Personalization impact
Number of Likes
55%
The client
Entre is a new professional social network designed for the future of work. Created to support the changing ways in which people work and network, Entre aims to help individuals and companies connect, collaborate and communicate in the digital world. Bringing together more than 80,000 entrepreneurs from all over the globe in just a few months, Entre enables its users to connect with other like-minded individuals to discover opportunities, find employees and monetize their expertise.
The challenge
Entre’s “For You” feed is the town square and the heart of the community. It’s where the users meet, exchange ideas, build relationships and kick off collaborations. To start with, Entre relied on a rule-based approach to prioritize the most popular content in the feed. However, as the community scaled to tens of thousands of users with diverse interests, Michael saw an opportunity to grow user engagement and retention through personalization - presenting each individual user with the most relevant content for them every time they log in, a vital aspect of bringing them back to the platform on a frequent basis. Personalization also enables content creators to benefit from engagement with their content from a relevant audience, encouraging them to create more and better content, creating a virtuous cycle.
The team run the numbers and discovered that users who consumed 30% more content during their first week on the platform, were 50% more likely to keep using Entre after 30 days. Michael got the proof, and Juan Carlos, Entre’s CTO, got the task of personalizing the feed. Juan Carlos knew that building a real-time personalization engine in-house could be a risky and costly endeavour, and decided to look for a customizable solution that could be deployed at speed and within budget. After exploring the solutions available on the market, he chose Superlinked’s personalization infrastructure. This was because, compared with other options, it was easy to integrate and allowed Entre the flexibility to leverage their user understanding to test and evaluate different approaches.
To help support our continued growth as a community, we needed a solution we could configure to suit our vision of the platform. However, we found that other solutions in the market were a “black box”. Unlike with Superlinked, we couldn’t configure how the content should be selected or understand how their recommendations were produced. Superlinked’s solution is flexible enough to meet the needs and interests of all our users.
Michael Marra
CEO/Founder of Entre
The solution
Entre’s team used Superlinked’s infrastructure to configure, launch and fine-tune a real-time personalization engine for their community, driving significant growth in engagement. This involved:
Identifying the key success metric: Entre’s analysis showed a high correlation between initial content consumption and longer-term user retention. Therefore Entre’s team tracked content engagement and feed scroll depth as they evaluated different configurations.
Integrating with Superlinked: Entre’s team defined the initial personalization strategy, balancing between content recency, popularity and most importantly relevance to the user, based on their profile and past interactions with the platform. To support real-time recommendations, the team mapped user interactions to Superlinked API, and connected the feed generation to the model results.
Launching and A/B testing an ML-powered personalization: Superlinked’s infrastructure allowed running Entre’s rule-based algorithm alongside the personalized model. The team evaluated the results on a weekly basis and adjusted the model accordingly, until a statistically significant increase in the target metrics was achieved.
Full roll-out: Following this process, Entre’s team identified the best-performing variant, which generated a 55% increase in content engagement, 59% increase in user profile views and a 12% increase in scroll depth, and rolled it out to 100% of the users.
Continuous optimization: Entre’s tech team can now continue adjusting the recommendation factors and weights, without any coding, and A/B test those against the currently best-performing variant.
Superlinked’s API was easy to integrate with and flexible enough to support our user schema, and the team were highly responsive and great to work with. Now that the integration is in place, we have the freedom to experiment without having to talk to Superlinked
Juan Carlos
CTO at Entre
Results
Entre achieved a statistically significant increase in the number of likes (+55%), comments (+22%), and engagement with other users (+33%). The average scroll-depth (number of content items viewed per session) increased by 12%, signalling that users want to see more relevant content.
The benefits
Real-time personalization
Superlinked’s ready-to-use infrastructure enables Entre to personalize social experiences in real-time so that content feeds are continuously improved. This means that instead of waiting for the next login, user interactions are used during the same session to update the models and deliver ever more relevant content, driving up engagement.
Quick-to-market
With Entre’s ambitions to achieve key metric improvements at pace, the team also appreciated that Superlinked’s infrastructure enabled the personalization to be deployed within days. All they needed to do is connect to the APIs to start testing the personalized feed in production.
Super-flexible
Superlinked’s machine learning infrastructure is flexible enough to allow Entre to easily configure a model combining topical relevance, contextual popularity and recency, without the need for in-house Data Science expertise.
Privacy first
Another plus for Entre was that Superlinked’s infrastructure is privacy first. This means that no personal data is required for personalization, helping protect the confidentiality of their users.
Superlinked’s platform allowed us to achieve a 50% higher engagement with the feed, and we are now looking to take full advantage of the integration and personalize other aspects of Entre, including community discussions and job listings.
Michael Marra
CEO/Founder of Entre
Advancing entrepreneurial success
The teams at Entre and Superlinked continue to collaborate to build on the success of the new personalization engine. As Entre’s user base and user engagement grow, they aim to leverage the power of real-time personalization to deliver the most relevant and engaging content in other parts of the platform. Entre’s team is now empowered to deploy and test additional model variants, helping to ensure that they maximize results - however, user interests and behaviors evolve.