Resources Administration & Configuration

Query Suggestions

This page and sections contain an overview of leveraging and configuring Query Suggestions from the Dashboard. Once Query Suggestions functionality is implemented, then that index’s relevance can be optimized with Rules, synonyms, and other index settings.

At a high level, implementing and leveraging Query Suggestions consists of 3 steps:

  • Configure new query suggestions index in the dashboard section:  Query Suggestions
    • This step is discussed in this page in the sections bellow
    • Note: Once you’ve configured your new Query Suggestions index, the feature starts building suggestions based on the source index (the one that contains your search results).
  • Building a UI for your Query Suggestions index. 
    • This step is performed by the development team. Refer to the docs for additional details
  • Optimizing your Query Suggestions index’s relevance with Rules, synonyms, and other index settings. 

NOTE

All metrics on all tabs are index specific. If you’re seeing unexpected metrics, make sure you’ve selected the correct index and Date range. You can toggle Include all replicas to aggregate replica index analytics.

  1. Load the dashboard section “Analytics”
  2. Select the correct index
  3. Click on "New Query Suggestions Index" in the top right
    Overview
  4. Select the source index that will be used for generating Query Suggestions.
    • The source index is the index that contains your search results. 

  1. Load the dashboard section “Analytics”
  2. Select the correct index
  3. In the top navigation menu select “Searches” tab
  4. This tab shows:
    • Query - the most popular query terms
    • Count - the number of times they’ve been searched for
    • CTR - click-through rate
    • CVR - conversion rate
    • Click Pos. - the average click position
    • Total Searches - the percentage of all searches made up by this query term
    Searches
  5. In the “Query” column, click on a search term to see:
    • Average click position and a click position histogram, in addition to the previously listed metrics from step 4.
    • Rules, if any were applied
    • Filters, if any were used 
    • The results displayed for the search term
    Searches GIF

  1. Load the dashboard section “Analytics”
  2. Select the correct index
  3. In the top navigation menu select “Searches without Results” tab
    Searches without Results
  4. You can view all the searches your customers made on the website that returned no results. This information can be used to identify synonyms that you may need to configure in order to improve search relevance. Additionally, these queries can serve as an indicator of a trend or customer demand for products or services not currently featured in your online catalog, but might be worse considering to be added to it.

  1. Load the dashboard section “Analytics”
  2. Select the correct index
  3. In the top navigation menu select “Searches without Results” tab
    Searches without Results
  4. In the "Query" column, click on the search term to see 
    • Popular results for this search
    • Popular filters for this search
    Popular results for Search

Q: What is “empty search”? 

A: Search performed without text input:

  • Search-as-you-type experiences trigger a search request even before users start typing to instantly show results.
  • Users browsing the website without typing search queries in the search bar

Q: What is the analytical value of the “empty search” metric?

A: Analyse the relevance of your index, including the ways users interact with an empty search. For example, are products that show up first in the empty search popular or not?

Q: When to exclude the “empty search” metric from the analysis?

A: When an “empty search” lacks analytical value or skews the data. For example, most blogs sort articles by publication date, so empty search metrics give little insight into relevance.

Q: How to exclude the “empty search” metric from the analysis?

Note: you might need to be assisted by the engineering team to perform this task.

Q: How to prevent sending empty queries all together?

Note: you might need to be assisted by the development team to perform this task.