We are working on Oracle Endeca 3.1.2 with Experience Manager. For pushing workbench content across environments and for managing version control of the Experience Manager content, we follow the same build scripts as provided by Endeca Discover application for loading the Discover workbench sample data (\apps\Discover\config\ifcr). We maintain similar folders structure in svn for content,pages and keyword redirects. We have sample content for keyword redirects but we do not have the sample content for thesaurus entries. If anyone has tried creating thesaurus related files then please share the json structure. We are not using export site and import site as we cannot take backups of specific page contents.