[THEME MUSIC] JOSH WOODWARD: All right. Hi, everyone. Thank you for coming. My name is Josh, and I'm a product manager on the Next Billion Users team at Google. And we're really glad you're here. We want to basically use this session to tell you a little bit about what we've been learning, building for the next billion in places like India, Indonesia, Brazil, and other countries around the world. Before we get started, just maybe a show of hands, who's flown in from India for I/O? Oh, awesome. I know it's very late right now for you all, so we'll try to keep you up. Anybody from Indonesia who's here? OK, a couple, cool. Brazil? Great. And I'm sure there's a lot of other countries represented too. So thank you all for coming. What I wanted to do just quickly by way of intro, before we interview some of the folks on stage, is tell you a little bit about how this initiative and this focus area for Google got started. Since the beginning, Google has always been interested in building products for everyone. And what we realized, looking around the world, is literally a billion new people come online for the first time. That changes how we think about some of the ways we develop products. And really, this session, we want to share some of those lessons with you. Some of the things we've got right, some of the things we got wrong, too. One thing that's changed is how we think about our existing products. If you think about YouTube, or Search, or Maps, or Chrome, or any of the others. What we found is a whole new crop of user behaviors that are becoming really common in [INAUDIBLE]. For example, today in India, 28% of the searches we see are voice searches, which is amazing. And it's growing, voice is growing everywhere around the world. Brings up really interesting questions about how you interact with computing devices, localization, and others. We've also seen in Google Maps, for example, our new Two Wheeler mode feature. Really popular, and spreading to other markets too. There's new modalities we're having to solve for transportation directions. That's a little bit about how we've started thinking about our existing products. But obviously, as smartphone prices drop around the world, as connectivity becomes more affordable, more people come online, and it opens up brand new use cases and problems to solve. And that's really what our three panelists up here have been doing the last few years. So they're going to talk to you a little bit about some of the products we've launched over the last six to eight months, and a lot of what's kind of gone into them, both from the engineering side as well as the design and research side. So what I'll do is just introduce them briefly, and then we'll get into some of the questions. And at the very end, we've got two mics in the room. If people want to come up, we'll save some time at the end, you can ask a few questions. So first, just quick intros down the row. I want to introduce [? Pankesh. ?] [? Pankesh ?] leads engineering for Google Tez. Tez is our payments app in India that we launched back in September. And there's a whole lot of stories he can share about that in just a few minutes. Up next, we have [? Davesh. ?] [? Davesh ?] leads our Files Go engineering team. Files Go is an app we launched late last year, that helps users free up space on their phone, find files really fast, and also share files offline when you're nearby people. So some very interesting engineering that went into that product. And then finally, we have [? Nithya. ?] [? Nithya ?] is on our research and design team. And she's worked on I think, every product at this point in some form or fashion. She's going to share a lot of lessons we've learned doing research, both in the field, but also we have a lot of local teams based around the world that build a lot of these products for us. Our kind of group, the Next Billion Users group, really thinks about how do we staff people in these markets so we can really watch these trends as they're emerging. We also give a lot of these small teams a lot of latitude to go after big repeating problems. We're an experimental group trying out a lot of new things. And of course, the three of them will talk about some of the products we've launched. Without further ado, let's start with [? Pankesh. ?] [? Pankesh, ?] maybe just give us an overview of what Tez is, and we'll start there. [? PANKESH: ?] Yes, sure. Tez is mobile payments app for India. And the word Tez itself means fast in Hindi and some other regional languages. It's built on top of a new inter-banking protocol. It's a real-time protocol that was introduced in India called up UPI. That stands for Universal Payments Interface. And almost all Indian banks have implemented this. It's 24/7, instant bank-to-bank money transfer protocol. And what Tez does is it puts a brand-new, very fresh user experience on a variety of payment use cases, on top of this protocol. So a Tez user links their bank account with their Google account, and UPI gives them a payment address, which is called a virtual payment address. Per the protocol, this payment address is inter-operable, so any service or product that is using UPI has to support this destination addressing mechanism. And the neat thing is that any payments that are done from this address or into this address are done from the bank account, or into the bank account directly. It's different than a digital wallet. And you don't have to load up a digital wallet first from your bank account, and then transact from that wallet. The other neat thing is that the council, which maintains and runs this protocol, has made these transactions free of cost. There's a big push by the government towards digitization of payments. And so effectively, what you have is instant, 24/7, free, kind of like a wiring ability. Literally at your fingertips on your smartphone. Why don't I show a demo of some of these features? OK. Well, we were having some Wi-Fi issues earlier, as they usually do with the demo. I have two phones here, if I could get the volt vision. OK. So this volt vision one is my phone, and I'm going to get into Tez. It uses the OS lock, so we will get here. I'll do the same on-- sorry about the orientation-- we will do this also on the phone on the left here. I will launch the Tez app-- again, use the OS lock to open it. And now, we have these two phones. And what you see right up front is a circular button. And let me explain this. We call these cash more payments. So imagine that a user has this phone, and they walk into a shop. And there's a merchant behind the counter who has their own phone. And so basically, two phones are nearby-- and one of my phones is working on India international roaming, so I hope that it works. But basically-- let's put this into the receive mode, and let's put it into the pay mode, and see if the phone discovers the other phone. So what should happen-- if WiFi was working-- was that one phone would be discovered by the other phone. Let's see. Boom. So as you can see, the phone on the left has discovered the phone on the right, which is my phone. So the user can tap to pay. And let's say I'm buying some chai, which costs a lot these days. I can put in a message, I can proceed to pay. UPI mandates that every transaction is protected by a PIN. And for purposes of this demo it will go through the bank's-- the destination and the receiver bank, and voila. And we're trying something out when you get this. So that's how these nearby payments work. What you can also see now, just focusing on the Chromecast, the phone on the left. You see these chat heads. And essentially, these are profile pictures of the payment peers. I can tap on someone-- let's say, this person-- and it will give me the entire history of the payment transactions that have happened between me and this person. So that's really convenient. Now if you go down further, you see a section called Businesses. And that has providers, like my phone provider, my internet provider. And we recently integrated bill payments on top of UPI as well. So every month, I get a notification that hey, your bill is due. I tap it and it's done in an instant. Likewise, you can check balance and other such things as well. So that's kind of like a quick overview of the app. JOSH WOODWARD: Nice. [? Pankesh, ?] can you tell everyone a little bit more about the cash mode, and how you guys built that? A little bit of the back story there? [? PANKESH: ?] So it started with some user insights. And the key user insight comes from the fact that paying by cash is still quite widespread in India and other [? MBU ?] markets. And you know if you think about it, paying by cash has its benefits. It's like pseudo private, anonymous, and it's just plain convenient. So when we wanted to do a digital mode of payment, we want it to have an experience which is kind of like cash in many senses. That's why we call it cash mode. And how it works is-- actually, there are various ways to make these nearby or proximity payments work. And like in the US, and various parts of the world, generally NFC has been the way to go. But phones in India, most of them don't have NFC. So now you can use Wi-Fi, you can use Bluetooth for pairing. Essentially, you're trying to make two phones who have never seen each other ever, don't want to exchange explicitly some addressing information, to be connected to each other. So what we do is we transmit information for pairing on audio. Actually using ultrasound. Let me go into a little bit more detail. What happens is, one phone sends a short ID. It broadcasts a short ID, identifying itself saying hey, this is me, this is me. And this idea is broadcast using what's called DSSS, or Direct Sequence Spread Spectrum. And it's a spread spectrum technique. What it is doing is it's multiplying the data signal by pseudo random noise spreading signal. And so what actually happens is that on audio, on an inaudible frequency-- nobody can hear it from a user experience point of view, that's very important-- what's been transported is white noise because of this pseudo random noise spreading. And it's very important for security because then a snooper can get access to the signal, but not make sense out of it. However, the receiver can use the same-- and multiply with the signal that it gets-- with the same pseudo random noise code, and recover the signal. So the receiver recovers the ID, validates the ID, and then does a sober look up to figure it out, OK, who does this ID belong to? And the server sends back more information, which is relayed to the receiving phone. Just like the other phone got, OK, you're paying to [? Pankesh, ?] and so tap. That's the user confirmation. So that's how it works. I think it's been great for us as you can imagine, that there's lots of challenges in making this work in terms of tuning and configurations. There's receiver configuration, broadcaster configuration, there's various audio settings. To kind of give you some statistics about this, when we launched-- and Tez launched in India in September last year-- we tested about 200 devices and hand-tuned them in our lab. And since then, we have evolved this process of tuning, and now it's a control system, like feedback loop, in which we just adjust automatically the parameters in the field. And observe how that goes. And to date, we have nearly 6,000 unique devices who have made a successful cash mode transaction, of which manually, we only configured 200. So I think it was extremely important to us that we get the coverage that we eventually ended up getting. And that's why we chose audio, over say, NFC. JOSH WOODWARD: Interesting, nice. Maybe one last question for you. Briefly, can you share a little bit about what you've learned working with a new app, a new protocol, and UPI, a lot of external partners? And I think there's a war story here, maybe? PANKESH: This protocol is very new, and a lot of the banks-- and we have the largest banks in India as our partners, and they were implementing it new as well. So we literally broke the bank when we launched, with such a high QPS. But you know, when you're building any distributed system, whether it's your components or external components, then you've got some health monitoring and various other kind of high availability things. So what we do is a couple of things, just to give a quick example each. User account creation time, when this address has been created-- if one bank is down, then we will go to the other bank. At the payment time, however, you're transferring from one Bank A to Bank B. And what's important is to not have the user in what's called a stuck state. For that what we do is we build ML predictive models to figure out whether any of the banks-- there's actually five parties in a transaction, the way the protocol defines it-- and any of them, if it is down, or almost done, then we would rather bring it up with the user, and say, hey, please try again. Rather than having the risk of the transaction kind of entering in a little bit of a stuck state, where nobody knows where the money has gone. JOSH WOODWARD: Nice. Great. Thank you for sharing. [? Davesh, ?] I want to shift gears a little bit and talk about Files Go. And maybe you can tell us just briefly what it is. I think you'll want to show us about it too. [? DAVESH: ?] Files Go is a file and storage management app built for [? MBU ?] markets for mobile first users. It also supports peer-to-peer sharing using nearby technology. It doesn't use the internet. The way it all started, we started looking at what the users in [? MBU ?] markets needed. And we realized first off that mobile data in these markets were very expensive. And what that meant was, users in these markets were not able to consume content on their phones, which was a primary or only device for them. Most of the time because of the expense. That's where the idea for nearby sharing was born. Then as we were researching more, we realized that once users got access to more content which they could get from their friends, a lot of these phones have very little internal storage. So users started running out of storage when they got content from their friends. Because of that, we started looking at storage management. The other problem that we saw was users were not able to find the content on their phone easily, because most of the existing file managers out there show a folder [INAUDIBLE] view, which is complicated for new users who are coming to the mobile first time, have never used a laptop or desktop computer before. So we decided to build something that would be easy for them to use, would allow them to keep their phones clean, and allow them to get content from their friends. JOSH WOODWARD: Cool. You want to show it? [? DAVESH: ?] Let me show it. JOSH WOODWARD: One thing while he's pulling up the demo-- I remember when this team came back from field research. They spent a lot of time on college campuses across India. And we found a stat which was shocking-- one out of three smartphone users in India run out of storage, or see a low storage warning, every day. So if you think about that, something like 80 million people are constantly up against the boundary of running out of storage. Anyway, we'll switch over to the Wolf Vision and you can show how it works. [? DAVESH: ?] Is the phone showing up? JOSH WOODWARD: Yup. [? DAVESH: ?] So this is a Files Go app. First stop, right on top, we show the storage available on the phone and what's the total storage. The cards that you see here are personalized additions for the users on what they can clean up from the phone, to free up storage. Let's go and take a look at the large files card. These are all the large files on the phone. Typically these are videos that users shoot, or the kind they get from their friends, like movies and stuff. So with one click delete, you can just get it off this user card back to 180 MB of data. Let's take a look at files. So one thing that we saw was users had a tough time trying to find the content on their phone. We came up with the scheme of categorizing the content based on how it was acquired or how it was associated in users' minds. For example, images. We don't show a folder view. You can see all images here, and then the other tabs show where the users think the images came from. Similarly for videos. Again, it's categorized in groups that users can identify with. Right down here, you see two buttons for send and receive. This is where users can initiate the nearby sharing mode to transfer content to or from their friends. Interesting thing-- just like [? Pankesh ?] talked about, doing a nearby connection sounds simplistic enough, but there's a bunch of stuff involved there. Because there's a wide variety of phones out there in the market. We use a combination of Bluetooth, Bluetooth Low Energy, Wi-Fi direct, Wi-Fi hotspot type of technologies. And based on the two phones, we are able to connect on Bluetooth and negotiate their capabilities, and create the fastest connection possible. For example, if the two phones both support 5 GHz band, then we can connect on 5 GHz, which will result in a file transfer rate of a GB around 40 seconds, which is pretty impressive. Which wouldn't have happened if we'd gone on the path of just using Wi-Fi, which is sort of the default path. The other thing that I wanted to talk about was the low-resolution and memes. So one thing that we noticed in our user research again was that users in these markets used chat apps quite a bit, and they receive a ton of content from their friends. Most of which is good morning messages, memes, jokes, which they don't want to really keep because it fills up their phone. At the same time, they also receive content from their friends which they want to keep, which might be pictures of each other that they took on a trip together. Which meant users were going through going through all these chat messages and deleting stuff manually by figuring out what was useful and what was not. We decided to try to tackle this, and we came up with this card that-- internally we use Google Vision API, Vision Library, to do text detection. And then add heuristics to that to detect memes or images that are not useful for the user which they view only once. And these are some of the things that get detected. And the user can use one click again to delete it, and they can keep the rest of their useful messages in tact without having to go through it manually. JOSH WOODWARD: Cool. One thing also-- just to call out from this that the team has seeing-- is that the average user, the first time they use the app, is freeing up over a gig of storage. And so that little bubble boy is dancing a lot when storage space is freeing up. It's become a really interesting daily and weekly habit people have gone into. Maybe one last question for you, [? Davesh, ?] before we move on [? to Nithya. ?] Can you tell us a little bit about what you feel like you've learned so far? This app has only been out about four months. It launched at the end of last year. [? DAVESH: ?] There's a bunch of stuff we learned. The most interesting for me was the fact that this app is getting used a whole lot in US and Europe, which we did not expect when we launched it. I recently read a book by Vijay Govindarajan called "Reverse Innovation," which talks a lot about building products specifically for these markets, which you can then bring back and [INAUDIBLE] other markets which you did not build the product for. And we are seeing something similar for Files Go, which is very fulfilling. JOSH WOODWARD: Nice. I guess, Nithya, you can take us home here with the last set of questions, before we open it up to everyone. Talk to us a little bit about some of the thematic work you've done, around connectivity and access with your sort of research and design experience. [? NITHYA: ?] So as part of Google's mission of bringing the benefits of the internet to everyone, there are two main products. There's Google Station, which provides high quality internet access in the form of public Wi-Fi and Datally, which is a mobile application, which helps users get more value out of their data. So I'll talk about Station first. Station was launched in 2016, and is now in India, Indonesia, and Mexico, in hundreds of hotspots like train stations, parks, malls, and public venues. Thousands of users, people come online for the first time on the station network. Datally really helps make the existing constrained internet more manageable and accessible. So it provides data transparency, so that users can understand where their MBs and GBs are going in their browsing activities. It provides data saving functionality, which helps users get more value, extend their data packs. And it also alerts users to public Wi-Fi hotspots when they're outside, so users can be online more often. And the underlying insight for Datally is that many people around the world treat mobile data as money. Because it is expensive, slow, and limited. And as a result, users resort to practices like turning off mobile data when they're not using it to cut costs. Or hesitate to get new applications, or have fears around using existing applications. So that limits their participation online. And Datally provides functionality to help users ease into their experience of getting more value from the internet. JOSH WOODWARD: Can you talk a little bit about what we've learned with these products, now that they've been out, and sort of the environment that they've landed in? [? PANKESH: ?] I can go on for long about this, but to simplify things-- there's people, context, and devices. So with regards to people, as internet access is growing all over the world, technology is touching new societies. And a lot of the people that are coming online are increasingly diverse. So there's diverse literacies, languages, aspirations, income levels, professions, geographic spread. And many of the assumptions that we've made around the first few billion users need rethinking. So for example, it may not hold true that users are well educated, or relatively wealthy, or English-speaking, even. Many people around the next billion users are not necessarily English-speaking primarily, but may prefer to use phone UIs in English, because English is seen as a language of upward mobility and aspiration. So in terms of design, how might we design interfaces that help users with the use of simplified English and more visually rich interfaces. With regards to context, like I mentioned, internet is often intermittent and constrained. Roughly only 43% of the world has access to LTE. So users are often on 2G and 3G, and sometimes even offline. So how do we think about designing for these constrained networks? And treating offline as not an error case, but as a normative use case. We also have to think about economic factors, like purchasing power, and, like [? Pankesh ?] mentioned, the prevalence of cash, and social factors like gender and the role of religion in using technology. And then finally with regards to devices. Today, you can buy an Android phone for $30 to $40. It is very likely to have a small screen, low RAM, low processing power, but it is the first computing device that many people have access to. And many of these devices don't get replaced as often as they might here in the West, because of cultures of frugality and gift giving. So how might we think about hardware and software design for low-end devices? JOSH WOODWARD: Maybe one last question, because I'm sure we've got a range of folks from different backgrounds here. But on sort of wearing your UX hat, if you had to pass along some research and design tips and tricks you've learned along the way over the last few years, what would you share with everyone? [? NITHYA: ?] First I'll alert you to our collection of research and design methods on design.google/nbu. You can find it-- it's part of the Google Design web site. And we regularly share new design and research methods, as well as product stories, in the behind the scenes look. With regards to research and design innovations, I'll talk about it through the lens of a product lifecycle. When we start to build a product, and we're trying to understand the problem space, and we want to create a vision for the product-- we really want a foundation of rich insights and understanding of people and the relationship with technology and the unmet needs. So here we've introduced techniques like intercepts, and the more traditional ethnographic immersion. Intercepts are spot interviews that are done with people in various contexts, such as where they live, where they commute, where they work, where they have fun. And by doing several of these we gather a firm understanding of people in that context and their use of technology. In Station, when we wanted to build a Wi-Fi service in train stations of India, we spent a lot of time hanging out in train stations. Talking to passengers, talking to business owners, riding trains ourselves so we could understand what that experience was like. And shadowing passengers who were embarking on these journeys. And really what we wanted to understand was who is in the station. What are they doing? What is the context of using the Wi-Fi? What is their understanding of Wi-Fi? And these insights led up to design principles, such as the service. The login service to get onto the network has to be really efficient and quick because stations are high-stress environments, with a lot of stimulus. The service has to be trustworthy and secure. We saw in our research that women were hesitant to give out phone numbers to log on to a public Wi-Fi network, so we really ensured that the network was secure. And that because of the heterogeneity of devices in a train station-- this is really the cross-section of the entire country-- the service was browser-based. It works as a captive portal model, and doesn't require the installation of any mobile application. An example of research and design techniques later on in product development is when the product is nearing launch. We've created a technique called Trusted Testers Studies. These are large panels of representative participants who are recruited to consensually make use of the application and provide feedbacks. And we collect qualitative and quantitative data. We ran several of these trusted testers studies for Datally in various parts of the Philippines, where we had hundreds of participants using Datally for three to four months. That gave us a really firm understanding of, what are the core use cases? What's the resident value proposition? What are issues that we're running into that we need to fix, before we launch the product in mock-in? JOSH WOODWARD: That's great. Cool. That's it. Thank you all! - Thank you. - Thanks. [MUSIC PLAYING]