Monitoring Social Media
Why Many Social Media Monitoring Tools Are Inadequate
We surveyed a handful of social media monitoring tools and found a wide array of interfaces for creating queries. For the most part, many of them were severely limited, but it may not be apparent until you’ve already laid out the money for the tool and then find out how difficult it is to limit the results to what is relevant because query generation is inadequate.
Boolean Search Query User Interface
Some tools will just let you directly type in a string of keywords and Boolean operators, as in Fresh Web Explorer:
However, even if they say they will accept a Boolean search expression, it is still important to see which Boolean search operators are available. In this example of Fresh Web Explorer, they do not accept parenthesis in their Boolean search expressions. Without the use of parenthesis, you can’t specify the order of how you want to search or parse groups of keywords that are variations of each other and the search falls apart.
Some tools will provide boxes or fields for you to add a Boolean search query with additional filters available at the time of the query, as in this example from the tool, Talkwalker.
Other tools will provide a field for you to add a Boolean search query and let you do a test of the query so you can see if it is returning the kinds of results that you want, as in this example from the tool, Brandwatch.
Some tools have an interface with separate fields where you can list keywords that must be included in the search and keywords you want to exclude from the search. But unless they allow you to tweak the Boolean search string that was just built by the tool, there may be no way to specify order/proximity, variations within those sets of keywords, wildcards or other factors.
The following screenshot from the tool, Mention, looks easy enough, but you cannot specify order or group search strings together with parenthesis.
Reducing Irrelevant Results
Some social media monitoring tools have licenses based on a per-month limit to the number of mentions that are returned. If a large percentage of your results are irrelevant, then you’re not only wasting time, but you’re wasting money!
Even if a tool has advanced query functionality to fine tune what to search for, we all know the internet is full of bogus web sites that are full of spam links and scraped content. Having a tool which can filter those out of the search queries will be very useful if you need to focus only on credible mentions of your brand name. Having to wade through your result data and delete all of the spammy, bogus, or irrelevant mentions is very time consuming. See if the social media monitoring tool you are considering has built-in spam detection or anti-noise technology.
Additional Search Operators You May Need
All social monitoring tools handle the basic Boolean operators AND, OR and NOT. Most should support additional Boolean functionality like quotes, parentheses, wildcards and proximity. Here is a summary of some additional search criteria that should be supported by the social monitoring tool you are considering in order to build an effective query:
1. Special characters – You may need to handle characters such as hyphens, apostrophes and ampersands in queries. For example, can the query include the special characters in brand names such as Hewlett-Packard, AT&T, and Macy’s without returning an error? Some tools provide a raw data operator that defines a search keyword that consists of a string of characters that includes special characters. Without such an operator, the query could return a syntax error, or the special character may be ignored.
For example, a search onGoogle+ may ignore the plus sign. In the social media monitoring tool Brandwatch, searching with raw:Google+ will include the plus sign. In the Talkwalker tool, it uses a plus sign to indicate raw data. So +”M&M” would include the ampersand (&) character in the search.
2. Case – If upper or lower case matters in the types of things you are searching, see if the tool can support searches where letter case is respected. In Talkwalker, two plus signs will respect case, as in ++”eBay”, where the B must be capitalized in the mentions retrieved.
3. Boolean Operator Words – Can the query differentiate between words that are part of the keywords you are monitoring and the Boolean search operators, such as AND, or, and NOT? When evaluating the social monitoring tool, Brandwatch, we discovered it could not tell the difference, however, it had enough other powerful search operators that we could easily work around this limitation. Other tools only treat such words as Boolean operators when they are in all capital letters. In Talkwalker, a query of free and easy will return mentions of the phrasefree and easy. However a query of free AND easy will return mentions where the word free and the word easy appear anywhere.
4. URL or domain – Sometimes we want to limit our results to be only from a specific domain name or specific URL, or alternatively we want to exclude a specific domain or URL from our results. Perhaps we have a restaurant chain in many locations and we only want to look for our company name on a specific review site. This may be expressed in a format like site:www.yelp.comfor an entire domain or perhaps a specific URL or subdirectory, for example if we only wanted to see reviews for our restaurants in New York City, url:www.yelp.com/nyc. Alternatively, we may want to exclude mentions coming from our own brand’s site, for example, NOT site:www.my-restaurant.com.
5. Author – If you expect to be looking for mentions related to a person, then you may need a tool which supports an author:operator. When detectable, it would retrieve mentions from specific authors. For example, you don’t want every mention of author Stephen King, but you want to find articles where he was the author. You would add author:”Stephen King” to the query.
6. Title – Depending on the tool, doing a search with the Title: tag may look for mentions of your query in the titles of an article, or it may look for mentions in the title of the web page, as specified in the <title> tag.
7. Geographic location – If our business is local or we are targeting a specific geographical area, we would want to limit mentions to sites from those areas or sites talking about those areas. Some tools have query operators just for location or country, some provide a way to automatically include variations of a location’s name (e.g., New York City, NYC) as part of the query, and others allow post-search filtering of the mentions.
Specifying Location in the Query
In the social media monitoring tool Brandwatch, specifying country:ukin a query would only find mentions that have been identified as from the UK. Brandwatch also has a pop-up that constructs location operators for specific areas, so entering New York City would create the location operator of city:new8.
Distinguishing Location Mentions from Origin Location
Other tools allow separate distinctions of finding the location name in the source, versus finding the country of origin of the source. For example, with the tool Talkwalker, a filter of articlecountry:dewould include all articles that contain the word Germany, while a filter of sourcecountry:dewould include articles from German sources or articles published in Germany.
Filter by Location After Query
The following screenshot shows how the social media monitoring tool, Brandwatch, allows you to filter by location after you have your query results – either by including locations or excluding them.
8. Ability to save your queries – After building a complex query, you don’t want to go through the trouble of remembering what you’ve searched on every time. The tool must be able to save your queries so you can rerun them again and again, as well as edit and fine tune them. Some tools may limit the number or queries you can save, based on the licensing version you have purchased. If you’ve put in a lot of work to construct a complex query, it is probably a good idea to save those query strings somewhere locally in a document just in case you need them again, for example, if you accidentally delete them or would like to reuse them for another client in the future.
The Key to Success
Even if a social media monitoring tool has great bells and whistles for analysis and reporting, if the initial set of data from the query is full of garbage you don’t care about, then you’re wasting your time and skewing your analysis because you’re basing it on data that includes irrelevant results.
Therefore the key to the successful analysis of your data is the query. The key to an effective query is the ability to include and exclude keywords and variations of keywords, specify proximity, and parse it all into a logical hierarchy so related parts of the query can be grouped together and operated on in the appropriate order. When you can narrow down the scope of your results to be focused on only relevant results, you save time. Also any analysis performed on the data will be more accurate.