Original URL: https://www.theregister.com/2009/09/22/long_tail_netflix/
DVD data backs Long, Poor Tail theory
More starvation tips for dippy e-tailers
Posted in Legal, 22nd September 2009 11:32 GMT
Last year WiReD magazine editor Chris Anderson's Long Tail theory collided with reality and came out with a black eye. Digital music sales showed the popular hits becoming more proportionately popular, while focussing on the 'tail' promised the e-tailer starvation with 10m of 13m songs were never even downloaded once.
Researchers found a similar effect on P2P downloads, where the punter had an near-unlimited choice, and no fee to pay. So how would the theory fare when applied to movie consumption?
Two academics at Wharton business school at the University of Pennsylvania have done just that, examining a sample of data from NetFlix, the home movie DVD rental service. And although their picture is less grim than for music, the bottom line remains the same: you're likely to go bust chasing Anderson's fantasy economics. Only when the demand was measured in absolute terms could a Long Tail effect be detected, with consumers pursuing niches at the expense of hits, Professor Serguei Netessine and doctoral student Tom Tan discovered.
The study examined movies that NetFlix users had rated. As has already been noted with online movie ratings, collaborative filtering leads to groupthink - people review what everyone else is already reviewing: a cybernetic feedback loop described here. This is apparent here.
Demand for the top one per cent rose from 14.2 per cent in 2000 to 25.5 per cent. Demand for the Top Ten per cent of DVD inventory also rose significantly, from 67.5 to 78.5 per cent. The bottom Twenty Per Cent of stock saw demand rise from 0.20 per cent to 0.30 per cent.
"These results imply that the demand for hits tends to be less and less concentrated in absolute terms, but more and more concentrated in relative terms."
But low-sellers remain low-sellers, now matter how you measure it. "Demand for niche movies decreases over time, whether popularity is measured in absolute or relative terms," they note. They also noted that punters "tend to be less satisfied with niche movies than with hit movies."
So how does this affect business decisions?
WiReD's plagiarist-in-chief, who viewed drafts of Netessine and Tan's study, has greeted its conclusions with the same response he has for all analysis: he says it proves he was right all along. That's because the study shows evidence of demand for inventory beyond the constraints of what a physical retailer can stock. Ah, say the authors - but that doesn't mean it's going to make any money.
There's another rule of business to remember: no matter what the shape of the market, your revenues must exceed your costs. Anderson's faith-based economics is singularly ill-equipped to deal with this reality.
You can download the 37 page analysis here [PDF, 550kb] and read a nice summary at Wharton's site here. ®