Managing Search Relevance in SharePoint 2013 and O365

Play Managing Search Relevance in SharePoint 2013 and O365
Sign in to queue


Improve the value of search results by managing Relevance. SharePoint 2013 and O365 provide a comprehensive set of tools to manage Search Relevance. SharePoint 2013 and O365 combine and improve upon the best tools available for Fast Search and SharePoint 2010, and introduce the notion of Query Rules that allow administrators to influence relevance for a single query or sets of queries using a straightforward user interface. In this talk, we will explain how ranking works under the hood, and give hands on demonstrations of how to use query rules, result sources, XRANK, federation to external search providers, and use the rank tuning tool to improve relevance and the value of your search solutions. We will explain the Relevance Lifecycle, the iterative process of finding problem queries and fixing them that is essential to optimizing relevance, and discuss which tools are most appropriate for various situations.



Right click to download this episode

The Discussion

  • User profile image

    Please feel free to post or send questions and comments.  We will use them to improve the content and to provide feedback on the products.  Thanks!

  • User profile image

    Small correction/clarification- the Rank Tool can be installed in on-premise environments (it is available in the store) but, as documented, it will crash after a known serious of events.  This should be resolved with SP1. 

  • User profile image

    Hi, thanks for the post.
    I have few questions
    Question 1: Are there are any drawbacks or limitations of creating the custom ranking model through RankingModelTuningApp in SharePoint Online?

    Question 2: Can you please share details on creating the judgement sets? I tired with the below method but the managed properties for HTML, DOC, PPT, XLS, TXT are not getting generated in the Add and Tune features section.

    The import file must be of type XML with the following schema:
    <QuerySet Name="testRM - JudgementSet">
    <Query QueryString="query1" >
    <Document Url="docUrl1" Label="Excellent" />
    <Document Url="docUrl2" Label="Good" />
    <Document Url="docUrl3" Label="Fair" />
    <Document Url="docUrl4" Label="Bad" />
    You can use four labels to indicate how desirable a result is for a query: Excellent, Good, Fair, and Bad.

Add Your 2 Cents