Hub - Structured Search & Bots

Search, navigation and discovery are the primary drivers for eBay’s intranet portal. As a system that doesn’t focus primarily on business processes and transactions, signposting is one of the platforms key functions.

Historically search had been a tough user problem to solve, understanding intent, gathering and indexing content, providing a system was meaningful were all big challenges. Through 2018 with the rise of bots we started to look at our own technology and see if we could bring something similar to the users.

The result of this was initially launched mid 2018, targeting specific keywords in the platforms search engine and interrupting the flow with a dialog window. This offered users a more specific path way to finding the answers they were typically looking for.

There was good positive feedback at launch and learned quickly about what areas could be built upon. A lot of the early content was hard coded into the system but it was clear that federating the content management was going to be key. Partnering with the content owners and business partners the team developed a ‘mission editor’ for content owners to create their own flows.

The UX here proved somewhat challenging, the need for content owners to understand the flow of one question to other, the key words they could leverage, connecting nodes were all conceptual difficulties for users. Ultimately the flow chart approach resonated and with close guidance and training soon we had people building out more missions and flows in the tool.

Latterly the system expanded to integrate with other systems causing more UI and UX challenges, managing things like approval flows and interactive elements. Each of these had to be designed out with the limited chat interface in mind.

The team also wanted to federate the deployment of the chat interface across other sites and platforms. We explored different ways to invoke and interact with the chat window beyond the initial interface design.