Is Twitter really ‘The Voice’ of the people?

the voice logoIn this guest post, Hugh Stephens analyses the conversation on Twitter during The Voice and asks whether there were more opportunities for television integration.

The Voice has certainly been one of the shining stars of Nine’s repertoire of late.

With this week’s final, I thought it would be worthwhile to have a quick look at the tweets to see how they changed over the course of the program.

I collected some 38,000 tweets to analyse. There was a distinct peak around when the winners were announced.

A quick look at the wordcloud will give you an overview of the content:

the Voice wordcloud

Winner Karise Eden was consistently overshadowed in frequency compared to Sarah DeBono, who came in fourth place. Considering that DeBono has been consistently Tweeting, Instagram-ming and generally keeping in close touch with her fans, it’s not a surprising result. More surprising (to me) is that she hasn’t converted that social interest into votes… an ROI begging to be measured (if only the votes were transparent!).

Not surprisingly, there were over 17,000 people who tweeted about the program over the two hour period that the show was most discussed, with the tweets very quickly tailing off after 11pm. If you believe impressions, over 29.8 million impressions (1.8 million of which were from NRL player Quade Cooper, 1.5 million from the show’s own account) to 5.1 million users were measured: a pretty impressive number for a very short time period.

An interesting comparison (certainly for me) is the results from the URL with the greatest number of impressions. With over 430,000 impressions (and second highest number of individual tweets), the URL to a duet between Keith Urban and mentee Darren received a sum total of only 453 clicks – a pretty low conversion rate.

A quick-and-dirty sentiment analysis found a slightly more positive spin – 3618 positive sentiment tweets compared to 2868 negative. But I couldn’t say much more than that without a more indepth analysis…

All in all, a pretty impressive collection of data for the producers at Nine, yet an averagely-executed digital strategy for one of the most-watched shows this year. I don’t think there was enough digital-only content developed, and not a lot of effort about engaging and developing a community around the program.
Additionally, it would’ve been good to see some more engagement from the top 10 finalists – an opportunity for some training and education as part of building their own personal brand as well as developing that of The Voice.

What should happen next time? I’d start with 4 things.
1. Develop more individual presences for the finalists. Twitter / Facebook chats / Q&As, individual accounts with the ‘experience’ day-in, day-out of the competition and preparation, behind-the-scenes photos and videos shot by the finalists etc. Teach the ones who don’t know how to use the common tools – Facebook, Twitter, Instagram, possibly even Pinterest.
2. More digital content. It’s not enough to just post snippets of the show that was aired. What about the rehearsals? The crowd reactions? It’d be great to have the judges answer some viewer questions via YouTube.
3. Online voting. It’s about time that shows like ‘The Voice’ start to embrace online voting. While it’s open to more bias, a solution can be created to allow people to vote via their Facebook or Twitter accounts – not everyone wants to use premium numbers.
4. Be more responsive to the community. What do they want? What would they like to see or hear? Who do they want to communicate and how? The Voice was amazingly popular, and as the show continues, a community needs to be fostered in order to avoid viewer burnout.
Hugh Stephens is a director at social media agency Dialogue Consulting.


  1. J
    20 Jun 12
    11:29 am

  2. Love that tweet from ‘enter quote author’

  3. Hugh
    20 Jun 12
    11:37 am

  4. Hi J – shouldn’t be enter quote author – works for me and at! –> it’s a beta service so might be a tad buggy…

  5. Ricki
    20 Jun 12
    11:50 am

  6. One correction. The Voice did allow online voting…via their FB page and website. How this changes the numbers I’ve no idea.

  7. Reality Raver (Emma Ashton)
    20 Jun 12
    12:03 pm

  8. Interesting and informative article. Thank you.

    However just so you know there was Facebook voting for The Voice. The first time this has been done by a reality TV show, I expect there will be more of it.

    What I would find interesting to know is the demographics of each voting portal ie Facebook, SMS, and via itunes.

  9. Simon
    20 Jun 12
    12:35 pm

  10. Out of curiosity, how did you acquire all the twitter data?

  11. Hugh
    20 Jun 12
    12:45 pm

  12. Thanks Emma and Ricki – hadn’t noticed the FB voting (maybe they should have promo’ed it harder!)

    Re how we collected – combination of methods. I dont like hootsuite’s archive function as I find it quite ineffective. We use a combination of existing self-installed tools that we’ve modified, Yahoo! pipes JSON/Atom feeds and a few other things if necessary. Our Amazon cloud-hosted tweet backup / archive tool is usually the best as it scales well for times like this – need to be able to capture a lot of data quickly!

  13. DD
    20 Jun 12
    12:53 pm

  14. Quade Cooper plays rugby union, not NRL.

    How do you get 1.8 million imps? Does tweeting 3 times during the show when you have 600K followers, really equal 1.8 million? Smells like BS to me.

    I follow but didn’t see the tweets. And on the flip side, what about the people who see tweets based on hashtag and name searches…

  15. leisai
    20 Jun 12
    12:54 pm

  16. I thought similar analysis was done of American Idol which concluded that whomever was tweeted about the most (irregardless of positive or negative) was almost always the one who was voted out that week…. Which would support the fact that DeBono was the most active and was placed fourth…

  17. Carolyn Hyams
    20 Jun 12
    1:01 pm

  18. I thought the social integration was great, but your tips for improvement are excellent. I hope their marketers are reading this.

    What you don’t mention is their very active presence on Facebook. They were certainly updating their page all the time with impromptu (if there is such a thing in marketing) videos, photos, interviews etc. You could also vote via their Facebook app. Quite frankly, between watching the tv screen, the twitter stream and facebook, it was a frantic time trying to catch it all.

  19. bob
    20 Jun 12
    1:04 pm

  20. Gee aren’t you jumping on the bandwagon in an uninformed way?!?! Did you look at the volume of behind the scenes video content on the show’s website, or the amount of content posted on facebook during the show?! The volume of likes and comments on these posts were huge

  21. Hugh
    20 Jun 12
    1:19 pm

  22. DD – Victorian here…so Union and NRL get confused in my mind at times.

    Re impressions, calculated as you suggested. It’s certainly not a hugely accurate number – like most SM impressions (excepting FB Reach) you don’t necessarily know if all those accounts are online – or even ‘real’ (eg just trying to sell you things to make you smaller or bigger). It’s a sum total ‘potential’ calculated and certainly a murky score. And yes – doesn’t take into account hashtag/name searches, so possibly some balance?

    While not a brilliant estimator of actual reach, IMO it’s better than (or at least useful as well as) the number of unique people tweeting – because 1000 people with 10 followers each (mean Twitter followers) doesn’t give you an indication of how many ‘influential’ people are tweeting about it. And social is all about engaging the key influencers.

    Leisai – saw that analysis but to me it seems a tad backwards. How the volume of tweets impacts voting depends a lot on sentiment – and unless you go through and hand-code the sentiment you often end up with an average quality result. Would love to test the hypothesis properly by running the analysis over the whole show’s duration… one day ….

  23. Hugh
    20 Jun 12
    1:24 pm

  24. @bob – there was some content but not much of it was overly ‘personal’ – these shows are all about the experience and relationship of the viewer…and by far most of the content was promo material or clips from the show. There were a couple of Google Hangouts (who uses G+ though?) which was at least a step forward… And yes, there was more content on FB than there was on Twitter/YT… I would’ve still liked to see more personality from the people themselves (it’s about the country falling in love with their fav singer) than only trying to build the voice brand

  25. Hugh
    20 Jun 12
    1:27 pm

  26. @Carolyn: agreed that it was pretty good…and their FB page is great (quite a young audience though – I’d love to see how it compares to viewer demographics).

    I also agree a lot about needing to centralise the FB/Twitter activity! We’re actually developing a tool like TwitterFall (twitter monitoring) that pulls twitter & FB data, for some of our event/conferencing clients….so there might be a tool out there *eventually* that would help out…

  27. Al
    20 Jun 12
    1:53 pm

  28. Same as everyone else; not the best “consultation” I’ve read… how well researched were your conclusions?

    1. All artists have Twitter pages and tweets from the main Voice account were constant. There seems to already be a Voice account on all the main platforms (Facebook/Twitter/YouTube)… even G+).
    And a quick Google search found the official Pinterest page:

    2. Lots of behind-the-scenes content online via Tw/Fb/Yt

    3. Facebook voting was plugged by the host guy every week!

    4. Reading through Facebook, their community is already active and engaged in a 2-way convo with “the Voice”. With auditions open for next year already, I think the traffic will remain.

  29. bob
    20 Jun 12
    1:56 pm

  30. have you actually looked at and watched any of the behind the scenes video content?!

  31. Floyd
    20 Jun 12
    2:12 pm

  32. So really you’re just trying to flog a tool for social media monitoring and engagement? There are plenty of these in market – and the only thing in common with them all is that their conclusions vary greatly – especially sentiment!

  33. John Grono
    20 Jun 12
    2:18 pm

  34. I know of another method. Watch the show and wait for Darren to read the name from the envelope. Quite an enjoyable methodology and surprisingly accurate.

  35. Hugh
    20 Jun 12
    3:21 pm

  36. @Floyd : flog our tool we are not. We don’t sell it – its a personal experimental version of yourTwapperKeeper (if you know it) modified to work and localise searches. YTK is a free/open source tool – i discovered it through the old service. The guys at ANU have done some great research into sociological networks on twitter using it and gawk – their analysis of the bush fires is well worth a read.

    Sadly I wish we had a fancy expensive tool, but its just modified open source tech, an installation of R an basic text analytics.

  37. Tarnya
    20 Jun 12
    4:27 pm

  38. @John Grono: wish there was a LIKE button for that comment.

  39. Angie
    20 Jun 12
    4:54 pm

  40. Winner Karise Eden was consistently overshadowed in frequency compared to Sarah DeBono. How do you measure Karise’s over shadowing ?

  41. Uncle Argyle
    20 Jun 12
    6:34 pm

  42. This is pretty personal. Joel Madden thanking a super fan for their support during the show!

  43. Hugh
    20 Jun 12
    7:50 pm

  44. @Angie: word frequencies in the tweets…

  45. Brandon Hillier
    21 Jun 12
    9:25 am

  46. …”and unless you go through and hand-code the sentiment you often end up with an average quality result”

    exactly. Social media monitoring tools are a giant load of BS.

    Volume doesn’t reflect intention.

    People cling to metrics such as number of followers and ‘Likes’ and determine success or failure. It’s largely useless.

  47. KP
    21 Jun 12
    10:13 am

  48. Hugh, your suggestions (outside of your lack of noticing the Facebook interaction) are certainly valid and I assume with 2013’s production these points and more will be covered for a richer interaction with the audience and viewers.

    Like many more will in the near future, I engage in Social TV for the humorous activity (although I cannot keep up at times) and certainly believe Twitter helped to raise the profile of The Voice however giving the contestants more of a personality during on-air time (not just in the V Room) would have assisted the individuals’ brand, for post-show movement. I think they were blocked by the host there doing her job for Vodafone.

    As for the winner and how social media worked or did not work, it’s common knowledge the underdog more often than not makes it to Number One on the night.

    Overall, impressive television viewing with the Twitter crossover — almost as good as Q&A.


  49. leisai
    21 Jun 12
    2:30 pm

  50. hugh: there is another article that demonstrates it throughout that entire season of Idol (sorry, couldn’t find it since it was published a over a year ago). Every week (except two) it was true

  51. Angie
    21 Jun 12
    3:14 pm

  52. Great comment KP! I follow and Tweet during Q&A. Can’t wait for some kind of dual screen for TV driven social interaction in the near future.

  53. Craig
    29 Jun 12
    6:49 am

  54. Nice analysis BTW, however by analysing only one channel, and by dint of the headline this piece has been given, the article becomes a little sidetracked.

    Why should there be a single channel that is the ‘voice of the people’?

    Everyone in the media and communications industry recognizes that media has fragmented.

    When top TV programs in Australia attract 3 million viewers this is considered large – and it is in a strict numerical sense. However by percentage it represents a far smaller share of ‘the public’ that in past decades.

    With the fragmentation of media comes a fragmentation of voice. The public aren’t all on Twitter, or even Facebook (which has roughly 7x as many subscribers in Australia as Twitter).

    Analyzing one channel tells us interesting stuff about that channel, shaped by its specific attributes, however doesn’t give a broad view of the ‘voice of the people’.

    I would be interested in seeking how the Voice contestants fared on Facebook, where (and which) voting took place, what blogs and forums were saying.

    Cross-channel analysis is likely to reveal more of the jigsaw puzzle of consumer voice. This Twitter analysis is useful and interesting, but don’t mistake the piece of the puzzle for the puzzle itself.

    BTW I like Twapperkeeper as well and also use services such as The Archivist and commercial social media analyst tools.

    However in my view no-one has yet cracked the market for providing excellent cross-channel social analysis tools. I keep hoping someone will :)