Exploring Content Recommendation and our Weekly Update 5/3
Turns out content recommendation can take multiple forms
Welcome back to another weekly update. This week has been a bit quiet. Some of that has to do with us hiring 2 new people and part of it has to do with waiting for experiments to play out. Regardless we have moved closer to our 3 main goals this week.
We also have an update about our road map and a couple of calls to action for the readers. If you would like to continue following our updates please subscribe.
User Accounts
This last week we moved from 119,945 to 128,911 accounts. Showing solid growth for users during the last month. We added on average 1716 accounts a day. Last week I claimed that we were "On Track" to reach the 1 million accounts goal, this is not completely true. We are boosting our daily account numbers, but we will need to get to around 3,000 - 4,000 accounts a day in order to hit our 1 million goals.
The account creation numbers above are from google analytics and it doesn't capture all of the account creations. Our data from the database is much more accurate and shows higher numbers. Regardless this data is useful to understand trends and right now the user account creation trend seems to be settling down to a new baseline of 1,500 daily sign-ups.
To make improve the output in the near term (This week) we will launch pro-metric ads for user account creation. After successful tests with pro-metric ads for premium sign-ups, We believe this will improve new user account creation by 40-80%.
Revenue
Subscription revenue has settled down from a recent high of 40-60 customers a day to the projected 15-25 a day. This next week will tell us if we can continue to see elevated sign-ups using our internal ads.
Advertising revenue is up this week slightly but still remains much below expectations. The team is started to spec out a new advertising system that will allow us to manage a larger inventory of advertisers. This inventory is limited to ExoClick, Crack Revenue, Internal Ads, and This new system will also give us the ability to more finely control the distribution of impressions on our advertising stock.
With more diverse inventory and better control of what we show to who, we expect to be able to increase advertising revenue by 2-5x over the long term.
Social Traffic (0.8 → 0.6%)
The number of daily social sessions to our site has dropped from a high of 0.8% back to 0.6%. Last week’s higher percentage seems to be from an abnormal increase in traffic from Facebook. After some research, I was not able to identify the single source of this traffic.
Social sharing has stayed steady since it started by used more recently and nothing major has launched yet that would drive more sharing.
Business Update
Our original road map was simple and highlighted areas that the team would be focused on as we continued to build towards our goals. We knew content recommendation would be on the list. What wasn't clear was just how varied this project could be. When we break this concept down into specific projects we get a widely diverse set. Below are two concepts for how to start implementing the content recommendations.
Updating the "For You" Feed
The main page is the "For you" page of scrolller. Right now it is the same for everyone and has 0 personalization. Tik Tok style recommendation is an obvious first thought, but it is hard to match their recommendation technology for a small company.
Other simple updates would include:
Adding and prioritizing content from your "following" selections.
Adding and prioritizing content from content similar to what you favorited plus what followed
Showing favorited content in your feed
Implementing the above could improve the experience, offer some level of personalization, and improve engagement for one of the most used parts of our app.
Personalized Content Emails
Another content recommendation touchpoint is the email inbox of users. Every day there are tens of thousands of new images, videos, and gifs being uploaded to scrolller. For those that opt-in, we could share what’s new and popular on a weekly basis. Improving retention among users and offering a new way for them to enjoy the content they are interested in.
Both of these projects offer the chance to start introducing content recommendations and build our competency in this new part of our business.
Call to Action:
What are your thoughts on the new content recommendation projects?
What is one feature that you wish Scrolller would support?