What?
Related searches was a project to deliver relevant related searches to the area just below the search box in the SDC header (see below). For the US we delivered about 7million new keywords sourced from eBay. We also updated the our site map and implemented canonical tags for our search results pages. This first phase of related searches covered 13% of searches, the second phase (Soft Match) launched in Aug and covered about 96% of queries (Miko will send out more details on the Soft Match release). Note, that what is currently in production for related searches is Phase 2 (Soft Match). Below is an example of the related searches for “toaster oven”.
When?
At the end of June, we launched Related Searches (KW URL 5) to production. In mid-Aug phase 2 launched (Miko will send out more info on the results of the Phase 2 Launch).
Why?
The previous related searches were targeted only by the category and attributes (i.e. not necessarily related to the keywords entered by the user), so they usually created a bad user experience. We wanted to have a much larger volume of keywords that were highly targeted to the user’s search query. The large volume was expected to generate additional SEO traffic because these “related searches” links are one of the main ways that Google finds the content on our site (Google bot does not enter search terms on our site.)
What Happened?
We saw a significant increase in SEO traffic for cross-category pages (where most related searches land) in the US.
Click-ins on product pages is shown as a point of reference (related searches don’t link to product pages).
The lift shown above only occurred in the US. We believe related searches was not as successful for SEO in the international countries because International countries have much less content which means that we created a lot of pages based on the new keywords but we didn’t have the selection of deals to properly fill these pages. We do have a filter to not show related searches that have no deals, but on average the international pages had fewer deals per page. But all the countries benefited from more relevant related searches for our users.
Here is the data for each country:
How?
This project was a great example of cross-pillar collaboration as it relied heavily on Research, FE, Search, Data Warehouse and SEO. Even though we had almost no dev support from the search team, because they were booked with Kadu qualification, Damon from community went above and beyond the call of duty to customize a version of Kadu to power related searches which allowed us to successfully launch this important project. Tamir creatively suggested that we leverage eBay’s keywords which saved a lot time in the end.
This project is also a great example of being able to leverage the Kadu platform (it runs on version of Kadu).
Who?
Special Thanks to the great work of the people on Phase 1 (Hard match) team:
Dev: Dmitry Yesin, Zin Lim, Joe Hanink, Clyde Jones, Damon Sauve, Tamir Rozenberg
Ops: Lakshman Mandali
QA: Morgan Weiss, Jane Lisinker-Korobov, Emilia Gentcheva
Research: Dominic Hughes
SEO: Johnny Jiang
Product: Tamar Bar, David Little
BI: Lana Schuler
eBay Support: Dan Kramer
