Blog > April 2010

Analysing Vodafone Retweets

April 28th, 2010. Posted by

As part of my first week here I’ve been doing some pretty intensive training, covering all aspects of Brandwatch.  Including getting into the nitty gritty and complexities of creating query strings which I found really interesting (yes, I am a bit of a geek).  But with so many operators so much is possible; the challenge is just working out what you want to track.

We found time between training session to catch up with James, who finally made it back from Chirp, the twitter developers conference in San Francisco. He was many days late due to the Eyjafjallajokull volcano (some people will use any excuse.)  We may have let him off but we certainly didn’t give him much pity after his run ins with celebs, free stop overs in New York and Boston and the fact that he is now the proud owner of an iPad.  Like I said, I am a bit of a geek and I seem to be fitting in very well here, as I joined in the general ooohing and aaahing over the Apple’s latest and greatest gadget. And it was good to discover that Brandwatch 4.0 works really well on an iPad (I’ve already suggested we get them for demos).

After getting on with some more serious work, Seb got asked if we could track just retweets about a topic on Twitter.   We decided to try it on Vodafone, who plan to sell the iPad, and, after a little bit of head scratching and beard stroking, we realised it was actually very simple to do, by using the operators ‘site:’ and ‘raw:’.   In a short time Seb created a query, which tracks RTs (retweets) on twitter which mention Vodafone, just to test this theory out.

When we looked at the data we found a few interesting things, firstly when we looked at the content of tweets we could see that the iPad really was the big story, with Orange and O2 also showing up as they are also going to be offering the iPad in the UK, see the below word cloud for the full data:

Word cloud of content of Retweets on twitter that mention Vodafone

Word cloud of content of Retweets on twitter that mention Vodafone

But then we thought “wouldn’t it be good to be able to know who is being retweeted the most?” and with a little bit more time and some more head scratching and beard stroking, we worked out how to do this too.  With a quick data download and some simple cutting up of the data in Excel we came up with a table that showed exactly that.  The below table shows that data and the interesting thing was that although many of Vodafone’s own twitter accounts ranked highly it was in fact IntoMobile that got the most retweets, 19 in total.

Word cloud based on volume of retweets about Vodafone

Word cloud based on volume of retweets about Vodafone

If anyone is interested in how we did this or would like help to do something like this for your own brand or query then let us know.  I look forward to posting up some more of these Brandwatch titbits over the coming months.

This entry was posted in Engagement and tagged , , , , . Bookmark the permalink. Post a comment or leave a trackback: Trackback URL.
  • http://www.webliquidgroup.com/ Nate

    What was the exact query you used? Something like:

    site:twitter.com AND raw:”RT” AND vodafone ?

    I can never remember, with raw: do you have to enclose single words in quotes, or is it just phrases? Can you raw: full queries in () like with site:?

    Also, if you used Brandwatch 4.0 for either of those word clouds, I would love to know how you did it!

  • http://www.brandwatch.com Seth

    Hi Nate, thanks for your comment always enjoy getting feedback.

    Yes spot on with your query string, apart from the ? at the end which would act as a wild card, although I don’t think it would do anything in that scenario. We’ll be blogging soon about some ways to use the “?”.

    With raw: you dont have to put single words in quotes, but you do with phrases and you can put it around full queries, so all of the below would work:
    raw:M&S
    raw:”Marks & Spencer”
    raw:(M&S OR m&s OR “Marks & Spencer” OR “marks & spencer”)

    We used http://www.wordle.net, which admitedly creates prettier wordclouds than our own however they dont have as much functionality.

    Any other questions then just let us know.

  • http://www.webliquidgroup.com/ Nate

    lol yes I realized after posting that the ? at the end was going to be an issue – it was because I was asking a question about the query, but again ? is a special character. A reminder to me that should I ever want to use a search query that looks for question marks (using raw:), I’ll need to know what Brandwatch’s escape character is.

    By the by, how did you get brandwatch’s word cloud data into Wordle? I was about to write a python script to “multiply” the phrases column in the .csv output by the volume column and output the entirety of that process into a single long string. My assumption was that wordle wouldn’t be able to handle an input structure that explicitly gave the word counts. Is there a better way?

    Thanks!
    Nate

  • http://www.brandwatch.com Seth Ball

    Hi Nate,

    Sorry I should have said I used wordle advanced, http://www.wordle.net/advanced which allows you to put in a list in the format “word:number”, so you can just use excel to join the relevant columns (e.g. =A1&”:”&B1) then copy and past the new column into wordle advanced.

    Hope this helps,

    Seth

  • http://www.webliquidgroup.com/ Nate

    Excellent! Thanks!

    -Nate

  • Request Demo

    Request Demo

    See Brandwatch live in action and discover how it can give your business the edge.

    Call us on:
    +44 (0)1273 234 290

     
  • Mailing list sign-up

  •