Twitter starts to predict contestant exits from The X-Factor

Next Story

Whoa: Nintendo Sells 1.5 Million Consoles During Black Friday Week

Glancing at my Twitter stream of an evening, I’ve been surprised at the number of Geeks watching The X-Factor TV show (in the UK) . But then I guess it lends itself extremely well to witty asides on Twitter. This actually makes the show worth watching, much more for the social media discussions around it than the show itself. It’s like being at a football match where the crowd’s chants are more entertaining than the game. If there is a business model for bland, manufactured TV I guess this is it.

So we make no apology for passing on the news that social media monitoring company Brandwatch is claiming that it can predict who is about to exit The X-Factor TV “musical competition” based on what’s being said about it on Facebook and Twitter. I asked what else they track, but sure enough, Twitter tops the list as a data source.

As Brandwatch points out, while the Bookies odds are on contestant Matt Cardle, it’s monitoring has found that boy band One Direction have been at the top of the Brandwatch chart for a few weeks now. They say they predicted the ousting of Katie and Wagner hours before the results were released on Sunday’s live TV show.

Brandwatch says they have been monitoring volume and sentiment of mentions on Twitter, Facebook, blogs, forums and news sites – but I daresay startups like Datasift would do this, and I’m not wholly convinced by Brandwatch’s numbers. [ Update: A clairification: These are not the total overall mentions in a week, but the top sites out of the 6,084 sites Brandwatch tracks.]

Here are the top mentions for Cher Lloyd in the last week according to Brandwatch. 2261 311 77 53 46 37 37 33 28 25
Total for top sites 2908

  • Jason Trost

    I think real-time, real money prediction markets such as ours ( are far bettor indicators of who will win rather Twitter mentions of a contestant. There is so much noise in social streams it’s hard to extract meaning from simple instances of a name. Early search engines use to score pages by the the frequency of words in each page until Google changed that with PageRank.

    • Reader

      Conversely, prediction markets might have access to a small and potentially biased sample that cannot reproduce the value of the broader Twitter audience.

      • Jason Trost

        Prediction markets are not a panacea; they are subject to biases as are all tools. However, as a tool, research shows they can be quite accurate even with small sample sizes because real money focuses traders’ activities.

      • Azeem

        BTW if you want my prediction on this based on the youtube data at about 3pm :!/azeem/status/8896762311872513


      • Azeem

        is my prediction from 3pm Sunday baed on the XFactor/youtube channel.

  • Jenni

    The numbers do seem extremely low, particularly when you take into account that there are Cher Lloyd groups & pages on Facebook with more than 100,000 odd members and comments posted every few minutes.

  • Giles Palmer

    the facebook numbers are for the public facebook pages – ie you don’t have to be logged into it so yes, the fb numbers are low – the others should all be right


  • Giles Palmer

    oh, and they’ve been filtered by english language only

  • Bobbie Johnson

    Less than 3,000 instances for the top ranking contestant in a TV show watched by 15 million people? Whether or not you can interpret action from that data is one thing, but it seems like a tiny sliver of the data that must actually be out there.

    Example: Google suggests more than 20,000 mentions on Twitter alone for “Cher Lloyd” in the past week, and 115,000 mentions of the hashtag #xfactor.

  • Adam Green

    Using Twitter as a predictor of popular opinion is finally being recognized. A couple of years ago I built a now-defunct site that tracked tweets for every US stock symbol. It only did peaks in activity coincide with price moves, in many cases they led the price move by up to 24 hours.

    The big problem, of course, is filtering out attempts to game this phenomenon. The Twitter API makes it easy to construct bots that reproduce any type of user behavior.

  • Jayne Coulthard

    interesting. we analysed 1/2million+ xfactor tweets, but then did contextual analysis – to ask why they were bottom 3 and to dig depper than sentiment analysis can ever do. brand aura contextual analysis doesn’t list the important words, the data decides. it brings out interesting issues surrounding brands. and helps answer why brand advocacy doesn’t always convert to sales.

  • Azeem


    Youtube views on thexfactoruk have been phenomenally accurate for divining the bottom 2 or 3 candidates. normally by sunday am there is enough data to make a guess.

    correctly predicted katie, wagner and mary this week, correctly guessing katie in bottom place

    apologies for not pitching my companies like the smarkets guy.

  • Is Twitter Predicting Contestant Exits From The X-Factor? | JetLib News

    […] Read the rest of this entry » […]

  • Chris

    The following that One Direction have is getting bigger by the day. Bookes currently have Matt Cardle as odds on favourite, but the clever money is backing Simon Cowell’s boys to take the crown.

  • Boon

    Its with great interest that I see someone else coming up with a model based around Twitter for predicting reality TV show winners.

    I used a similar model I built myself for Britain’s Got Talent earlier this year. I must say that it is quite a good predictor – although one has to adjust for demographic, as Twitterers are not representative of the whole voting demographic of reality TV shows.

    I find that in order to get a good prediction, you need to get the immediate live feedback and sentiment scoring of the tweets right after the show and when voting takes place. The above model by Brandwatch only takes into account what happens between episodes. What happens during the episode can have quite alot of bearing on the final voting patterns.

    • Giles Palmer

      yes, fair point — must look into that….

  • SecrDig
    Here you can say your secret out,

    secrets will be automatically announced to Twitter,

    but they are anonymous…xD

  • Matt

    Boon is right, you do have to factor in demographic. A quick search of Twitter would tell you that One Direction obviously would have the most active group of Twitter followers (call it the Justin Bieber factor if you will) followed by Cher, etc etc as opposed to someone like say, Mary. One Direction could easily mobilize all of their fans and win the competition if they had a viable social media strategy.

    The real question is how does all the Twitter chatter actually translate in to actual phone calls aka votes.

  • Paul Simpson

    I suspect it’s missing negative comments about One Direction because they often refer to the group with a differently spelt but identically sounding “Wand Erection.” Add that to the mix and I think we’d have a different picture.

  • Mobile & Social TV- War of the app platforms « @CharlotteClark

    […] Twitter starts to predict contestant exits from The X-Factor by Mike Butcher […]

  • Twitter starts to predict contestant exits from The X-Factor | Brandwatch

    […] We make no apology for passing on the news that social media monitoring company Brandwatch is claimi… This entry was posted in Press. Bookmark the permalink. Post a comment or leave a trackback: Trackback URL. « London Bike Hire Scheme, Boris’s Bikes […]

  • Admin

    Twitter is not as well as facebook now

  • Social media and The Apprentice | kitcattnohrdigitas

    […] relationship between TV talent contests and social media on the face of it, appears to be more significant. People talking about a show such as The X Factor also have the opportunity to vote for their […]

blog comments powered by Disqus