Teachers Pay Teachers: building a rich and flexible search platform to scale a growing marketplace

The challenge

Sophisticated users, big catalog, better filters and facets. 

As a marketplace with a wide catalog, TpT’s main challenge is to connect users to the right content. User queries usually provide thousands of results, especially as TpT’s users are very sophisticated and will search for very specific resources. TpT’s teams need to make sure users have all the facets and filters they need to refine their search and drill down into relevant results in a matter of a few clicks. 

TpT’s team was using Sphinx, an open source solution supported by a full-fledged search team for the last 5-7 years. The company needed to move to a more modern search engine. In addition to the above requirements for a new search solution, scalability and performance were key as Sphinx had already proved to be unreliable during busy sales periods. Sheer improvement in the user experience was also key to ensuring smooth interactions with buyers. 

The solution

Flexible and scalable Search-as-a-Service platform to support the marketplace’s needs 

After evaluating different search vendors, TpT teams chose Algolia to power their search. They were confident that Algolia would support the scale they were projecting, as well as require a low maintenance effort. But most of all, they believed that Algolia’s Search-as-a-Service model and roadmap was the most aligned with their goals. 

“We want to be experts in educational resources, not experts in search systems.” Harlan Harris, Data Science Team Manager @Teachers Pay Teachers 

Seeing that Algolia is a SaaS solution, the TpT team was pleasantly surprised by how flexible and customizable the platform is. As their users tend to use fairly complex queries, TpT needed to use a combination of small and large changes to solve relevance issues. With Algolia, they were able to implement granular relevance changes, and implement a complex ranking strategy matching their performance criteria. 

It took about 4 months for Algolia to be integrated on TpT’s marketplace, before rolling it out on their school-facing product. In addition to using Algolia as the main search for educational resources on the marketplace, TpT also uses Algolia to filter reviews and ratings in the listings, as well as to make the sellers’ stores searchable. From a single search platform, TpT is able to power its whole catalog (or a relevant subset) on multiple channels. Post implementation, TpT worked with their sellers’ community to make sure they had visibility of the changes and were happy with them.

The result

Performance and reliability means more time to innovate, while increasing revenue 

After implementing Algolia, TpT was happy to see their efforts rewarded. They registered a 2% increase in conversion rate - which is very meaningful in their case — as well as an increase in Click-through-rate. They also measured improvement in search efficiency — the time needed by a user from first search to checkout: more users were able to get what they wanted in 5 minutes or 5 queries or less. As a result, TpT’s sellers were happier, and their buyers were able to see more relevant resources. 

Perhaps most importantly, Algolia has contributed to improving the user experience, by making engineering teams able to focus on the user rather than search systems. Engineers can make changes rapidly and iterate faster on their future strategy (faceting, indexing, and synonyms, to name a few examples).

What’s next

Using Algolia AI studio to connect users to relevant content

After this first launch, TpT is excited to test the latest Algolia AI features such as Dynamic Re-Ranking. Being a beta tester on the Dynamic Synonym Suggestions has helped TpT solve some of their most complex user queries. They’re also thinking about a potential change in taxonomy, and its implications on search. The TpT team is excited to experiment and to discover more in the years to come.