Can social media be used to predict the X Factor Winner?

The X factor 2012 final to be aired across a pair of two-hour shows on 8-9 December.At Brandwatch we’ve decided to turn our hand to a little online palm reading. With three weeks to go before the X Factor final, we wanted to see if we could predict the final results. Even if you’re not hooked on James Arthur, read on; you might just get some ideas for how to use a social media to benchmark and predict your brand’s success.

Last week we ran some analysis on the online buzz of the five quarter finalists, and then sat back to see if our predictions came true. Some background for those of you who didn’t catch the results show this weekend: Rylan Clark and Union J were in the bottom two after receiving the fewest public votes, and Rylan was then voted off the show by the judges.

Want to know your future? Learn from your past

We chose X Factor to demonstrate how you can use social media data to predict the future because there’s plenty of data we can learn from. When working with social data, benchmarking is paramount – how can I know what good looks like if I haven’t seen bad?

We studied the last month’s X Factor quarter finalist discussion and compared it with the five acts that made it through to last weekend’s stage in 2011, looking at exactly the same period of the competition.

Buzz volume alone is not enough to predict a winner

If we look at discussion volume for the five acts leading up to the quarter final, we can see that James Arthur was by far in the lead, with his greatest competition for the title coming from boy band Union J.

However, if we look at last year’s buzz from exactly the same point in the competition we can see that the buzz leader then was Amelia Lily (who?!) and Janet Devlin, who was given the boot that very week.

Our eventual winners Little Mix were languishing in third at that point, so it looked like all of this year’s acts still had everything to play for.

But that’s okay; we can look at much more than just volume…

To take this one step further, we took a look at which of the acts from this year and last year were most associated with winning at this stage in the competition.

We looked at anywhere that terms associated with “winning” appeared within 5 words of each of our acts. Since we’ve already established that volume isn’t everything, we’ve worked out each act’s winning buzz as a percentage of their total buzz to make a fair comparison.

Winning Buzz before the Quarter Final Show:

The results are in:

When we looked at who was most associated with winning the week before the quarter finals last year, suddenly we saw our winners Little Mix at the top and poor old Janet in the relegation zone at the bottom.

In fact, for those of you who don’t remember last year’s running order in perfect detail, this winning buzz chart displays the quarter finalists in exactly the order that they were eliminated from the competition (bottom to top).

Ranking this year’s acts in the same way, we were able to accurately predict which two acts were in the bottom two from the public vote, two days before the live show. Of course, the buzz ranked Union J lower than Rylan Clark (who went home on Sunday), but once the bottom two are chosen it’s the judges who decide who leaves, meaning that buzz tracking could not predict this part of the show.

So who’s our money on to win?

If you’d have asked us last week, our money was squarely on James Arthur, but there have been some interesting developments in the last two days that are starting to make us wonder if Christopher Maloney may cause a stir in this year’s competition.

Controversial contestant Christopher has avoided being in the bottom two, despite consistently gathering criticism from the judges, as he’s been saved each week by the public vote. After being saved yet again on Sunday, speculation is rife that X Factor saboteurs are at work once more (remember the “Rage Against the Machine” Christmas number one plot?). The idea is to discredit the franchise by voting for the less talented Christopher to win, something which Simon Cowell is keen to avoid.

If we look at each contestants winning buzz percentage for the last two days, the picture looks very different:

Winning Buzz: After the Quarter Final Show

So, will the campaign to bring Christopher to victory prevail, or is it all just hot air? We’ll be keeping our ear to the buzz closely over the next two weeks to see if we can figure it out. Stay tuned!

  • http://twitter.com/JohnFell7 John Fell

    Interesting read. I did something similar using Brandwatch a couple of years back for Britains Got Talent. It was pretty accurate for all the semi finals. For the final it didn’t predict the winner as being Jai (Not sure of what his surname was, his star seems to have fallen since), but it predicted a much younger contestant. The important thing for me was that the people on social media were not necessarily the ones voting – probably because they aren’t the ones who pay the phone bills…

  • Joel Windels™

    Thanks John – here’s the post John is talking about: http://www.brandwatch.com/2011/06/brandwatch-predicts-the-winner-of-britain%E2%80%99s-got-talent/

    We’ve also successfully predicted things like the US election and the London Mayoral election, though with votes such as the X Factor, we find the correlation between tweeters and phone voters rather close.

    Here’s a post on the value of social polling for elections: http://www.brandwatch.com/2012/11/social-media-opinion-polling-should-be-considered-as-valuable-as-offline-equivalents/

    Joel

  • http://twitter.com/rachelhawkes Rachel Hawkes

    You mention that you looked at ‘terms related to winning’ within five words either side of the contestants name, but can you clarify whether the context of the post / tweet was considered? For example, before the semi finals James Arthur had the most ‘winning’ related buzz, but post the show that was Christopher Maloney. But words like can’t, shouldn’t, never etc., may have also appeared alongside the word win or winning, thereby changing the context entirely.

    At the start of the live shows last year, we decided to look at the same thing that you have done… whether social media could predict the winner of The X Factor http://bit.ly/rY7cW3 (published here on The Wall http://bit.ly/sLZIuT). We focused only on rudimentary figures such as Likes, Followers, Views et al and how they changed each week (electing not to use a measurement tool). As we wrote at the time and as John has rightly said, the social media audience are not entirely indicative of the voting public, and not all Followers / Likes are supporters – but it is one quick and easy way of getting the “lay of the land” so to speak.

    Just looking at crude numbers, which by themselves can only ever be an indicator as you can’t determine sentiment and intent just by looking at numbers (and why as marketers we should all be advising our clients to move away from this obsession with Likes and Follows), we were able to correctly predict the winner weeks ahead of the final.

    You make a really excellent point in that benchmarking is hugely important. We have to have a consistent measurement of data over a period of time so that we can best identify trends, good and bad periods and take action as appropriate.

    Looking forward to your clarification on the verbs alongside “winning terms” and whether that was factored into the buzz measurement. As a huge James Arthur fan, I hope there isn’t that much buzz about Christopher Maloney winning. Thanks for an interesting read.

  • Rebecca Campbell

    Hi Rachel,

    Thanks for the great question. You’re totally right, the context of the post could change the meaning as to whether the author wants that act to win or not to win. However, for this study I decided not to restrict by context and look at a macro level at whether there was an association between each act and winning, whether or not that association was positive or negative. The idea behind this was to see if just being talked about in the context of winning the show would give us a proxy for which acts were most likely to make it to the end (and test the old adage that any publicity is good publicity!)

    My decision to use this methodology was also informed by my previous experience of tracking X Factor buzz in 2010, when I noticed that association with winning was more telling about the final result than volume/sentiment. In 2010 Matt Cardle won the X Factor, but in terms of volume of positive online comments One Direction were far ahead of him throughout the competition. However, Matt Cardle was much more associated with winning in online conversation, especially in the last few weeks. This suggested that the huge following of (mostly) teenage girls that were raving about One Direction online weren’t actually voting (as you and John have rightly noted, just because you’re talking online doesn’t mean you’re voting, and vice versa). Again when I looked back at 2011 for this article, association with “winning” in online conversation actually provided a better indication of the voting than mention volume or sentiment. It will be very interesting to see if it plays out the same way this year, especially with Christopher throwing us a bit of a curve ball!

    With two acts now leading in “winning” conversation for two very different reasons, I think you’re right that we’ll need to look further into the context of the conversation over the next two weeks to get a better idea of who will come out on top. Also, without a precedent to benchmark against in this situation, it will be more difficult to predict how what people say online will translate into voting behaviour on the night. Really interesting stuff!

  • Pingback: Writing A Blog Tips That Can Save You Many Headaches! | Wholesale Retail Marketing Blog()

  • Pingback: What Is Offered By Website Design Leeds How Will It Improve Business? | Wisdom Health Prosperity()

  • Pingback: Can social media be used to predict the X Factor Winner? | Betteridge’s Law()