 Hi, everyone. My name's Tal Oppenheimer. And I'll be talking to you about how you can make web experiences that work for everyone around the world. So as we probably know, the world is a pretty big place. Looking at statistics from the end of 2014, we have about 3.2 billion internet users around the world. But when we're actually developing and building our experiences, some of us might have a picture of the world that looks a little bit more like this. And if we're really sort of expanding what we're thinking about, we might throw Europe into the mix, too. But if you actually look at where the internet users are, it's a lot more distributed than that. So this is a map that shows the number of internet users currently online, where darker means more internet users around the world for each country. So you'll notice that while the United States has about 279 million internet users, there's a number of other countries that are also quite prominent. India, for example, at the end of 2014 had about 243 million internet users, so a close runner up to the United States. And what's more interesting is we actually look where the growth is happening. This is a similar map that's looking at users that came online in 2014. And here in the United States, we had about 17 million internet users come online for the first time. But in India, we actually had 30 million internet users coming online for the very first time just in 2014. And to put that into perspective, 30 million, that's about half of the population of the United Kingdom, coming online in India alone in just 2014. And what's more, if we actually look at where we're expecting the growth to continue to happen moving forward, this is the number of people who are not yet online by country. So you'll notice that here the US is less prominent. It's a little bit more saturated. We still see that about 42 million users may be coming online. But this is compared to India that we're expecting about 1 billion and 24 million internet users who are not yet online to be able to potentially come online. So in terms of where we're expecting growth to be able to happen for new internet users, the United States isn't necessarily the main focus here. There's a lot of people that will be coming online in very different areas. So sure, those are some big numbers. But what does this actually mean? And what might this mean for your experience that you're building? So I just want to look at the e-commerce in India and some of the growth that we're seeing here. So a lot of users coming online might not necessarily mean that this has a big impact for you. But this is just a case study of e-commerce. So what we've seen in India is that e-commerce over the last years has really skyrocketed and is continuing to grow. So these are some headlines from a number of e-commerce players in the India market today. But just in terms of projections, there's actually expected that by 2020 the e-commerce market in India will be a $100 billion market. So we're seeing really huge growth and actually having functional impact on the experiences and how users will be interacting with your site in some of these areas. But there are some challenges. So a lot of us, as we mentioned, we're often sitting developing with fast Wi-Fi connections on our laptops, but the users that are coming online for the very first time may never have used a laptop. I was actually just in India talking to a number of users about what works for them, what doesn't, how they use the web, how they use their phone, and what some of the challenges, what they want, what they don't want, just to get a better sense of what's going on. And I was sitting down with a 32-year-old woman in Pune. We were sitting in her living room. And I just asked her if she remembers her first time using the internet. And her answer was yes. It was about a year ago. She had borrowed her husband's phone and used the internet for the first time. And this is very different than my first experience with the internet and probably a number of your first experiences with the internet. We have people who are coming online interacting with the web for the very first time on a mobile device, not on a big laptop, not on a Wi-Fi connection. And oftentimes, when we think of mobile devices, we might be thinking about something like this. We might have this in our pocket or something similar. This is the Nexus 6P. It was released in 2015. It's 5.7 inches, so a pretty large device. It has 3 gigabytes of RAM. And storage can vary depending on which version you purchase, from 32 to 128 gigabytes of storage. Now, when I was just in India, we also talked to a number of store owners and to get a sense of what sort of phones do they have, what sort of phones do they sell, and what do they suggest for someone who's coming in to purchase a phone for the first time and connect to the internet for the first time. And this is just one example of a phone that they sell. So this is the Samsung Galaxy J1. It was also released in 2015. It's 4.3 inches, so a little bit smaller. It has 512 megabytes of RAM and 4 gigabytes of storage. So just as a quick comparison here, we have about 1 1⁄6 of the RAM. And best case scenario, if you get the smallest Nexus 6P, we're talking about 1 1⁄8 of the storage space. This is a very different device profile than what you might have in your pocket, what you might use on a daily basis, and what we often end up envisioning in our head when we're actually building these mobile web experiences. The good news is, the web actually solves a number of these problems that some of these different devices can introduce. Installs, as Darren mentioned, are not an issue on the web. You don't have to worry about your actual experience getting downloaded on the device, which can be a real problem when you're talking about some of these devices that have limited storage space. What's more is you don't actually have to worry about the exact APK size and worry about shaving off megabytes on that APK so you don't get blocked because users don't have quite enough space to download your experience. What's even better is once users are interacting with your experience, you don't have to worry about getting it up to date. Because again, with native apps, you often see that installs can be challenging because they also require more storage space and connectivity. So the web does a lot of things to actually solve some of the problems or challenges that these devices can introduce. But there are some other constraints that come into play that I just want to talk a little bit about. The first is connectivity. So a lot of us here are probably connected to Wi-Fi right now. It might be a little bit slow because it is a conference. But in a lot of these areas where users are coming online for the very first time, 2G is incredibly prevalent. And the 2G speeds that we're talking about here are about 60 kilobytes per second, or sometimes ranging up to about in a good condition 450 kilobytes per second. And just to put this in context, globally, 62% of mobile connections are on 2G. And when you're actually looking at the countries where users are coming online for the very first time in India, for example, you see 87% of mobile connections are 2G. So this means that a user's very first interaction with the internet, borrowing their husband's cell phone, buying their own first smartphone, they're going to most likely be connecting on very different network conditions than what we might be using and what we may be testing and experiencing on our own devices. And just to put this in context, you guys may already be familiar with some of these stats. But a one second delay in page time can lead to very concrete impact in the user experience. So a study found that this one second delay can lead to 11% fewer page views on an experience. And this actually also has a qualitative impact on the user's experience, where it led to a 16% decrease in customer satisfaction. And this is a one second delay, which when we're comparing 2G networks to 4G networks or the fast Wi-Fi we might have, the seconds may be an order of magnitude or two larger than one second delays. So making sure that you can really make your experience as fast as possible is very important. And I'll talk a little bit more about that later. The second is even when users are connected, the connectivity can be quite expensive. So in India, it can take 17 hours of minimum wage work to get a 500 megabyte data plan. So just put this in context of web and what we're thinking about here, looking at the average size of a web page. This roughly translates to one hour of minimum wage work in India, loads about 15 web pages. So this is a web page, not a web site, so every single click on the web would count as a separate web page here. So if you take a moment to think of maybe how many web pages you visited this morning, or how many web pages you visited yesterday, do a quick mental calculation of how many hours of minimum wage work you may have been expected to work to be able to afford that. This can become a really large number. I got terrified when I did this myself and would not be able to work enough to pay for my browsing habits. So what this means, both the speed, the connectivity, and the device is that users are interacting with their phone and the internet somewhat differently. The first big consideration is that users are making trade-offs regularly. Choosing to use your experience comes at the cost of choosing to use some other experience, because they might not have a connection later, or they might be out of data for that day or that month. And so as a result, we actually, I was talking to a student in Brazil about how she uses her phone throughout the day. And what we found is that during the day, as the day starts, she actually doesn't browse as often. Instead, she just only uses her phone when it's absolutely critical to message a friend that she's meeting up with or something like that. Later in the day, when she has a sense of how much data she's already used and how much data she has left, she lets herself explore more, use the browser when she gets home, and some of these other experiences, because she knows she has the data and won't be stuck in a position of not being able to do a critical task, be it a search or communicating with her friend, when she actually needs to. Similar to this, because of the trade-offs that people are making constantly, users are also turning their data on and off. And we've seen this one user that I was also talking to in India mention that they actually go into airplane mode about 14 times in one day. This is because in addition to just the intermittent connectivity, the concern over data usage is very real. And there's a hesitation for releasing that control and trusting that the experiences will be data conscientious. And so users are just turning off data so that data is not being used unless they know and need it. And finally, as a result of these two and a number of these other constraints, every extra step that a user has to take on their phone takes time because of the connectivity might be limited or intermittent, and actually takes money because each one of these page navigations takes data, and that correlates directly to hours that they need to work to afford some of these data plans. So this may seem a little bit daunting. There's a big opportunity, but there's a lot of unique constraints. So what can we actually do to make the experience better? The first is removing steps to re-engagement. So Darren talked about some of the new features that we have, and those can have a really big impact. So this is just a simulation here of navigating to a web page on a simulated 2G connection. So what you'll notice here is that I'm navigating to Google.com, though there's a number of other entry points for how I could get to a web page, be it a bookmark or from the new tab page or another experience. And every step that has to happen here on a 2G connection just takes time. So this is actually a faster 2G connection that we're simulating here. There is a broad range. And so I'm specifically trying to navigate to a voice memo app. And the big thing here is that each step, in addition to just taking the time, that this video in total will take about a minute for me to get to my final destination, is also costing money. So I had to do two additional page navigations before I actually got to the experience that I was going for. And this is going to directly relate to users having to make those regular trade-offs. If it's taking extra steps, that's just extra effort for the user that they need to go through that is compounded by some of these constraints that we see. So Darren talked about this a little earlier. But the features we have, such as the add to home screen feature, can really help the user actually get what they care about as quickly as possible. So in addition to letting you re-engage with the user easily by just having this access point, you've also just removed a number of steps for the user to get to your experience in the first place. So you're reducing those trade-offs that they have to make and the extra time and money that they have to spend to just get to your experience. Additionally, with the notifications that Darren also mentioned, and we'll be talking about in more detail later, you can make sure that for the experiences that are most time sensitive or most critical to surface to a user, they don't have to spend that time within your app taking those extra steps as well. You can directly link a user to the experience within your app that is most important and critical to them. And so with notifications, this is an example of Facebook one, you can make sure to deep link and remove not only the steps to your experience, but the steps within that. So you don't aren't asking users to take this extra time and money to get to some of the experience that may be time sensitive. But once users actually get to their experience, you want to make sure that it's as fast as possible and that your data conscientious in what you're actually creating. So Chrome actually has a number of features that are aiming to help already. Chrome Data Saver on Chrome for Mobile is already launched and is a proxy service for HTTP content that can actually save up to 60% of data on average. And what we see is that users around the world are actually using this and enabling it and showing that they are more sensitive to data. So globally, Data Saver is enabled for more than 10% of users, and we see much higher percentages in some of these areas where users are coming online for the very first time. What's more is we're actually currently doing a field trial for Data Saver users who have expressed particular concern with saving data to make sure that we're proactively giving them control of where their data is going. So in this particular field trial, we're identifying particularly slow connections and replacing most images on the page with placeholders so that the user can then choose to long press on an image if they want it or click load images to load all the images on the page. What this allows us to do is that on those connections that are particularly problematic and for users who have expressed particular interest in saving data, we're actually giving them back the control so that they can make sure that the experience works for them and they're using the data where and when they need to. We also have a lot of functionality for developers to help with this as well. The page speed module is already available. And with just a single line of code, you can add this in and make sure that you're getting these data saving experiences for both HTTP and HTTPS content on your web page experience. And so what this lets you do is we've run some trials with this. And this does all the transformations for you with functionally no visible impact on the page for the user. And for users who have clients that have WebP enabled, we've actually seen that on average this causes 37% data savings with a single line of code and no user visible change. So this is a huge impact with just adding this to make sure that for all of your users, you're as data sensitive as possible and really creating great experiences. Beyond that, we also have a number of hints, client hints that are available. These are coming out in M46. So we have DPR width and viewport width, which can allow you to optimize your experience depending on the device that the user actually has. So what this lets you do is make sure that for users who are on maybe on smaller devices, you don't need to send that whole large image. You can actually make sure that you're sending just the image and the bytes that are actually necessary for that user. What's more is we're also working on a save data hint that will be coming out in M49. In Chrome, this save data hint is going to map directly to that data saver feature that we have in Chrome. So that if a user has data saver on, you will know that the user is particularly sensitive to data, and you can make sure that your experience works well for them. So just to give a few examples of some of the things you could do here, so for a particularly data sensitive user, you could load only the critical images to make sure that the default experience really works for them and start building that trust that you are aware that they're sensitive to data and you're taking proactive action to make sure that you're building the best experience possible. You could also do things like changing the resolution of the video or removing autoplay on any videos you have, so that the default experience isn't the one that assumes that users are comfortable using data, but rather is optimized for the user that you're seeing. To make sure that they feel that they are in control and that you can gain that trust, that the users know that when they come to your experience, it says data conscientious as possible. We also have a number of features that you can use to help optimize the speed and the experience on your web page, regardless of the connectivity quality that a user's experienced. So Downlink Max, it was coming out in M48 as part of our NetInfo API, allows you to not only see the actual navigation connection type, but see what the maximum expected throughput of that connection type is, so that you know if a user is currently connected to one of those 2G connections, and you can listen to this. So as users change based on where they are or what their connectivity actually looks like, you can make sure that your experience works well. So if you do want to have an experience that assumes a sort of Wi-Fi or very fast bandwidth, you can, but you can also make sure that your experience is really gonna work well for users who may not currently be in that condition, and they can adjust as those situations change. And what can be challenging with a number of these things is we are connected on Wi-Fi. We do have our 3G, 4G data plans, and so it's often hard to simulate and actually see what your experience might be like. With DevTools and Chrome, you can simulate various network conditions. We also have PageSpeedTest, and I know we'll do a deep dive tomorrow on various speed-related issues, but I won't steal too much from that. But with PageSpeedTest, you can actually make sure that your experience is fully optimized and you can see what it looks like on real 2G and 3G connections, on devices actually in a number of these countries. So you can fully simulate, experience it firsthand, see if it's the experience you want, and adjust it if not to make sure that it's one that you think will work for your users, regardless of where they might be. So a lot of these things have been assuming that a user is currently connected. They have a data plan that they're worried about, or they have a slow connection, but sometimes that's just not the case. And if you've made re-engagement really easy, hopefully users can click right into your experience, but what you might want to avoid is them tapping right into your experience and being met with the offline dyno page. Now, I personally love this game as much as the next person and I'm happy to compete for high scores for anyone who wants to later on. But if you're actually trying to engage and create an experience for your users, something more like this might be preferable. So this is the same voice memo app, and when I'm offline, I can fully see all my memos, listen to them, add new ones, and interact with it. And so with Service Worker, which I know we'll be doing a deep dive into, I believe, right after the break, you can do a lot for this experience to actually make sure that when the users are completely offline, they can interact with your experience. And you can also, what's interesting, is make sure that as the user transitions back to online, that is seamless transition, and they can also, any sort of actions that may require you to be online to complete are transitioned over smoothly. So as users are going in and out of connectivity variations, whether it's speed or actually being completely offline, and then back to online, you've created a seamless experience for them across the board. So with all of these things, you've created an experience that's ideally optimized for all devices, not assuming the sort of big devices and the profiles that we might be expecting or used to or have in our pocket. You've made it really easy for your user to re-engage so that you're not adding additional steps that can actually just cost time and money. And when they're actually in your experience, you're furthering this by making sure that any clicks or interactions within your experience are very sensitive to time and the loading speeds that users might be experiencing and the cost that comes with that. And finally, you've also adjusted to make sure that on any type of connection, your experience still works well and can fluctuate and adjust throughout these. And so with all of these things, you can really make an experience that works for everyone, whether it's a user coming online for the very first time or someone who's on that speedy Wi-Fi on that great connection, you've adjusted and built something that works for all of the billions of users that we're hoping and expecting to come online. Thank you very much.