Comments on: Amazon and Netflix, great together, but not why you might think http://ricmerrifield.com/2010/02/amazon-and-netflix-great-together-but-not-why-you-might-think/?utm_source=rss&utm_medium=rss&utm_campaign=amazon-and-netflix-great-together-but-not-why-you-might-think A business manifesto for cutting costs and boosting innovation Sun, 29 Jan 2012 00:52:31 +0000 hourly 1 http://wordpress.org/?v=3.3.2 By: Ric Merrifield http://ricmerrifield.com/2010/02/amazon-and-netflix-great-together-but-not-why-you-might-think/#comment-100 Ric Merrifield Wed, 03 Feb 2010 20:35:54 +0000 http://www.ricmerrifield.com/?p=1049#comment-100 Well buying a successful and growing internet business is pretty good gravy to get if your primary goal is a better recommendation engine . . . but I agree, it probably wouldn't cost $3.22 billion just to buy that, but with time value of money, who knows how long it would take to get as far as NFLX has gotten with it. Well buying a successful and growing internet business is pretty good gravy to get if your primary goal is a better recommendation engine . . . but I agree, it probably wouldn’t cost $3.22 billion just to buy that, but with time value of money, who knows how long it would take to get as far as NFLX has gotten with it.

]]> By: Gary Manfredi http://ricmerrifield.com/2010/02/amazon-and-netflix-great-together-but-not-why-you-might-think/#comment-99 Gary Manfredi Tue, 02 Feb 2010 01:48:27 +0000 http://www.ricmerrifield.com/?p=1049#comment-99 Or they could just acquire better recommendation technology without spending $3.22 billion. Regarding the recommendations on the item pages, they employ "item-based collaborative filtering" to associate items to items without actually taking in your particular history (there's lots of reasons to for this, one of them is performance over creating a new recommendation set for every single visitor looking at that page). Yet in their "recommendations just for you" section: https://www.amazon.com/gp/yourstore, they do attempt personalized recommendations, but it's quite shallow and rife with problems that have been reported by others. I figure the conversion rates are lower using this method to warrant greater integration. But thankfully, there's better recommendation solutions, especially ones for media, using newer technologies. There's lots of activity in this space, with us being a new entrant. So you never know -- they may wake up and employ one of these... cheers, Gary Or they could just acquire better recommendation technology without spending $3.22 billion.

Regarding the recommendations on the item pages, they employ “item-based collaborative filtering” to associate items to items without actually taking in your particular history (there’s lots of reasons to for this, one of them is performance over creating a new recommendation set for every single visitor looking at that page).

Yet in their “recommendations just for you” section: https://www.amazon.com/gp/yourstore, they do attempt personalized recommendations, but it’s quite shallow and rife with problems that have been reported by others. I figure the conversion rates are lower using this method to warrant greater integration.

But thankfully, there’s better recommendation solutions, especially ones for media, using newer technologies. There’s lots of activity in this space, with us being a new entrant. So you never know — they may wake up and employ one of these… cheers,
Gary

]]>