The Top 17 Image Recognition Tools
By BrandwatchOct 3
How has consumer behavior changed in 2022, and what does this mean for brands?
In order to surface useful and accurate social insights and analysis about your brand, or competitors, you need to be capturing good quality and relevant data.
In Brandwatch, you find that data using Queries: the search strings used to match mentions of your brand.
If you are lucky enough to be searching for an unambiguous brand then you can often get away with quite a simple Query string that will not require too much syntax.
IKEA is a favourite example of a unique brand term that captures 99% relevant mentions with just one term in the string (the other 1% is for the Scandinavian furniture enthusiasts that use IKEA in profile names, and a song about everyone’s favourite flat pack furniture company, by Jonathan Coulton on his Album Smoking Monkey).
When searching for mentions about an ambiguous brand or a search topic however, Query writing can become quite a complex, frustrating and time-consuming task.
In the Client Services Support team at Brandwatch we play around with Boolean operators (operators used to indicate the relationship or format of different terms or parts within the query string) on a daily basis, writing Queries with a broad range of search objectives for clients and internal research projects.
Requests can be anything from standard brand tracking to broad research topics such as “how do people talk online about managing their money?” or “what are the biggest concerns about the physical signs of ageing for women over 50?”.
For us the more obscure and tricky the Query, the better!
We find it really satisfying to write a good complex Query for a particularly challenging brand name. SEAT, NEXT and MAC (cosmetics) are examples of particularly challenging brands we’ve had to write Queries about.
Through much experience and testing we have developed several techniques to help us write better Queries and make the process more efficient and enjoyable. Here are a few top tips to share with you:
Search for info about brands on Wikipedia, look at their own websites, and find Twitter and Facebook brand profile pages. These sources will provide a lot of the information you’ll need to get started with your Query, such as:
Include as many different ways that the brand can be referred to in your Query string. Remember to also include any common spelling mistakes for brand names. For example;
MarksandSpencer* OR (Marks NEAR/1 (spencer* OR spencor* OR spenser* OR spensor*))
OR MandS OR “M and S” OR “M n S”
OR raw:(M&S OR M&s OR m&s OR “M & S” OR “M & s” OR “m & s”)
Always research and test whether your brand/inclusion terms need context terms in the Query in order to capture a high proportion of relevant results.
Unlike Google searches, which use a similar structure to Brandwatch Queries, we aren’t just interested in just the first couple of pages of the results; all mentions captured are important.
If your brand name is ambiguous, you will need to link the brand terms to additional terms in order to capture the right mentions. We call these context terms.
For brand searches requiring additional context, you should include terms and phrases about about a range of topics surrounding the brand, not just products e.g. PR, marketing campaigns, corporate, and charitable activity. This way you’ll get a bigger picture of what’s being said about the brand online.
Also check for any sub brands, unique products or campaign names and test whether they need to be linked to the main brand terms or can be searched for on their own.
E.g. Apple has several sub-brands that are often referred to without a mention of the main brand: iPod, iPad, iMac etc. Unlike the main brand name, the sub-brand names don’t have any other interpretations and can therefore be included separately in the search string:
(Apple NEAR/15 (Computer* OR laptop* OR “lap-top” OR …..)) OR iPod OR iPad OR iMac
For each term you add, try and think if it is specific enough to the context, and if it could possibly also pick up irrelevant mentions.
If you are not sure how well certain words will work as context terms, test them by temporarily removing all other parts of your string (make sure you copy and paste the removed part of your string to a text doc so you don’t lose it!), then search for the term/s you are unsure about with the brand terms on their own and see what mentions are returned.
This is the end of Part 1. Come back tomorrow for Part 2, when I’ll talk about exclusions, testing and refinement. You can also read the full PDF of this guide by clicking here.
If you’re already using Brandwatch, hopefully the above tips will help you with your Queries. If you ever get stuck, please don’t suffer in silence.
There are lots of things Brandwatch can do to help: User Guides, extra Training or you can even get us to write them for you. Speak to your account manager or the support team to find out more.
Similarly, if you’d like more information on how Brandwatch can help give your business the edge in social media, just get in touch to find out more.
Keep calm and Query on!