Queries are usually the starting point of any social media monitoring or listening initiative, so they are incredibly important.
Writing an accurate Query means accurate data – and who doesn’t want that? Analysing inaccurate data full of irrelevant mentions is time consuming, misleading and, really, pretty pointless. There’s no point in gaining insights from data that is full of useless mentions that you don’t care about.
So what is an accurate Query? Well, an accurate Query is one that pulls in data that is relevant and useful to your needs. It will bring in relevant mentions of whatever you are searching for, be it a brand, topic, competitor etc.
There’s no such thing as a ‘perfect Query’, but we usually advise to aim for about 90% accuracy. There will always be a few mentions that sneak in, and some mentions you miss, but this level of accuracy should be good enough for most needs.
If you’ve got a simple brand name that is unique, then writing an accurate Query should be no problem at all. For example, Toshiba is going to be pretty easy to write a good Query for.
But if you have a less distinct brand name, or something which is referred to in lots of different ways, then the Query writing process becomes a bit more complex. For example, think about how tricky it would be to search for brands such as Apple, Orange or Next.
Listen again to our Query webinar
We know that writing accurate Queries can be complex, so we’re always trying to give you as much help as possible. This week, we hosted a webinar with social media marketing agency DragonSearch to talk all about writing good Queries.
Within it, Jannette Wing Pazer from the agency talked us through some research she recently did for a whitepaper, using the case study of searching for P.C. Richard and Son. She walked through how she created the Query and the ways round some of the limitations she came across.
Our in-house Query expert Gemma Cooper then talked us through some of her top tips, such as how to research for a Query and the best ways of improving mention quality.
In case you missed it, you can watch/listen to the webinar again below. You can also find the slides here.
Below that, we’ve also detailed some of the questions we were asked or are frequently asked about Queries, along with their answer.
Want more tips for writing Queries? There are many on this site – here are some of the more recent ones.
Q: What would you do if you were looking at a Query like rheumatoid arthritis or multiple sclerosis, where people usually use common acronyms like RA or MS to talk about it?
A: You will need to add in lots of context terms, as obviously just searching for ‘ra’ or ‘ms’ is going to return a lot of irrelevant data. You could also search for those acronyms (using raw: to make it case sensitive perhaps) on specific sites, if you know that those acronyms used on those sites will be relevant (say, specific medical forums). To do this, use the site: or url: operators.
Q: How will you know that you are “done” with creating the Query?
A: As noted above, it’s impossible to make a perfect Query. Normally, if you have done the research suggested by Gemma, and included as many variations of the brand name and/or context terms you can think of, you should be ‘done’ Make sure to test your Query, and then check through the test results – aim for about 90% accuracy, and continue to refine (such as adding exclusions) if it isn’t accurate enough.
It’s impossible to know what exact volume of data you should be returning, but you can get a good idea by knowing which brand it is. For example, for a brand like Apple you should be expecting at least 100s of 1000s of results, whereas a small local brand is probably looking at more in the hundreds, up to maybe a few thousand.
Q: Constructing complex Boolean queries can be hard for many people in our industry. Is there a prompt or assist type function available in Brandwatch to help build the Queries?
A: We try to help our clients as much as we can. The Query editor has a list of all the operators and their functions, and we recently introduced the colouring of syntax when you’re building your Query so you can better follow the structure.
We also offer initial training to all clients, and can offer additional training specifically for Queries (at an additional cost – contact us for more information). Your Account Manager will always be happy to help too.
Q: What’s the best way to track a hashtag competition?
A: If it’s on Twitter, use the site: operator to track Twitter and then the hashtag. If the hashtag is very unique, you probably don’t need to use raw (for example: site:twitter.com AND thisismyhashtag). But, if it’s a word that will be used without the # often, you probably want to use raw:#hashtag. Remember that this makes it case sensitive, so you’ll need to include all the variations of the way people will write it (for example: site:twitter.com AND raw:#dog OR raw:#DOG OR raw:#Dog).
Q: How do you exclude a domain/URL in a search?
A: Use NOT and url: or site: and then enter the link, e.g. NOT site:www.brandwatch.com (not that you’d ever want to exclude our website, obviously!)
Q: Does it make a difference whether you’re using capital letters or not?
A: Only when you are using raw: which makes your search case sensitive. Otherwise, you can write you Query in whatever case you like (or a mixture) and it will match any case.
Q: Will NEAR and raw: operators ever be supported together?
A: You can use raw: with NEAR/n as long as it is used on both sides of the Query. For example, ‘raw:dog NEAR/5 raw:cat’ would work, but ‘dog NEAR/5 raw:cat’ wouldn’t. Unfortunately, the software we use does not allow the possibility of mixing raw and non-raw terms in a NEAR search.
Q: Can you limit by location?
A: Yes, you can restrict by location to the country, state, city or town level either at the Query creation stage, or with filters once you’ve saved you Queries.
More questions? Feel free to tweet us – @brandwatch
Not a Brandwatch client and wish you were? Get a demo.