Talend is an innovative vendor of big data integration software. Their documentation was published on the web, but in a linear, book-type format. Their customers often used Google to look for support, and when they clicked through to the documentation, the information and the links on the page were often insufficient to provide the needed answer.
To solve this information and findability deficit, Talend moved to a granular information model, based on articles rather than books. An article could be a paragraph, or a very long web page, but it always had a clear purpose and an appropriate level of detail to answer a customer question. Structured content consultants Mekon helped Talend develop a taxonomy and metadata framework to tag the content at various levels.
This tagging was used to generate "recommended links" that were more relevant than those generated by simple keyword matching. It also helped in personalizing content at the block or even phrase level, so that people looking at the same page would see accurate information for the particular tool or product set they were using. Talend achieved what many organizations hope for — a robust, extensible taxonomy that drives improved customer experience across several touchpoints.
The problem that Talend’s users faced was the familiar one of findability, but with particular issues associated with technical communications: the adaptation of a book-oriented publishing architecture to drive web content in a knowledge base.
Talend used an XML architecture to manage and publish a large number of online user manuals, tailored to each particular product variant but with a lot of near-duplicate content. For any given page title, there would be dozens; possibly hundreds of very similar pages, differing in feature details or the context that the page was intended for (for example, a particular product configuration installed on a particular operating system). Google and other external search engines had no way of knowing which page would suit each searcher, and the duplicate pages also reduced each others’ page ranking. Once a user landed on a page, they were often unsure whether it was the correct page for their particular situation.
For this first problem, Mekon helped Talend to design a new information design; one that drastically reduced duplication. Whenever there was a procedure or overview that was common to many products, only one page would be written. Where subscription-based products offered enhanced features over the open source offering, these features were highlighted in the content, providing clarity to users as to the content’s applicability to their own situation, and perhaps indirectly encouraging users to purchase a subscription. Always, the general applicability of the page would be indicated with displayed tags at the top. Both the highlighting and tagging features were driven by a central taxonomy, managed in the PoolParty semantic suite.
The taxonomy also provided crucial relationships between the various published articles. The one benefit of the old book-based model was that once in a book, there was some sense of structure and flow between the various pages, even if that was a linear structure that would not suit every user.
Talend were able to create new relationships between pages, more suited to web content, by leveraging the taxonomy concepts applied to each page to drive intelligent "See also" links. More than simple keyword matching, the approach combined facets with Boolean logic to drive more relevant recommendations. As one would expect, the tagging was also exposed in faceted search UIs that were simple enough to suit all users, but powerful enough to quickly drill down to the exact article that a customer needed.