Sorry. This page is not yet translated.

Engagement Part III: Connections and Inventory

Data Science Team

Previous posts in this series introduced the fundamentals of engagement and explored content creation and sharing (production in Figure 1). In this post, we discuss the importance of connecting users and examine the implications of various levels of content inventory.

16 1

CONNECTIONS

Once your product has a healthy population of content creators, the next step is ensuring they connect with their fellow users. The more connections made between your users, the more inventory you can offer. Moreover, these connections are not restricted to individual users. Connections to other entities that create and share content?—?such as celebrities, news pages, groups etc.?—?also increase available inventory.

Product teams should therefore treat users who have a low number of connections as “needy,” and focus on encouraging them to friend other users, follow celebrity and news pages, and join groups. It is also valuable to identify “tipping points” after which users better understand the product and are more likely to retain?—?for example, Facebook works to ensure new users connect with at least seven friends in 10 days. Of course, this will only work if the seven friends are actually producing meaningful content that the consumer cares about.

CONNECTIONS: KEY METRICS TO TRACK

  • Number of friend connections in first 10 days: It is more than likely that users will churn if they are not able to make a minimum number of connections early on.
  • Percentage of friend versus group versus page connections: Different types of connections are valuable for different people at different times. Understanding this in the context of their time spent and number of sessions is useful.
  • Number of engaged connections: Not all connections are equal. Close friend connections are likely much more valuable that a connection to a group that you rarely use any more.
  • Number (percentage) of active connected user pairs: Understanding how many unique active connections there are and how it is growing over time is valuable.

INVENTORY

As each user’s connects to more content-producing friends, celebrities, news sources, etc., the volume of their inventory, or available content, grows with it. Inventory is extremely important because it is directly correlated with the amount of content a user consumes and therefore the amount of time they spend with your product. In an activity feed environment, the total inventory for all users is a product of content produced and connections made, and significantly influences overall consumption (see Figure 1) and ad revenue.

When people only have a few stories that they can see, they usually consume most of them. So, in general, the percentage that is consumed when the inventory is very low is high but the absolute value of the number that is consumed stays low. On the other end of the spectrum, when people have too many stories that they can see, they just don’t have the time to consume all of it. The following sections explore four levels?—?very low, low, high and very high inventory.

18 2

Very low inventory

When a user’s inventory is very low (less than 10 pieces of new content per day), they will likely consume all available content each time they log in. This is most often the case when people first join a product and don’t yet have many connections. As mentioned above, the Product team should consider these users “needy” and focus on suggesting relevant connections and content that would delight them. In this situation, ranking content is useless and, if low inventory continues as a cohort ages, it will be very difficult to retain these users.

Low inventory

When a user’s inventory is low (say less than 50 pieces of new content per day), several issues may be to blame: they may not have enough friends using the product, the product may not be recommending the right connections. Engagement may also suffer if there is too little content that appeals to the user. If such a user is consuming most of the content available to them, consider that a significant opportunity and work to offer them more. Conversely, when a low-inventory user is not consuming all of their available content, that may reflect more on their engagement with the product itself.

High inventory

Generally, a high-inventory (say more than 50 pieces per day) user is connected to more friends, pages and groups. They also tend to consume a greater percentage of their available inventory, have more sessions and spend more time using the product. These users will generally have higher DAU/MAU and will visit the product regularly. Ranking algorithms are invaluable levers for increasing engagement in this group, as these users cannot consume all inventory available to them. It is also important to deeply understand the types of content they prefer and offer the right mix of social, entertainment and informational.

Very high inventory

While users in this group will have more inventory than others (say more than 200 pieces per day) and will generally consume more content, the percentage of available inventory consumed is relatively low; these users have far more content than they can possible consume. As with high-inventory users, ranking is critical to maintaining and improving the engagement of this group.

INVENTORY: KEY METRICS TO TRACK

  • Amount of inventory available
  • Number of connections
  • Consumption of available inventory
  • Number of posts consumed
  • Percent of users who are inventory constrained

When tracking inventory metrics, segmenting your data by country, gender, type of connection (friend, celebrity, business, group), content type (text, video, photo, etc.) and platform (iOS, Android, desktop) may lead to a clearer picture on how your inventory is performing.

TAKEAWAYS

  • Increasing the right connections helps drive engagement.
  • Identifying needy users and ensuring they have a great experience is paramount to long-term product success.
  • Understanding the consumption and availability of inventory for different segments and identifying specific opportunities will help serve them better and thereby improve the overall product.

Follow us on Medium for weekly updates.


This work is a product of Sequoia Capital’s Data Science team. Chandra Narayanan and Hem Wadhar wrote this post. We would like to thank Jamie Cuffe and Jenny Wang for their contributions to this post. Please email data-science@sequoiacap.com with questions, comments and other feedback.

Related Article Data-informed product building playbook.
国产人妻熟女偷在线视频_人妻熟女中文字幕在线播放_人妻熟女中文字幕在线播放