The Making of YouTube Go

How reimagining an app’s experience led to five key lessons for building international products

When our YouTube research team met with Indian users in 2014, we heard a common complaint: YouTube’s data usage was mysterious. In interviews, people described how they could buy a 100MB data pack, stream a video, and unknowingly deplete the entire thing in one sitting. Similar stories poured in from countries like Brazil and Nigeria. One 20-year-old aspiring entrepreneur from Lagos said he avoided opening the YouTube app altogether. When we asked him why, he told us that “data just flies away.” Across India and Indonesia, YouTube users experienced similar frustrations. “That data is burnt,” said a 24-year-old IT specialist from Jaipur after streaming a music video.

We knew there were opportunities to improve YouTube and make data usage more transparent. And, we learned that adding or subtracting features from the existing app would not be enough—we needed to reimagine the app’s experience from scratch. This is the story of YouTube Go, a two-year journey that resulted in a video streaming app specifically for emerging market users. While developing the app in the field, we learned five key lessons for building products for global audiences. Though YouTube Go’s features may not apply directly to your product, these takeaways can still help guide your team’s user research and product development approach.

A research participant in India demonstrates his video streaming habits.

Retrofitting may not always work

User feedback was essential in YouTube Go’s early design phase. We wanted to make video consumption more affordable, so our first concepts featured lower-quality videos. Our research team tested a design displaying low-resolution key frames overlayed with an audio track. Users could view the core information—audio and key frames describing the video—without having to spend as much data.

Just because I have less data, does not mean I want a bad experience—video is video. 

— A research participant and mother of two from Gurgaon, India

Unsurprisingly, participants reacted strongly against this concept. Users had downloaded high-quality videos before, and naturally didn't want to settle for a lower-quality product. Inspired by these conversations, we pivoted our design and explored options for delivering an improved experience in spite of data and connectivity constraints.

Respect people’s budgets

According to a recent study, ninety-six percent of people in India use prepaid data plans. Data is expensive relative to average income, and prepaid plans help people manage the cost. YouTube’s research participants described buying data from a local kiosk every day and carefully planning how to spend it. Many people said they avoid data-heavy activities like streaming videos, which can consume the daily allotment of data in less than an hour. As we continued our research, we watched participants search, start a video, quickly realize it was not what they wanted, and click on another result. Some people peeked at the video without watching it entirely.

Our team learned that viewers need more than just a thumbnail to decide whether a video is worth a data investment of 200MB. To solve for this we developed “video preview,” a new feature that allows users to glance at the key frames of a video and  get a better sense of what the video is about, before deciding if they want to spend data to stream or download it. The preview modal also includes the size of the file, allowing users to calculate the exact cost of the video. For those users buying small amounts of data every day, this awareness can help them maximize the media they download with their data.


Video preview helps users decide if they want to watch the content and how much data it will cost. In this example, a user can choose to save money by spending 1.9MB for a low quality video, or spend 12.6 MB on standard quality content.

Help users connect with each other

Throughout India, Nigeria, and Indonesia, we noticed that participants often searched their friend’s video gallery rather than browsing the internet, because it offered a curated list of options. While YouTube suggested content, users often ignored it in favor of recommendations from people they already knew. Finding information from local resources and recommending content in social settings is so common that our team coined the term “Human Information Network” (HIN). In this network, people rely exclusively on each other for directions, recommendations, and news. When we began building YouTube Go, we realized we needed to leverage this network in order for the app to be relevant to our users.


To ease the boredom of long commutes, strangers in Mumbai create hotspots to connect their phones and exchange downloaded videos and songs with one another.

In HINs, friends not only recommend videos but share them as well. Despite unreliable networks, interview participants told us they were already finding ways to share their favorite videos without investing data. Participants showed us how they downloaded videos when they had Wi-Fi then shared the media offline through exchanging SD cards or sideloading: a process of transferring files between two nearby devices.

Our YouTube team explored ways to extend this HIN and developed a feature called “nearby sharing.” This feature connects people to their real-life social networks and enables them to wirelessly transfer offline videos to people nearby, using a minimal amount of data (15KB for a majority of transactions). For example, after downloading a video, users can open the sharing menu to see who’s in the vicinity, then tap a name to send the video by creating a local and secure Wi-Fi network. Nearby sharing also recognizes the cost of data and helps users conserve it, by allowing them to receive videos from friends at a fraction of the data cost of the entire video. By learning directly from participants and seeing how they share information, news, and internet content, YouTube Go expanded the amount of available content in the offline sharing ecosystem. 


With the “nearby sharing” feature, users can send saved videos to friends who are close by, without using the internet.

Offer offline experiences

When users tried to watch videos on YouTube over spotty connections, the app perceived that they were offline and sent an error message to try again later. We knew this wasn't a good experience, and we wanted to ensure that people could use the app regardless of their network.

YouTube Go doesn’t treat being offline as an error state. We worked to develop a feature that enables users to watch video previews offline, by compressing and caching the metadata for all the videos on their YouTube Go homepage, plus an additional set they can pull to browse even when offline. When browsing offline, users can also add videos to a queue for download—and these become active downloads when the user next comes online. Rather than prohibiting all actions offline, we enable users to accomplish more without an internet connection.

Feature locally relevant content

In conversations across India, Nigeria, and Indonesia we found that YouTube content didn’t always feel relevant to users. In addition to international content, users wanted to watch videos that were popular in their region, but couldn’t easily find these options on the homepage. We needed to do a better job of incorporating cultural norms and expectations into YouTube Go, so we designed an experience that tailors video recommendations to user preferences, in people’s own language. We also created a home screen that features trending and popular videos in the user’s region. By glancing at the home screen, people can find and discover videos that are relevant to themselves and their community.

The creation of YouTube Go was inspired by conversations with the people who use our products, and a better understanding of their day-to-day lives. The app’s initial success is inspiring, and demonstrates the need for more global products that support data saving and content discovery in locally relevant ways. Now that we’re launching in over 130 countries, we’re working to meet our users wherever they are, and we encourage you to do the same.

User Research Emerging Markets