Skip to main content
All CollectionsUser GuidesSocial Listening
Best Practices for Building Searches in Listening
Best Practices for Building Searches in Listening

Learn how to use the various keyword fields and filters to dial in your social listening analysis.

Seth Bridges avatar
Written by Seth Bridges
Updated over a year ago

We designed our Social Listening tools to help you measure and understand conversations and customer sentiment online. This guide will walk you through the different filters and keyword fields you can use to create meaningful searches and extract the most relevant information for your listening analyses.

Choosing Keywords

One of the first things to consider when building a search is to choose the right set of keywords. Our tool provides you with three different keyword sets to narrow down your social listening inquiry:

FIND posts with any of these primary terms

The first input enables users to find posts or documents that include any of the primary terms in the body of the post or article (but not the title). The more words you include in this field, the broader your search will be.

Also, the primary terms are crucial, as all sentiment analysis will be computed relative to any one of the primary terms defined.

Consider, for instance, a use case like a brand health analysis. You might want to include the name of your business, social media handles, or branded hashtags commonly used in discussions about your business. To look for financial results of public companies, including the stock ticker symbol may also be useful.

Example

To study retailer Nordstrom, we can start with the name of their brand, their primary social handle, and a branded hashtags.

INCLUDE posts with any of the following terms

This second field helps increase the precision of your search. It reduces the document set found by the primary terms to include only documents that contain at least one of the terms included in this input within either the body OR the title of the document. (Note that some types of documents, like Tweets, don't have a title field.)

Of course, you don't have to specify anything in this field, and at times, leaving it blank might be a good choice. If your primary terms aren't ambiguous (e.g., Nordstrom, Cetaphil, or ColourPop), then you may want to start your analysis as broad as possible by leaving this field blank.

However, for brand names easily confused with common words, you'll need to use this field to focus our analysis.

Example

Returning to our Nordstrom example, we don't necessarily need to use any include terms to focus the search on the brand because the term "Nordstrom" isn't particularly ambiguous.

That said, if we'd like to limit our search to a particular subset of the conversation, we can use include terms.

For example, to limit our Nordstrom search to discussion of shoes and footwear, we can use various footwear-related terms. Wildcards can help us match more related documents.

EXCLUDE posts with any one of the following terms

This third input helps refine your query further by excluding any documents that contain any word or phrase included within this field. It is useful for removing irrelevant or off-topic discussions.

Example

Continuing with our Nordstrom example, we can use exclusions to further focus our search and remove unwanted noise.

For example, we use various exclusions about "coupons" and "promo codes" to remove spam. Removing the terms "stock," "buy rating," and "research report" help us remove stock analysis posts.

Tip: Use the Associated Terms panel in the query editor to quickly identify noise terms and exclude them.

Selecting Sources

Another critical part of building a search is selecting the data sources. By default, the tool will include data from all available sources, but you can choose any combination of sources you want, including just a single source.

Reducing Noise

The last set of inputs in our search builder is designed to help reduce noise.

Exclude unwanted content

We offer a set of checkboxes to help users exclude various types of typically unwanted content, such as ads, sponsored posts, job ads, coupons, pornography, auto-posts, and profanity.

Exclude high-volume authors

The tool also includes a feature to exclude noisy or high-volume authors. The application automatically provides a list of authors who post most frequently about your currently defined search.

Domain exclusion

Similarly, there is a domain exclusion capability. This allows you to exclude any domains that contain spam, noise, or irrelevant information for your analysis. The tool automatically provides a list of high-volume domains within the data set.

Did this answer your question?