Archive for the ‘Analytics’ Category
June 15, 2010
Within an increasingly sophisticated argument about digital piracy and the motivations for stealing intangible assets like music and film, there is a growing number of research projects focused on the question of software piracy and sales displacement. In other words, what does the data, rather than vested interests, have to say about the potential loss in revenue as a consequence of software piracy?
The most recent study by Joel Waldfogel at The Wharton School, University of Pennsylvania, concludes by quantifying the displacement effect at between -0.15% and -0.28%, depending on the artist, genre and importantly, the music interest (music sophisticate versus mainstreamer versus lifestyler versus influencer) of the individual in question. That is, an additional stolen track reduces paid consumption by between a third and and a sixth of a song.
The methodology included sampling university students and their music consumption. The incidence of illegal music was obviously high, but so was the legal version of the same track. Interestingly, the first insight of the study revealed a definite, positive correlation between the percentage of students who owned a legal copy and those who possessed an illegal version.
For example, of the 50 tracks used in the survey, Coldplay’s Viva La Vida had the highest incidence of ownership, with 24.5% possessing a legal copy versus 17.5% who had the shared copy. Lifewise, the legal version of M.I.A’s Paper Planes was 17.7% versus 22.5% for the illegal option, and so on. Interestingly, the average valuation placed on Viva La Vida by the students was $2.22 – well above the iTunes retail price of $0.99 (the consumer has achieved a real, legitimate surplus in this instance). Correspondingly, the average value placed on Miley Cyrus’ 7 Things was just $0.36.
On this point, but beyond the scope of this paper, much needs to be discussed about current pricing structures of digital entertainment products and their real market valuations (as determined by consumers). There is definitely room to discuss the deadweight loss of certain music offerings which are simply being priced well beyond consumer expectations.
To conclude, we have an instance here where consumers purchase an average of 5.54 songs and steal an average of 6.71, while purchased songs have an average valuation of $2.76! Based on a displacement figure of -0.28, and in the context of no illegal music filesharing, the number of purchases would rise to 7.42 songs, not the 12.25 songs (5.54 + 6.71) in a simple one for one scenario.
The fascinating aspect of this research, and subsequent analysis, is the implications for all digital entertainment offerings. While we assess the implications for digital music sales, we must also be aware of the implications and learnings for filmed entertainment and associated distribution options.
January 31, 2010
“will be celebrating Australia Day with some gardening, a barbie, drinks & a swim with friends and tripleJ hottest 100 playing in background.”theother66, Twitter, January 25th
“They really f**ked it up this time. Can’t believe the effect that spoiling the Hottest 100 is having on youth around me.” Downesy, Twitter, January 22nd
“Did I mention that I am disappointed with the @triplej hottest 100? The top 10 especially offends me. And probably gave me herpes too.” Mversion, Twitter, January 26th
In 2010, the “world’s biggest music democracy”, Triple J’s Hottest 100, smashed all previous year’s efforts with an incredible level of listener engagement. More than 131,000 individuals cast 1.1 million votes for what its audience considered to be 2009’s hottest 100 tracks. Both the number of voters, and hence votes, were up 30 percent and 46 percent respectively.
Passive listening is one thing, but to vote and use the whole program as the basis for celebration on Australia Day – 4,100 Hottest 100 parties is not an apparition – is a testament to the public broadcaster and the eclectic tastes (and passions) of Triple J listeners.
Celebration and criticism of the final rankings aside, is the Hottest 100 more than just a gang-up of Muse and Art vs Science fans to game the results? Is the list a potential bellwether of illegal music consumption and the popularity of those tracks? What would a Darkest 100 look like?
Tracks in this year’s Triple J Hottest 100 accounted for more than 17 million illegal downloads in the past four weeks; that’s out of a total of more than 35 million in a ‘usual’ four week period. An impressive ‘market share’ for a public broadcaster!
While Mumford & Son’s Little Lion Man topped the Hottest 100, the track was ranked fourth in the illegal chart, with 698,528 downloads. The top spot in our Darkest 100 was taken by La Roux’s Bulletproof with 1.3m illegal downloads, up five places from its Hottest 100 ranking. By comparison, Hilltop Hoods’ Chase That Feeling came in third on both charts (760,312 downloads).
The tracks swinging wildly between their Hottest 100 spot and the illegal chart included Calvin Harris’ I’m Not Alone, ranked fifth in the illegal download chart (626,142 downloads) - a massive 54 places up on its popular vote (ranked 59 in the Hottest 100) - while Foo Fighter’s Wheels, ranked last by the Triple J audience, came in at number 12 in the Darkest 100 chart (452,583 downloads). Jet’s She’s A Genius jumped 62 spots, from 77 in the Hottest 100 to 15 in the illegal ranks (412,332 downloads).
The high profile results of Unearthed bands in the Hottest 100, however, weren’t exactly replicated on the Darkest 100. Art vs Science, coming in at number two on the Hottest 100, dropped 53 places on the illegal chart, accounting for just 75,048 downloads. Despite the drop, the band is still number one out of all the Unearthed bands when it came to downloads. Seth Sentry accounted for 66,880 downloads, a big drop of 29 places on the Darkest 100, Philadelphia Grand Jury accounted for 37,882 downloads (down 83 spots) while The Middle East had 24,737 downloads (down just 6 spots). Interestingly, Art vs Science’s Friend in the Field ranked 73 on both charts.
In terms of music type, the Darkest 100 is better suited to dance, rock and hip hop, while the indies were, to put it bluntly, murdered by the alternative chart. On average, indie tracks dropped 25 places on the illegal chart, while bands with more commercial radio play, tended to score higher positions on the Darkest 100.
The Hottest 100 is a testament to popular culture, a love of great music and a countdown, in more ways than one, of how modern Australia chooses to consume music.
Source: PeerIn-Big Champagne Australia 2010
August 12, 2009
In the context of marketing and sales, the concept of ‘engagement’ carries increasing gravitas, and yet is still difficult to conceptualise through a single definition. Quantifying the blighter is harder still. The fascination with engagement stems from its potential to make-up for any shortfall in customer contact a brand may have in the areas of reach and frequency. What a brand fails to achieve in market share or regularity of contact (or feedback), the brand might compensate through share-of-mind. In the case of mobile phones, market share (unit sales) is dominated by the likes of Nokia and Samsung, with Blackberry and iPhone well behind. The put the quantity of sales into perspective, Samsung Australia sells approximately 3m units a year (second only to Nokia), compared with the 400,000 iPhones currently in operation in Australia (just under 4% of the market). Yet, in terms of digital chatter, comprising opinion, advice, questions and general consumer feedback, there is a massive discrepancy between brands, this time favouring Apple and Blackberry handsets. In short, what these brands lack in market share, they more than compensate for in terms of communities openly engaging in a dialogue, sometimes quiet passionately, about a mass-produced consumer item. The Sentiment Index graph below illustrates the scale of digital chatter between brands, with the iPhone alone having a 10-to-1 advantage against most competiting brands. By quantifying engagement in this manner, does this incentivise brand managers to do the cost-benefit of investing in an engagement strategy ahead of traditional reach and frequency filters? 
Source: Victrix Media
August 10, 2009
The clean, crisp lines demarcating the radio stations we listen to, or the television programs we watch on free-to-air, tend to blur and mutate the more our media behaviour migrates online. In other words, taking a screeching left off the autobahn, the viewer becomes immersed in a very dynamic, less predictable media-experience.
What compounds the issue of unpredictability, and at times confusion, is the lack of comprehensive data on comparative behaviour across websites. In other words, online media consumption is still overwhelmingly assessed by singular points of ‘engagement’ rather than in the context of the person’s repertoire of online publications and applications.
In the context of online music audiences, the plethora of music-related websites makes the measurement and understanding of engagement even more acute.
In Australia, our understanding, let alone analysis, of these online eco-systems, or repertoires, is very limited, despite having evolved through a combination of habit, relevancy and external promotion. We know, for example, that just over 390,000 Australians visit The Pirate Bay in a single month, and that about 226,000 will visit the Take40 website, but in the context of an online entertainment category with a reach of 9.4m Australians (source: Nielsen Online), where individuals spend on average three hours a month in the category, are these audience figures respectable, impressive or positively underwhelming?
Each month, Australians visit more than 450 entertainment-type websites, and this ignores the plethora of social media, sports and news sites which tend to leverage the entertainment dollar. Given this level of fragmentation, a monthly audience of almost a quarter of million visitors seems more than respectable.
In the US, understanding media relativities in a digital context is clearly more mature than what we commonly experience in Australia. A quick snapshot of Pandora listeners aged between 18 and 34, for example, throws up some fascinating site affinities.
While shared audiences between Last.fm, Radio Time, emusic and Limewire seem self-evident, there are also some strong audience relationships with the likes of Music of Faith, All Gospel Lyrics and the entertainment sites, Bet.com, VH1 and Hulu (Source: Quantcast).
For MTV.com viewers, again aged between 18 and 34, the online repertoire is a little more secular. Sites like Bossip, Skyrock, Atlantic Records, Perez Hilton and NY Mag are all regarded as having some of the highest affinity scores with the cable group’s online viewers.
Locally, the situation is made more complicated by the regular ‘leakage’ of many Australian music consumers to a number of high profile international sites. Perez Hilton pulls in approximately 320,000 local visitors each month, compared with 215,000 for the Huffington Post, and the 3.6m Australians who switch onto the emerging social media giant RockYou each month.
Understanding online music consumption, let alone the potential adoption of music-related applications and technologies, is near ineffectual without better understanding the market’s overall immersion in entertainment content and the predominantly international brands that have secured local engagement on a frequent basis.
May 4, 2009
An interesting article published in the UK last week asked the provocative question: are we drowning in music? In other words, has the currency of music been irretrievably de-valued by the rising tied of mediocrity? (And no, that wasn’t a mixed metaphor).
This post isn’t about to answer that question, but the insight does provides a nice introduction into the world of music entertainment and the growing plethora of online, music-related sites. In Australia, the ranks of music streaming and retail services will shortly be joined by the likes of Guvera, MyDJ and the gorilla of them all, MySpace Music - the latter scheduled for release before Christmas.
We’re not short on services, but what about demand? At what point do we reach saturation where the audience hygiene audience factors start to drop, and in turn, the CPM rates for ad-based music services also start to weaken? But that’s a whole other can of worms.
There are two market metrics Victrix uses here: Nielsen and Comscore. The former suggests online music entertainment has in fact peaked, and now skims along at about 3m uniques a month (March 09) - down from 3.4m in March 08.
In comparison, Comscore considers the glass half-full, and reckons the audience for online music entertainment has grown 21% in the last 12 months, to 4.3m uniques a month (March 09). Again, this isn’t about picking a fight with one or both of the world’s largest panel methodologies; I’ll leave you to ponder the reasons for the discrepancies in audience figures before a more informed opinion is expressed. Both figures exclude online music retailers (and associated applications).
However, where one platform seems to reflect a depleted, almost exhausted market for music entertainment, the other is reporting double-digit growth; not just in terms of the overall sector, but for individual players, including Last.FM - up 115% to 342,000 uniques a month - and even locals visiting MySpace Music (US) - up 72% to 935,000 uniques a month.
Other strong players include MTV Networks (again AUS visitors to US sites) up 60% year-on-year, Universal Music Group up 52%, and the ABC’s Triple J, up 25%.
These audience stats aren’t the makings of an exhausted or confused listening audience. Far from it. In fact, with improved AI and smarter curatorial processes (behavioural targeting), online music entertainment will only evolve into a more relevant service offering.
May 4, 2009
Ok, so Victrix is a little late jumping onto the Twitter bandwagon, but given the hysteria surrounding Swine Flu - with little justification as to why the WHO went to DefCon 5 - maybe a little ‘late to market’ isn’t such a bad policy after all?
This is a case study of how Twitter Intelligence (TI) can sometimes having a profound impact on our (the brand’s) understanding of the consumer’s perspective as well as the questions and concerns which are typically raised in a word-of-mouth scenario.
And this is a major point: I’m not aware of anyone yet describing Twitter technology as a method to measure and monitor word-of-mouth, especially in metric terms such as ‘rate of diffusion’, ‘positive/negative’ sentiment, brand and competitor awareness.
In this example, we have comments, opinions and chats associated with one of Samsung’s most recent mobile phone releases, the Omnia touchscreen. Using PeerIn’s Sentiment Metrics platform we have tracked and quantified these references online - across typical websites, but more interestingly across social media and user-generated content. Included in this catch-all is Twitter communications.
This analysis captures references to “Samsung” and “Omnia” and identifies more than more 20,000 items across blogs (including micro-blogging), forums, websites and news services. In percentage terms, the split is 51%, 26%, 15% and 8% respectively. Micro-blogging, or Twitter, accounts for just over 1% of all comments, which is relatively high for a product-related ‘event’.
However, what it lacks in scale, the Twitter channel more than makes up for in sheer chutzpah. In others words, when you’re limited to 140 characters, your opinions, questions and answers are going to be couched in a very concise, direct and forthright manner. There’s simply no room or patience to waffle, procrastinate or pontificate. Though there’s plenty of opportunity to market, or dare I say it, spam.
In the context of understanding a mindset, the Samsung-related Twitters are pure gold.
“Phones I have my eye on Samsung Omnia and HTC Touch Pro2″
“got lots of games and most importantly a GPS for my Samsung Omnia that is free…holla!!”
“Does anyone want to swap a Samsung Omnia for an iphone?? It would make my day and I would actually think you are crazier than me”
“I returned the samsung omnia and got an htc touch pro…spent all day yesterday unlocking and flashing it…sweet as betch now”
“Goddamn Samsung Omnia. You’ve broken my heart and I’m thinking about leaving you for a Blackberry Curve. It’s not me, its you.”
“Nokia N97 comes with disappointing hardware. Samsung Omnia HD looks better and better every day.”
In these few examples, you have cases of competitor positioning (and awareness) as well as consumer insights into hardware and software configurations.
There are many other examples, but one thing Victrix has learnt very quickly about Twitter: don’t take everything on face value. While no doubt a flexible, responsive and convenient communications channel, Twitter is also increasingly susceptible to astro-turfing techniques, where product placements clumsily infiltrate innocuous Twitter comments. Case in point:
“What a beautiful sunset. I’m sitting by a blue lake and listening to Michael Buble from my Samsung Omnia.”
And you thought the close-up of the Ford badge in Casino Royale was gratuitous!
March 7, 2009
Well, its been almost six months since the last entry on this blog, and I’m pleased to resume my efforts on all matters digital, analytical, and here’s the twist, entertainment-related. The latter is more a reflection of my new role as Managing Director for Peer Group.
September 4, 2008
Events at the Republican National Convention (RNC) in the US have seriously electrified the blogosphere and comments pages across the Internet. More accurately, it has been Sarah Palin’s introduction to the Republican ticket and her acceptance speech which has created some very serious seismic activity, outdoing even Steve Jobs and his iPhone hyperbole.
From Yahoo!’s poll of polls and TechPresident through to Politicaltrends and Wonkosphere, the measurement of online commentary and sentiment has never been so prolific or detailed. In short, the sudden reframing of this campaign has crystalised some very stark differences between all four candidates across the two tickets, effectively polarising general opinion, and more importantly, reducing the independent vote to a negliable number.
The result? A massive influx of individuals prepared to step forward and be counted by their words, many of whom, it might be assumed, have previously been observers (readers) rather than contributors.
This is a digital scream-fest, consuming the worlds of both UGC and professional media. Every story, video, sound-byte attached to the RNC, and Palin’s speech in particular, is being hammered by both liberals and conservatives with a blend reasoned opinion and out-right vitriol.
And the result? An immediate change in online sentiment, with McCain racing to intersect Obama’s lead. Palin’s introduction has been a game-changing event, for better or worst, and online momentum has resoundingly swung in the Republican’s favour. It is this real-time measurement of opinion (though skewed in so may ways) which acts as a very powerful barameter on issues, personalities and tone.
It also puts into perspective the relatively static social media space, in particular Facebook and MySpace and the weight previously given to such stats as Obama’s and McCain supporters (Obama outnumbering McCain more than 5 to 1 on Facebook and more than 6 to 1 on MySpace).
No doubt there is some correlation between funds raised and friends enlisted, but the true test seems to be whether these ’supporters’ are resolute enough to go in and fight for their candidate as the two party machines open up all guns in a broadside battle? The Palin factor was the opening salvo.
August 10, 2008
Those of you who are more perceptive of systems change (others will refer to it as the Zeitgeist) will recognise the Internet’s morphing from a web of documents to a web of people, sweeping aside the traditional, but authoritative role of documentation for the emotive and subjective language of opinion and assumption. Remember, we’re all experts now!
In essence, this change is about electronic word-of-mouth (WOM) or consumer generated media (CGM) ripping holes in the delicate fabric of corporate positioning strategies. Reputations built more on spin and less on actual service and product quality are first in the firing line.
Given then how this ‘ripple effect’ is an obvious business risk, where’s the commercial will to employ analytics to quantify both the speed and direction of this consumer shock waves? Where’s the emphasis on sentiment as a precursor to consumer behaviour?
And that’s the key.
Sentiment analysis (and its poorer cousin, sentiment mining, which is done by the likes of BuzzMetrics, BlogPulse etc.) is the canary in the mine. If done well ( meaning comprehensive, robust samples, with more human interevention than software application), sentiment analysis identifies more than just frequency of expression; it determines market drivers and inhibitors to commercial and organisation growth. It’s as simple as that.
The Twitter example above, highlights a rudimentary form of sentiment analysis, this time focused on the commentary swirling around recent movie releases. A more ambitious analysis would seek to develop correlations or relationships between sales performance or brand equity and the quantums associated with positive/negative sentiment.
The difficulties of using automation to conduct this type of analysis are obvious. Much of the current debate dealing with sentiment analysis is focused on how software can reduce the number of false positives without compromising on the efficiencies of scanning/spidering billions of characters in a cost-effective manner.
The alternative is a hybrid solution involving people (i.e. analysts), with the obvious drawback on volumes and expense. Nevertheless, the rules of sample methodology are equally applicable here as they are elsewhere, suggesting conclusions with high confidence levels can be reached without necessary trawling every online comment or opinion piece.
The bottom line: which is more effective when it comes to semantic orientation classification - man or machine?
Fortunately, this isn’t a battle the machines are likely to win. Quiet simply, the value of sentiment analysis is only fully optimised when context is taken into account, alongside the sentiment of opinion/comments being scrapped. In other words, the value in determining an overwhelming negative or positive sentiment is only realised when we also have insights into the beliefs, personalities, community, influence and authority of the individuals making the actual comments.
And this can’t be condensed or modelled into an algorithm - no matter how hard Google tries.
August 1, 2008
On a month-by-month comparison, the number of online display campaigns in June 08 was approximately 86% higher than the corresponding period the year before - from 2,219 to more than 4,300 campaigns a month, or almost 11 new campaigns a day! (Source: Adrelevance)
Into this mix sits the automotive sector, of which one of the largest online display advertisers is Toyota Australia. Coincidentially (if you believe in coincidences), the number of automotive campaigns in June 08 was also up 86% on the same period a year before. However, it is the broader automotive sector which seems to be providing impetus to that growth - in particular auto publishers/classifieds groups, like Drive and CarsGuide.
This is to the detriment of manufacturers, which in June 08, made up 59% of all auto campaigns, compared with 61% from the year before.
Toyota, as a case in point, slips to 14% of all manufacturing campaigns, despite an increase in the real number of online display campaigns. This four point slip in share-of-voice amongst auto manufacturers is predominantly the result of exceedingly aggressive behaviour from brands below the top three, including Honda, Jeep, Subaru, VW and Mazda.
In the case of Honda, for example, there have been some spectatcular jumps in online campaign activity. In March, its online campaign index jumped more than 1500% relative to its performance in January 08, followed by a similar increase in June.
With auto groups increasingly committing significant advertising dollars to online, at the expense of above-the-line, there is added pressure on media planners to justify higher CPM charges and lower response rates associated with an increasingly saturated automotive media. Unfortunately for planners, auto groups are unlikely to reverse their media thinking given the increasing investments they are making in website-initiated CRM programs.
