Using Promoted, Outschool increased conversions in search by +11.2%, increasing Outschool's total revenue and unit profitability by several hundred basis points. Next steps include optimizing directly for user LTV, adding new surfaces, and more ML investments like categorization.

Outschool is an online learning platform that empowers kids ages 3–18 to learn on their own terms with the widest variety of teachers, topics and classes to suit any interest. They offer interactive, small-group classes that are fun and social, video classes that can be taken any time, and clubs that connect learners around their favorite topics.

Outschool has raised over $240M in venture capital, most recently in a Series D, from world class partners, and is among the top 20 largest private consumer marketplaces. Today, Outschool offers more than 140,000 live online classes to more than 1,000,000 learners in 183 countries worldwide.

Kejia Zhu
Product Lead, Outschool
Erwin Alberto
Engineering Manager, Outschool
“Our journey in improving and optimizing our search conversion led us to Promoted and we couldn’t be happier about how this turned out. Promoted allowed us to deliver an ML powered search product much faster than if we had built it ourselves. This resulted in us delivering relevant search results to our users. Additionally, the Promoted team is a delight to work with. The whole time we were working with them we considered them as part of the team and more than just a service provider providing a solution to us.”

— Erwin Alberto, Engineering Manager
Search, Discovery, and Merchandising
“Working with the Promoted team has been nothing but a pleasure. Our team has been working on optimizing our Search platform and hypothesized that an ML based platform would take our search to the next level. Promoted allowed us to accomplish this at an accelerated level. The integration process couldn’t be easier. Promoted’s SDK was simple to work with and well-documented. Their team was always a delight to work with and readily available for discussions whenever needed.”

— Christopher Soto, Staff Software Engineer
Search & Discovery
Christopher Soto
Staff Software Engineer, Outschool

Outschool’s mission is to inspire kids to love learning. They believe the best way to do that is by linking learning to interests and giving kids the autonomy to pick their own path. To do that, Outschool wanted to better match learners with educators to increase search conversion rates, total GMV, and profitability. Existing recent investments in search quality were promising, and Outschool wanted to extend these investments to further increase conversion rates.

These investments using had already brought Outschool search conversion optimization to the state-of-the-art in terms of systems and practices. Any further improvement would need to be an order-of-magnitude investment in a combination of user metrics collection, data processing, modeling, and real-time serving.

Outschool primarily compared Promoted to building search and discovery improvements in-house. Outschool chose Promoted because they weighed the cost and risk of the aforementioned infrastructure investment against partnering with the highly experienced and recommended Promoted team to get a working, managed solution that they would benefit from immediately.

Structurally, Outschool is a complex two-sided peer-marketplace with time-sensitive, unique, and variable inventory. Its web app combines Segment for measuring user engagement with in-house search retrieval solutions using Elasticsearch and OpenSearch.

Promoted maximized conversions (+11.2%) in search by applying the state-of-the-art in conversion optimization ads engineering to the problem of marketplace search optimization. Promoted created a unified organic and promotion system for Outschool in three steps:

1Collect and join metrics

Outschool installed the Promoted client and server SDKs to collect more, higher quality metrics like insertions, impressions, and clicks. Then, they integrated existing third-party event systems like Segment to add conversions. Promoted’s Metrics system automatically joined and aggregated these event streams in real-time for use in optimization and reporting.
2Re-rank search results

Outschool sent its search results to Promoted's Delivery system for optimized re-ranking and blending. Promoted and Outschool ran A/B user studies to measure progress and total impact.
3Continuous optimization

Metrics and Delivery create a feedback loop to continue to optimize and improve results over time. Through this framework, Promoted can support new allocation rules, product logic, and data. Outschool has the flexibility to add new listing types, merchandising, and internal promotions to its feed and search at any time, and Promoted's optimization will continue to work automatically.

Our Client Metrics Libraries integrated with Outschool's web app to deliver the real-time data that powers Promoted Delivery. When integrating web frontends with Promoted, you can use our state-of-the-art client libraries to get up and running quickly and correctly. Our libraries are tailored specifically for Promoted Delivery.

The Promoted Delivery API optimizes listings in search, feed, or paid promotions. Call our Delivery API where you would call a database or a search service like Elastic. We re-rank on every request to deliver the best possible listings, personalized to each user and context in real-time. To get your marketplace optimized faster, use shadow traffic to safely train and test our Delivery systems so you can be confident before live production.

Visit our Client Metrics SDKs and Server SDKs pages to learn more.
© 2022