Using Twitter and word clouds for competitive intelligence

  Social media in my work is an everyday activity such as connecting with clients and colleagues through LinkedIn, sharing interesting websites or news through Twitter, or reading useful blogs online. For me, it's now a normal part of my day on top of the usual activities I do to earn my pay. This year, however, one of my objectives is to research social media and its potential uses within our team, and to implement a strategy based upon this investigation.

  A particular challenge with understanding social media is interpreting what is being said about your business on Twitter. There are so many tweets occurring every minute of the day that you need to tame this flood of information. Then you need to present the data in a way that people can grasp quickly.

  One particular idea I had to help make use of the flood of information was to use 'word clouds'. You often see 'tag clouds' on blogs or websites which highlight which tags are being used on blog posts more often, for example.The more often a tag is used to describe what's contained in a b log post, the larger it appears in the tag cloud. It's a quick and easy way to see topics you are interested and which ones are written about more often.

  Word clouds are the same principle and I decided to make use of these to help people quickly grasp what is being said about our business on Twitter and also what is being said about our competitors.

But, before creating a word cloud you need to capture the tweets. I used a desktop tool called 'The Archivist' which you can download for free. Once installed, you type in a keyword into the search box and it will search for tweets containing that keyword or hash tag. It then lists the tweets and allows you to export as an XML file or an Excel spreadsheet.

In the spreadsheet, you can then copy all of the tweets, URLs included, into a Word document to strip out the keyword you used in the search so that it does not completely dominate the word cloud you are going to create. Next comes creating the word cloud. Copy all of the words on Word document (without your keyword).

I used 'Wordle' to create the word clouds. In Wordle, you simply go to the 'Create' webpage and paste the copied words into the box below the instruction 'Paste in a bunch of text:' and then click the 'Go' button below. Next, you will see the word cloud which you can customise to an appearance you like. 








Here is the word cloud I created using the keyword 'proquest' which I picked from tweets on 15 July 2011 at 1.14pm. There were 123 tweets. 

What stands are the large 'RT', 'via', ''AddThis' and ''ProQuest'. So, what does that mean? 'RT' is twitter-speak for re-tweet which is when people share a tweet with their followers. 'via' is a term usually used when people share or retweet information too. '@AddThis' is another sharing service on the internet. '@Proquest' shows the people are sending tweets directly to our business. 

Re-tweeting of your tweets is what you are hoping to achieve. When people re-tweet your information it means it has gone beyond your followers to their followers. This is great. 

So, what about our competitor's word cloud? Here is the Ebsco word cloud taken from 129 tweets at midday on 15 July 2011:

The words which stand out here are 'AL' (which is taken from a hash tag), 'Discovery', and 'Chrome'. The AL hash tag goes with the other hash tag on the page, 'ALjobs' which comes from a jobs page on a website about Alabama, USA. So, they are hiring there! 'Discovery' and 'Service' go together. This is a new niche which our competition are fighting for against us with our 'Summon' product. When I looked at 'Chrome' in the tweets, it shows that customers using the Google Chrome web browser are unable to use the Ebsco services with it. There's a fix on the way, apparently. 

There is more useful information available to indicate the language used in the tweets which helps to indicate from which country the tweet was sent. 

Although, this is not scientific, it is a very visual way of understanding feelings, thoughts and perceptions about your business or brand. It did not show some of the really negative tweets in the sample, such as 'F**k Ebsco' or 

'I know he's at home all week, but why is he muttering "f**king ebsco useless f***ing muppets couldn't write a working hello world script". On ProQuest, the negative comments were more like this one 'Is it just me or is the new ProQuest platform incredibly buggy from the back-end...? #reallygettingannoying'.

Nevertheless, using word clouds with Twitter data is simple way to convey sentiment about your brand.