18.10.2019

The 5 most common mistakes when defining queries for social media monitoring with Boolean search

In this article, we're going to look into the most common issues that come up when you're building queries for social media monitoring.
It's aimed at those who are already familiar with Boolean search and how it helps you to get the mentions you need. Although the basics are not covered here, there's some information below that might be useful for people who are relatively new to the topic and those who have been using it for a while.

The mistakes were drawn from the day-to-day practice of reputation professionals at Topface Media who specialise in social listening. They are epidemic in businesses that have been collecting mentions in-house before outsourcing brand monitoring, and often come up when training new insight managers within the agency.
Boolean search is available through a large number of social media monitoring platforms like Brandwatch, Awario, TweetDeck, Sprout Social and YouScan.
The most basic operators that are used uniformly across all platforms are OR, () and "". Other popular operators include:
Defining how close keywords should be to each other, i.e how many words can be between them. The most popular symbols on different platforms: ~N (Tilde), /N, near/N (where N is the number of words between the keywords)
Excluding words. The most popular symbols on different platforms: NOT or AND NOT, -
Searching for posts that have two or more keywords together. The most popular symbols on different platforms: AND, +
Some platforms also let you search for a specific social media network or site (FROM, site:), country (country:, location:) or language (lang:). Other platforms only allow to filter out these parameters in the main dashboard once the search has been completed.
Despite the differences in syntax and number of operators available from platform to platform, the advice on how to build queries can be applied to any of them.
1. Getting the spelling right is wrong
Misspelling common words is a popular way for companies to get desired domain names and make their brand names unique, but it becomes a problem when people start talking about your products and services, and, specifically, make posts about them. Examples of these types of brand names are very numerous: Tumblr, Flickr, Cingular, Kool-Aid, Car Hirez and so on.
While famous brands like Twitter and Google mean there's almost no chance of people not knowing the correct spelling, it's not true with less known brands, and especially for customers abroad.
Apart from that, there are lots of names that you wouldn't be able to spell correctly even after hearing or seeing them a couple of times, like M&M'S or Dunkin' Donuts. With these names, social media users most likely won't bother typing up apostrophes or resist the natural temptation to spell "donuts" as "doughnuts".

One of the most problematic industries, when it comes to spelling, is traditionally pharmaceuticals.
Boolean search in social media monitoring platforms doesn't work like Google search (which will help you find anything no matter how you spell it), so it's important to think of all the options your customers might come up with.
2. Building a query that's too narrow
When you are collecting mentions for your brand, usually the task is to get as many of them as possible. Once you have the full picture, it's easy to filter mentions and categorize them. In this situation, adding other words will limit your search.

For some reason, words that unnecessarily impair the search results can often be seen in queries for unique brand names.
Unique brand names are those which don't resemble any other words, like Aciclovir, Revlon or Hitachi. If you type them in Google you won't find anything apart from relevant results. The opposite of a unique brand name would be something like Happy View or Mountain Dew.
So, let's say we have a company that sells hair loss products. Very often people would build a query like this:
And it's not correct. Here, we will get results for the unique brand name within the context of the brand category, which will significantly limit the search. Our mentions will include phrases like "Alerana spray review" or "bought Alerana shampoo yesterday", while posts like "love Alerana, it saved me from hair loss" won't be found.

How to do it right:
3. Building a query for a brand name that's not unique
Brand names that contain or consist of common words create a lot of issues when building queries. Usually, it means that you will need to adjust your query after the mentions have been collected.
One of the most common mistakes when defining the context for the search is not grouping the words properly with brackets.
Here's an example of an incorrect query for the makeup brand Bell.
In this case Bell is going to be searched along with makeup, while all the following words (cosmetics, lipstick, mascara, etc) will turn out results without the brand name. Obviously, we don't want to collect mentions of all lipsticks in the world.

Here's a correct query. We added two more brackets to tell the system that we need each of those words to appear with the brand name.
4. Not excluding words
This concerns both unique and non-unique brand names, especially when you are dealing with a large amount of mentions that take time to process and analyze. Some platforms can show preliminary results informing you of mentions that don't relate to the brand.
In all other cases, it is better to think about what might get in the way and run a quick Google search for your brand name.
First of all, you should check for brands that might have a similar name (then you can exclude words that describe this company's products or services), or other meanings that your brand name might have (for example, for brands like Swag or YOLO, will need to find a common pattern in the context where these words are used as slang to allow the exclusion of unwanted mentions).

It's also useful to think about what kind of posts you are going to analyze. For example, do you need to know what people think about the company as an employer? If not, you can exclude words like "jobs", "vacancies", "employee".

If the brand name is not unique, there are words that you can exclude right from the start even without a research, for example, "porn". It might sound surprising but insight managers who deal with social listening for different companies on a daily basis name it as one of the most common words to exclude. Turns out Rule 34 (if it exists, there is porn of it) checks out.
5. Double checking the syntax
Sounds extremely obvious but this happens all too often to people setting up queries on a regular basis. Make sure your ORs and ANDs are capitalized and don't forget the golden rule of Boolean search: the number of opening brackets should always be equal to the number of closing brackets.
By clicking this button you grant your consent for processing of your personal data and agree with Privacy Policy
Useful articles
Show more