Establishing durable connection between readers and authors of a smart newsfeed

I’m a product designer in Yandex.Zen, one of the most popular Russian digital media and blogging platform. Zen is a personal newsfeed. Smart algorithm collects articles from thousands of bloggers and media. Then it forms an individual feed for each reader according to what he or she has already taken interest in. The more you read Zen the smarter it becomes.

This is how the feed looks like. Each card represents an article. After reading one the feed suggests related publications.

Yandex.Zen product development is mostly data driven. One of our key metrics is 7-day retention. Besides there are plenty of other metrics to follow: who, when, how much clicks, how long people spent reading an article, when and how often they come back etc. It is vitally important to remember that behind those numbers there always are real people with their hopes and fears, with their people’s problems.

When our business performance metrics stop to grow or stop to grow as fast as we expect them to, we start following related product metrics even more carefully. When we see sufficient deviations we start digging deeper. What happens? What exactly people’s problems stand behind these analytics graphs changes?

A year after a fabulous launch Yandex.Zen started to lose its growth rate. People came to enjoy the product just once and never came back. This metrics behaviour made us ask ourselves what people’s problem can cause it. We were to answer a ‘why’ question.

At this point the most important empathic part of research and development process starts. We talk to people, watch them using our product and try to understand what prevents them from enjoying it more.

While analysing findings from user interviews, we identified the following most common reasons to return to the application:

• content from favourite authors
• discussion of content

Close and durable connection and interaction with authors makes the app usage experience valuable enough to come back for it.

Having the cause found we started to generate hypotheses to eliminate it. Then we were to validate them according to whether they are truly useful. We had several parameters which we validated each solution according to:

• Does the solution solve the problem in the moment?
• What measurable impact will it have?
• Won’t this solutions spoil other scenarios in the moment and in the long run?
• What are the time and cost development estimations?
• What impact will this solution have on the whole product in the long run?

We started brainstorming desperately and came up with most wild ideas. We suggested that comments, chats and stickers could arouse interest to come back and continue the discussion.

Hypotheses: comments, stickers on articles, chat and badges for reading in readers profile.

We rejected those hypotheses because each of them was more or less of a certain niche. Extrovert like to chat both in real life and on the internet, but introverts are completely indifferent to this kind of activity. Besides our engineers gave enormous development time estimations of all these fabulous ideas.

In order to establish a durable connection between author and reader we decided to provide an opportunity to subscribe to any blogger or channel in Zen. From now on a user recevies an opportunity to subscribe to blogger, channel or media inside Zen. From now on people can see publications from their favourite channels in the feed more often and they can check the new articles of their favourites within a click.

At this point we agreed on the metrics (and its numerical growth) to show whether the problem is solved. It makes sense to pick several metrics both in a short and long term run. In this case the main metric was the number of returning users.

Among the interaction solutions there was a burger menu, a horizontally scrolled gallery of channels and a tap bar.

Burger menu, horizontal galley and a tap bar.

By that time the burger menu was already used in Zen to keep profile settings and it was not popular at all. It would be short-sighted to reuse it for more important functionality. A horizontal gallery promised to be lovey, but in case of too many subscriptions it would be boring to scroll. Besides it had no room for any discovery scenario or search opportunity. Regarding all this we picked the tap bar solution.

This is the first visual design version of main product pages: the subscription list, the search and the channel page.

When product hypothesis is validated within the team it’s time to build a prototype and start showing it to people. At this point we not only verify whether the solution is useful, but whether it is usable as well. If we have enough time there may be several iterations of research.

First prototype used for research.

The research proved the prototype to be usable. Though it became obvious that some more interaction animations are needed to make all the processes smother and more obvious.

I also went through several iterations on color scheme and icons. The blacks were substituted by the whites to keep the whole application light. All caps were reduced to drive accents from text to icons.

Main steps towards pixel perfect tap bar.

Icons were redrawn by my talented colleague Misha Anikin to empathise the geometry and create an elegant rhyme with interface elements. The check symbol shows the user is subscribed to the channel. And the same symbol is used to mark the subscriptions list in the tap bar.

The early version (left) and the final version (right) of the tap bar. The accent color in the tab bar was changed to red for two purposes: to keep consistency with Yandex guidelines and to visually separate the tap bar from the whole page.

As soon as the solution is approved by the team and our readers it’s time to ship it to our beloved audience.

The subscriptions tab, channel search with suggest that were shipped as the first version of subscriptions back in 2017.

The introduction of subscriptions made Zen even more personal and we could see it as our key metrics — 7-day retention — started to grow. It showed the growth of 8% in a month period.

The prototype of subscriptions functionality.

Subscriptions made it possible to establish durable connection between readers and authors. Now when people received a guarantee of they favourite content in their feeds, it appears a real reason to come back.

These are the main steps we took towards this result:

1. Mention the metrics deviation
2. Research what people’s problem stands behind it
3. Generate hypotheses on solutions
4. Verify each hypothesis and pick one
5. Prototype
6. Verify prototype
7. Iterate for pixel-perfection
8. Production
9. Follow the metrics