Introduction:
It is well known that the notification system plays a vital role in enhancing user interaction and engagement rate with products (and other users as well).
If I want to deal with this project very quickly, I must point out that "Aparat" did not utilize a proper notification system. despite the high number of its active users; almost 70 million active users and +4 million daily visits! Notifications were sent only when users followed the channel or when someone got a video warning from our operators; such as “Your video has been removed!” without any reason and so brief!
Define Problem:
The problem started when a product with this volume of users in the field of VOD, definitely needed a clear and reasonable notification system that most importantly meets the needs of its users. The feeling of the need to change was felt from the very beginning, a change that will cause to cover the likes, comments, new followers and operator messages with detail.
Research:
We asked our "Support department" to share the most repetitive feedback from users about their problems with us to find some useful messages that could help us design the new notification system.
After reviewing all the related messages, a few important things were identified:
The high number of messages and contacts with the support department to find out
why their videos are blocked, warned, removed and etc. They looked for a reason! that we weren't sending it to them, In fact, that was one of the Notification system's roles that we did not take advantage of it.
let me point out some of the users’ questions:
- What caused my video to be reported?
- How did my video get deleted?
- Why did I receive the video warning?
- Why did I lose access to my account?
- Why am I blocked?
- And ...
(At least 20% of calls received by our "Support Department" were related to the above factors, sorted into a "Bad notify " category)
And some other related messages about the useless notification system in Aparat:
- I can’t use it
- I want to get notified when a channel posted a video or starts a live stream.
- My audience will not be notified when I upload a video.
- ...
Due to this research and data analysis, showed the low rate of interaction with videos (likes and comments) and the low rate of reciprocal followers, which caused the project to start.
Beginning & Ideate:
The project has started! Using the previous data as well as data from the support team and "analysis" teams... I started to classify the types of notifications.
- Notifications that create more engagement and interaction of users with each other's channels and videos.
- Notifications that reduce support calls.
- Notifications that are critical and should be notified to users.
Before explaining these categories, let me mention that we have launched a new service for streaming gamers at Aparat! Which was well received and increased the complexities of the notification system! And different categories of notifications were then created! Like: new subscription, donate and etc.
So the users who received the notification were divided into 2 types:
- Channel owners
- Live streamers
Each of these users had different behaviours with the service and formed different personalities. This duality created different conditions in providing notifications to individuals based on their behaviours.
Following my research on the types of users and their behaviour, I went to work and specified the types of notifications that we should send to users:
Notifications that increase engagement and interaction with each other's channels and videos:
Like, Comment, Follow Channel, New Video From Friends, Start Playing Live Friends
Notifications that reduce support calls:
Operator messages; report video, block account or warning with reasons and solutions!
Notifications that are vital and should be communicated to users:
status messages, rules updates
Streamer-specific notifications:
New subscription, new donate
Each type of notification had a set of needs that I identified after research and analysis. Once everything was identified, I placed them into a table and went to Design ...
Design:
In the table above, I specified the characteristics of each type of notification and saw them as a component that can have multiple variants, such as Thumbnail, CTA, Time, and Avatar User.
The initial sketches started and after determining the structure, I designed the wireframe. We applied several changes to it after carefully examining it, resulting in a version that met all the requirements of the "Aparat Notification" system.
Finally, I turned the raw design into the final design in the software, and the output was something like the image below. I provided this image and a document explaining how the notification system should work to the development team.
Launch & Test:
After a while, the development was completed and the first version of the system was released.
After two or three days, several different feedbacks were sent to us, and apparently, a number of content producers were offended by part of the system's performance. By reading the messages, we realized that Like notifications do not have a pleasant function for some Aparat users who use this service differently!
These users were members of the republishing system and uploaded more than 80 videos per day on their channel, In this case, if each video received 5 likes per day, there would be only 400 Like notifications that day. Although there were not many of these people, we decided to solve this problem so our notification system would work better.
Refinement:
Therefore, we created a feature to reduce the number of notifications sent in some cases while still keeping the user updated on the number of likes they receive; That feature was "grouping".
In this way, instead of sending a notification for each Liked video, a group item is created for that video, which is updated every time the video is Liked and when clicking on this group item, it goes to its internal page, where displays individual users' likes.
As such, if 10 videos receive 100 likes each, the notification centre will have only 10 unread items.
This drastically reduced the number of unread notifications of users and the percentage of user satisfaction after this notification update experienced a significant increase.
Moreover, we implemented this feature on Operator Message notifications after reviewing the positive impact of this update.
The service was also enhanced by creating a Setting section that can be used by different types of users and can be customized for their specific needs. So Channel Owners and Live Streamers can easily personalise their notifications.
Result:
After the project launched, I looked at its data (on Yandex) and Support Department reports to evaluate how the changes to the notification system affected our previous data.
Fortunately, this time data indicated a specific decrease in total calls received volume by Support Department by 18% and a certain amount reduction in the “Bad notify” calls category (almost 43%).