was one of our flagship programs, and Ant has really kind of put us on the map as an organization. But also, I think, being a flagship for the National Department of Health in South Africa. And that was our beginning of our journey with with National Departments of Health and working with government instead of sort of alongside well
Jim [21:55]
And I think the other point you made, which is, and we often see this, when a group of men get together to design something for women, maybe they miss some of the heart of what’s going on here. And so I just, I think the idea that you had that lived experience helped you go, you know, people are going to do this because they should do it. We have to create a reason why they want to do this.
And you knew what the transaction was, we’re going to give you great information. Oh, and by the way, we’re also going to register you with the government, which is going to have some benefits down the line. Did MomConnect also become answering questions? Was that like the initial part or did that evolve later?
Debbie [22:37]
No, it did actually, and that’s actually why we launched a bit later, because the minister said, I’m not launching this unless we can have a help desk associated with it, which was great insight. And we also said, oh, you know, actually on our Mama program, people message back, even though we never expected them to, we expected people would realize there was a machine at the other side of it, but people thought it was their health worker, so they were messaging their health worker to ask them questions.
So that insight was the foresight of the minister of health. We thought, okay, we have to make this bidirectional. We have to be able to also answer questions. And we have to allow people to be able to lodge a complaint, give a compliment, like really understand what’s happening on the ground, because you can’t, you can’t have this direct communication from the national department of health to a matter, and then expect that they’re just not, not going to want to give complaints. There’s going to be issues on the ground, and it’s going to be useful for the national department of health to understand what the issues are on the ground. And so that was also a really important part of the program. In fact, the minister of health, it’s one of my favorite anecdotes around MomConnect was the minister of health when he launched Mom Connect in August 2014 said, “MomConnect is a bazooka in every woman’s hand,” which is a little dramatic.
Jim [24:02]
Wow, must have been an ANC politician.
Debbie [24:09]
He was indeed, and definitely from the struggle. What he was absolutely right about is it was incredibly empowering. The ability for a woman who has probably been neglected in her care at various points in her life, has not been listened to, is underrepresented, can actually speak directly to the National Department of Health and say, I have a problem. And there was a really tight turnaround time on resolving those issues. The complaint that came through, he wanted a report on that every month. So there was direct action from a national level directly to the individual. So it was incredibly empowering.
Jim [24:49]
Wow. And I love that sort of power shift, right?
So, MomConnect’s been around more than 10 years. Yes. And you guys are at national scale. So, every year, I bet you’re sending out a lot of messages and serving a lot of women.
Debbie [25:05]
Yes, so far, we’ve served 5.2 million mothers over the course of the program.
At any one time, we have between 300 and 400,000 active users. Obviously, mothers cycle out as their babies get older. We don’t have very few opt-outs. At our peaks, when we have our peaks in terms of registration, we can register between 60 and 80% of the mothers who give birth in a year. And it’s at 95% of the 10X.
Jim [25:45]
So now, of course, MomConnect, the signature sort of initiative that’s really well known, but there’s more, you know, so you want to move on to talking about what happened during the COVID pandemic?
Debbie [25:59]
When COVID-19 started, and particularly when it first hit the African continent, we were sitting there and we just thought, we have to do something. This is absolutely the opportunity for mobile health. So what do we have? And we realized we had two things. We had MomConnect, where we had millions of users. So we could message them immediately and let them know about things. We’d seen this when there was a listeriosis outbreak. And so we realized we can reach out directly through the channels that we have. The other thing that we had is we’d been developing a product called Turn.io, which we’d spun out. And we’d been developing this to allow us to very quickly roll out WhatsApp messaging programs in particular. And we had been realizing that we could actually spin up a WhatsApp line and get stuff going in a matter of weeks. It was actually pretty quick because we had the space of this technology. And so we thought, okay, we must leverage as many people as we have right now. We must send the messaging. But for those parts of the population that we’re not reaching yet, let’s set up a WhatsApp line. We’ll call it contact NUH. It’s not for COVID-19. It’s for contacting the National Department of Health and being in contact with them. And we’ll put the line up. They will add COVID-19 messaging to it and we’ll see what happens.
And so within two weeks of obviously in partnership with the government, and in two weeks of the pandemic being seen on the African continent, we’d launched this line. Yeah, so it was a super quick turnaround. We needed content. Content didn’t exist really because so many people were still working out so many things about COVID-19. So we reached out to the WHO and we said to them, do you have content that we could use? Because we’ve set up this mobile line. We need content. The National Department of Health is struggling to keep up with it. How can we leverage content that you have? And they said, you’re welcome to our content, but can you do something like this for us too? Because we also need this. So within a few weeks
Jim [28:32]
Lots of other countries.
Debbie [28:34]
Yes. Well, it’s the WHO. So we launched within another three weeks, we had launched a platform for the World Health Organization, a global platform in six languages, the six UN official languages. And we’d launched that platform for them. So suddenly we had the National Department of Health, World Health Organization, both of these really big institutions in the health space. We’d managed to move quickly, we’d rolled out these things for them. And we were seeing massive traction. So we were seeing hundreds of thousands of people onto each platform like every day.
It was absolutely exploding because people were so hungry for the information, they really wasn’t anywhere else where they could get it because we’d managed to move so quickly. And so we published a little article about this. And next thing we knew, we had 40 different countries requesting the same service. We were at the time 40 people. So that was a challenge. But within three months, we’d launched in 11 countries.
Jim [29:46]
And so so the essence of this is that people could come in and ask questions and it would have a menu tree that would direct you to the information you wanted. So kind of a classic, you know, interactive response, but for texting.
Debbie [30:01]
Yes, exactly. Okay. Very simple.
Jim [30:03]
And you were already in the process of building up WhatsApp capability, so it just happened that at the moment this hit, you were ready with the technology to spin it up in two weeks. Because obviously, if you’d started from zero, then you wouldn’t be able to do that. So basically, Turn.io is that basic technology makes it easier to deliver material. So what are you guys up to now? Are these things all still operating? Or have they morphed from being COVID information platforms to general health platforms? Has the technology shifted? Because I’m guessing things are moving fast. You guys are moving fast, right?
Debbie [30:44]
Absolutely. We learned a lot. And some of it was fantastic. And for example, you know, we still get more new users on Contact NDOH in South Africa every month than we do on MomConnect. And that is literally just because the line, the advertising for the line still exists out there. We haven’t done anything with it. We haven’t done any advertising. But people started putting the number on stickers and on posters. We had it put on bread packaging. Everyone got hold of it. The number was really catchy. It was 0600 123456. So another marketing technique used really simple to remember number. And so we still get enormous numbers of messages through there and through the World Health Organization line.
And we have now pivoted those lines to being more general health. So whatever the priorities of the National Department of Health are at the time, we make sure that we have content available for that. We can do outgoing messages to alert people at certain health issues. And we’ve kind of started evolving those platforms in that direction. Unfortunately, we also learned a lot about what doesn’t work. Or fortunately, I don’t know, but what we did learn is that in those other 10 countries where we rolled things out, we moved very quickly. We knew we weren’t in those countries. So we couldn’t be on the ground. We didn’t have the connections into the information. We didn’t have enough staff to be able to be on the ground or in managing all of those platforms. So we worked with partner organizations to manage the platforms. And so we gave them the content. We gave them the technology. We did a bit of training, and then we said go. And unfortunately, while the actual technical elements of it worked really well, the sort of strategic elements of growing the platform, as well as maintaining the platform as a continuously evolving platform that becomes a general health platform, didn’t really happen in those other 10 countries. And so we had amazing success in the countries where we were not only doing the technology and the content, but also the implementation. And where we weren’t doing the implementation, we had mediocre success or even failure.
And so what we started to look at is we were thinking like, we were a technology organization. We provided the technology. We did it in a really replicable way. We did great content. We provided that. Why didn’t this thing work? And what we started to realize is that, and we should have known this before. We did kind of know this before, but I guess it took this kind of realization for us to take it seriously, was actually the hard part, the harder part is scaling and sustaining. So how do you implement a program that is scaled? So getting people onto the platform, the hard job of getting people onto the platform, something that as an organization that came from the marketing side, we kind of did naturally, didn’t really think too much about how we did that, sustaining it, keeping it fresh, making sure that it’s always solving a problem for people, making sure that we understand what is it that users want, and we’re adapting to what users want rather than staying static, and then sustaining as in making sure that you’re working closely with government, you’re working around their priorities, and you’re making it cost effective. So how can you make it cost effective at scale?
So those things were actually just as important, if not in some ways more important than the cool technology that we had built. And so we realized if we’re going to replicate, we can’t just invest in building cool tech. We have to invest in monitoring, evaluation, research, and learning platforms, like sustainability strategies, demand generation, capacity building, partnership. These are the things that we have to actually invest in, and not only the technology. And so we were lucky enough to get some funding to invest in that, and it’s interesting because technology is such a tangible thing to invest in, right? And then to say, oh, it’s actually not the technology that’s cool, it’s all this other stuff.
Jim [35:32]
10% of the challenge, right?
Debbie [35:34]
Exactly. It took a little bit of persuading, but we were likely enough to have some sympathetic ears in some funders, and we were able to invest in that. So we said, okay, we’re going to have a much more proactive way of engaging with governments. We’re going to make sure that the government says, this is what I want. Then we’re going to bring both the technology and all the other frameworks that we’ve developed, and we’re going to work with them to build out something much faster than we did for MomConnect, because we learned many lessons from MomConnect. It took many years to get to that point.
We’re going to work with governments to expand. So we expanded into Mozambique, with a MNCH program, into Zambia, with a pandemic preparedness program, and into Kenya, with a maternal and child health program. Then we’ve got Nigeria, Tanzania, Rwanda, all asking for the program, and we’re busy raising funds for it. So we have expanded. Wow. We’ve got some experience in actually replicating and scaling it, which is our first one, which was Mozambique. Then we’ve got a lot of interest from governments who say, this is definitely what we want. We’ve seen what you’ve been able to do. We buy into how you want to do it, and this is how we want to move forward.
Jim [36:56]
Wow. So the really interesting thing is that it didn’t turn on the technology, it turned on the program and the execution that your attempt to franchise what you did, there was more to the secret sauce of the franchise than just here’s the technology and here’s how it operates.
Debbie [37:11]
Yeah, absolutely.
Jim [37:12]
So I think that, I mean, that’s a generalizable sort of challenge. I guess the other thing is we kind of talked to what are some of the big issues in the field.
You guys are also known as a leader in using AI and not from a, oh, AI, it’s going to solve everything, but instead from deep experience of operating at scale. So how are you using AI and what lessons have you learned that you think are generalizable to anyone who, because everyone in tech for good is now pretty much like, what am I doing about AI, right?
Debbie [37:45]
Yeah. So we actually started using AI back in 2018 because we had a big problem. We broke the help desk of MomConnect because we had so many users using WhatsApp. They were so engaged. We just completely broke it. And so we needed to be more efficient. We had two help desk operators to serve everybody. And we had tens of thousands of messages coming in and we needed to plan. And so we also had lots of data. And so what we did was we used natural language processing to sort through all the queries that were coming in. Originally, they’d all been answered by hand, sort through them, classify them, respond to the ones that we could respond to, and escalate the others. And that was how we initially used AI way before LLMs. But very much based on a need.
Debbie [38:41]
So I guess that’s our general approach. How we’re using it in some really interesting ways is, again, focusing on some of the problems we have. So one of the problems we have is the more we personalize our communication, the more likely we are to have an impact. So the more we understand where somebody is coming from, the more likely that we’ll be able to give them messaging that will resonate with them, which will help them to change their behavior. Now, in order to do that, you need a lot of information from that person. And because of the large scale that we work at, this information has to come directly from the person. It’s not coming from a medical record system, for example, because there are very few medical record systems at national scale, and these sorts of challenges. So we have to ask people a lot of questions. The more questions we ask people, the more likely they are to disengage. So we’ve got a catch 22. We’ve got: the more information we can get from you, the better the service will be. But the more information we ask of you, the more likely you are to leave the service.
So how do we actually optimize that? So we’ve been using GenAI in a project with OpenAI to kind of use natural language, sorry, LLMs to engage users in an optimal capacity to gather information from them and to make sure we’re not overloading them with questions. So that’s like one of the areas. The other area that we’re seeing is around multimodal work. So I’ve spoken a lot about what we’ve done on WhatsApp. As you said, we also do a lot on SMS and USSD. We’ve not done a lot on voice. We’ve done some, but not a lot. But now voice is becoming more and more important on WhatsApp. If you think about how people use WhatsApp, it’s not just a text-based solution anymore. And WhatsApp have opened up their API recently to allow for voice calls. They allowed for voice notes. So it’s now becoming a multimodal platform. Now, this is amazing for us from an impact perspective. So for example, in Mozambique, 50% of people speak a language that is not written. So we can’t communicate with invitex. It’s impossible, but we can absolutely communicate to them via voice. And so now we can use the same channel to engage with them over voice. And suddenly we have 50% more users that we can engage with and likely have more impact with. So that’s the type of stuff that we’re using voice for. Now voice is notorious, if you think of the old school call centers, for being highly resource intensive.
Jim [41:20]
Yeah, expensive.
Debbie [41:22]
Expensive, expensive, but with AI, now suddenly we can use GenAI to be far more effective, cost-effective. A lot of the things we’ve developed from a text-based perspective like identifying intent and answering questions without escalating into a helpdesk operator, these sorts of things can now be done with voice as well.
And so for us to do voice multimodal at scale, we have to be using generative AI.
Jim [41:50]
But we’re not talking about voice, and yes, I mean, obviously you’ve got English or you’ve got Portuguese and Mozambique, but a lot of people don’t speak those languages. So as you’re working your way through the national language groups in these countries, you’re actually getting voice recognition from AI that actually works well enough to actually do kind of the same work you would do if people were texting you in the number one language in the country. No? Okay. Are you getting in that direction?
Debbie [42:27]
Yes, but we are heading in that direction. So through partnerships with organizations that are really focused on gathering data around resource-poor languages, through partnerships with them, as well as the fact that we are operating at large scale, we already get a lot of data. People have been sending us voice notes for years. So we have a lot of data that we can use as well to train models.
So in partnership with people who are dedicated to increasing resources for resource-poor languages, as well as the fact that we have large scale reach already and we can leverage our own platforms to be able to train. That’s helping us to reach people. And so we’re incrementally improving how we can communicate with people in resource-poor languages using AI. But we still have a way to go. It’s one of the big challenges in the field.
Jim [43:26]
But the backdrop of this is you have a program that works, you have tiers of ways of responding that are increasingly more expensive from delivering vetted content to recognizing a question or helping someone navigate to the question and getting a canned answer, to conversational AI trying to get you your answer without needing a human, to a human help desk operator, to a referral to go into a physical clinic or whatever. So you’ve been basically optimizing this entire sort of healthcare system, and every year you find ways to respond to more and more of these at a cheaper tier, I’m guessing. That’s what I’m hearing.
Debbie [44:06]
You’re absolutely correct. Yeah, that is exact. That’s the plan, okay. Started from every SMS got responded to by a helpdesk operator 14 years ago, to now I think only 10% of our queries gets forwarded to the helpdesk.
Jim [44:24]
Wow, okay. All right. So that’s it’s amazing and and I’m guessing, you know, like if I express that in you know cost, you know dollars or Rand per Person responded to that number just keeps going steadily down and You’re probably getting factor of 10 or better cost reductions compared to humans answering them Well as we so we’ve we’ve had wide ranging conversation covered a lot of territory So as we wrap up, are there any any other issues that you want to like bring up of something that you want to share with our listeners?
Debbie [45:01]
I think this is something I’m learning, not something I have learned yet. And I think that’s really, you know, looking at the state of development at the moment, we’re all facing tough times. And I think we’re all reassessing, like, are we doing things right? Are we doing things in the right way? How can we be more effective? How can we be more focused? Because, you know, resources are constrained. And that is the reality for the next, the foreseeable future.
And what I’m starting to think a lot about is more about working in collectives. And I think that there are amazing organizations out there like Reach for Little Health. I can name many that I really admire. If you want to find out about them, just look at the Skoll awardees for the hospitals. There’s being recognized there. And we’re all working really hard to solve a problem. But we’re all pretty much scratching the surface. That’s the reality, really, is we are all scratching the surface of the problem. And so I don’t think, you know, in these sorts of resource constrained times, all of us scratching the surface separately is not going to get us to where we need to get to. We absolutely have to start working together. We need to start pooling resources. We need to start pooling learnings. We need to start pooling research. And there are, and a fantastic example of this is the Health Workers Coalition. They were a Skoll awardee last year. And I think what they’ve managed to achieve by bringing organizations together and sort of their radical Canada approach and radical collaboration approach has been really incredible. And I see, you know, you can see the impact on the continent, not just at a country level. And so I really think that sort of the future of us doing really impactful work is starting to work more within collectives. And that’s going to take a lot of stepping back and a lot of checking our egos at the door. But I think it’s going to be the way that we have to move forward in these resource constrained.
Jim [47:25]
I love hearing about that because our system encourages people to claim that they’re going to single-handedly solve a gigantic social problem. But if you really want to see the systems change, it’s going to be hundreds of organizations, it’s going to be governments and a society moving in these other directions. And we’re going to go in that journey, it’s going to be together.
And I think CHIC is a great example, right? They’re working on not just the technology, the technology was the community health toolkit or something is a key part of that, that Medic, another Skoll awardee helped pioneer. But it’s not just the tech piece, it’s also all these other systems. In their case, it’s getting community health workers, getting the tools and the recognition and the pay that they deserve. But I think you’re right. And I was talking to a tech company foundation this week, it said, well, what if you had $100 billion? And one of the things I said is I would invest more in some of this common infrastructure that we all need so that we’re not recreating the things that we all share, whether that’s stripping out personal information, which the state of the tools are kind of weak and we should be doing more of that, or minority languages, under-resourced languages, and the list goes on and on. So I think that’s great advice.
Well, Debbie, this has been a really exciting conversation. Thanks for investing the time. Absolutely. I wish you a lot of luck in continuing to figure it out, even though you kind of have a tracker or a couple of… At great scale, and you are still trying to figure it out. All right, thank you very much.
Debbie [49:06]
Thank you.
—
A huge thanks to Debbie Rogers for sharing how Reach Digital Health is using mobile technology to provide answers, hope, and safety to millions of mothers.
We are now approaching the end of season three. In a couple of weeks, we’ll close the season as we normally do with some personal thoughts from me about the incredible innovators we’ve heard from during the season.
If you enjoyed this episode, please follow, rate, and review the podcast on your favorite platform and share it with someone who cares about maternal health, digital equity, or scaling up tech for good.
If this conversation resonated with you, I’d love for you to check out my new book, Technology for Good, How Nonprofit Leaders Are Using Software and Data to Solve Our Most Pressing Social Problems, from MIT Press. You can find links and more details at fructurman.org.
I also want to acknowledge the generous donors who support Tech Matters the Organization and Tech Matters the Podcast, especially Okta for Good.
I’m your host, Jim Fruchterman. Thanks for listening.