At the beginning of 2017 Twitch’s Commerce team launched Gift Subscriptions, a way for users to purchase Channel Subscription for other viewers. Subscriptions came with perks like ad-free viewing and custom emotes, making them a great way to bring new people into the streamers community, while also supporting that streamer financially.
Gifting became a popular way of supporting creators as it meant you were receiving recognition from both the creator and the viewer you were gifting to. Some of the most interesting user behaviour we saw was from users who would give multiple gifts in succession on a channel. They created exciting moments on the channel, and was also responsible for the majority of revenue for the product. Considering the number of steps users had to go through to do this it was clear we had a big opportunity to improve the experience.
We knew the next step was to make it far easier to give multiple gifts. But what was the right way to do it? How could we make it simple for someone to give not just 5 gifts quickly, but 20, or even 100?
This was a high impact project where success was dependent on optimizing the user experience for customers. Design was going to be key in making sure the project was successful. I had to consider how to showcase the experience without overloading new users with too many options, and find ways of conveying the value gifts would have on the recipients and the streamer.This project was incredibly successful for Twitch and our creators. It’s some of the work I'm most proud of from my tenure at the company, and really helped me in understanding the power that delight and recognition can have in making a product resonate with users.
The first version of subscription gifting let users buy an individual gift for a specific user. This was the simplest way of adding gifting to Twitch, and allowed us to quickly gauge impact before we decided to invest more resources into this product category.
There were two ways of gifting to a user. After subscribing to a channel the ‘Subscribe’ button would switch to ‘Gift a Sub’. Opening would still allow you to change your subscription tier, but now included an option to gift a subscription. Selecting this would take you through to a user search, with initial suggestions pulled from your friends list. You could choose to type in a username, and see live suggestions as you typed.
The second option was to select a specific user from the stream’s live chat. This was a fast, simple way of gifting to unsubscribed users who were currently watching the stream. If you saw someone who was being particularly funny or interesting you could gift to them with just a few clicks.
The launch of subscription gifting was incredibly positive, and we saw a number of interesting uses emerge. Three months after launch we were seeing that 40% of gifters who have bought more than two gifts have bought them within 3 minutes of the previous gift. These clustered gifts accounted for nearly 30% of all gifts given in the month…
We knew that simplification of the existing experience would be key to making multi-user gifting work. We saw that currently around 35% of users got through the gift purchase funnel. Our hypothesis for the low conversion was the friction in the flow. It required multiple clicks to select options, searching for a recipient, and lastly entering payment information. 50% of users were dropping out in the “gift recipient search” step, which we believed was down to choice paralysis. Potential gifters had no help choosing who they should gift to, a list of usernames wasn’t proving compelling in helping them decide who deserved a gift subscription.
In the earliest concepts the experience was a straight-forward expansion of single-user gifting. The same search box UI, now with the option to select multiple users. We knew it would be less than ideal for someone to search for multiple user names so we also proposed smarter lists to replace the friend recommendations. By default users would see a list of unsubscribed viewers who were currently watching the channel. If they wanted to expand out from there they could switch over to a list of unsubscribed followers.
If a user was still unsure of who to choose I proposed a ‘Random’ button that would pick someone for them. This led to the team exploring an ‘Add Everyone’ button, which would select everyone currently in chat who wasn’t subscribed. What started out as a bit of a joke, we quickly realised could be a powerful feature. We were still focused on gifting to 5 or 10 viewers, what if someone genuinely wanted to gift subscriptions to 50 viewers, or more?
In the next round of designs I expanded the placement given to the random gift option, exploring UI options for adjusting the amount chosen. This still felt like the right focus, but the UI presentation wasn’t particularly exciting. It also wasn’t doing a great job of conveying the impact you would have on the channel if you were to give a large quantity of gifts. A slider, or text entry, didn’t translate that excitement very well.
The final design focused completely on mystery gifting, with searching and selecting multiple users removed entirely. I renamed the product to Community Gifting to better align with the value and intended benefits of giving so many gifts. When a user opened the overlay they would see the two gifting options, with brief descriptions and illustrations to clearly indicate the differences between the two offerings.
The bundle layout took inspiration from offer presentations in video games. Games, especially on mobile, did a great job visually representing the increasing value of each option shown in a list. The stacks of gifts helped give a real-world reference point to what you were about to do. Choosing 100 gifts was a lot of money, but translated to a lot of people you were about to make very happy. The bundles also helped remove even more cognitive load from a prospective purchaser. They could see gift amounts and prices without having to enter any values themselves.
The last piece of the puzzle was the post purchase experience. How to spotlight these gifts for the rest of the channel? For single gifting we used a notification similar to the one used when someone subscribed to the channel. My initial plan was to use individual notifications up to 5 gifts, then group them together after every 10 to stop chat from getting overwhelmed with notifications. Ultimately I decided to lean into the overwhelming. As each of these gifts were worth $5 and special to that recipient, I believed each one deserved to be recognised individually. Spending hundreds to ‘break’ the stream by filling the chat with notifications just made the moment even more memorable and exciting.
Community Gifting’s launch day was incredibly strong, and continued to grow from there. On the first day over 75,000 gifts were given, 3x the number of individual gifts. This was far, far beyond our expectations for the product.
We had believed we were on the right path, but hadn’t fully comprehended how Community Gifting would resonate with Twitch’s users. The act of giving a gift, getting recognition on stream, and supporting the streamer were individually valued, now we had launched a product that offered all of them simultaneously. And all of that was far more important than picking who the gift went to. There were people willing to spend hundreds of dollars to have an impact on the community they cared about, and feel a sense of belonging through their actions.
This surprise success was great, but it also showed we had missed the mark on understanding the true impact multi-user gifting could have. It was important for myself and the rest of the team to do a retrospective on the project and understand why the signals we were getting from single-user gifting hadn’t resonated with us as much as they ultimately did with the community. If you don’t take the time to understand why you were successful then you’re only ever going to get lucky.
The biggest takeaway for me was that building products that prioritize clear customer behaviors and needs are going to be the most successful, especially ones where the product’s value can be quickly understood by a user before they use it. Secondarily I learned that someone seeing a product get used is a far more powerful way for them to understand its value than being told about it.