Monday, January 7, 2013

New AT&T U-verse "Screen Pack" Illustrates "Value-Based Pricing" Dilemma

AT&T is offering U-verse TV customers a new Screen Pack feature, costing $5 a month, and offering access to unlimited viewing of a library of about 1,000 movies. The content can be viewed on U-verse TV, U-verse.com and on the U-verse app for tablets and smartphones, AT&T says. 

In one sense, that is a simple business decision by one Internet service provider. There is no absolute reason why any particular service or application has to be sold at retail on any basis directly related to the cost of providing the service. 

In fact, there is no reason why a firm cannot sell a product at cost, or at a loss. Still, service such at Screen Pack, and streaming of Netflix movies, do raise questions. Ignoring marketing and fulfillment costs, what does bandwidth and content licensing really cost, for such services.

And what level of actual average usage is a "breakeven" business case? By some estimates, a standard two-hour high-definition movie might consume about 3.6 gigabytes. A standard definition movie of the same length might consume only about 700 Mbytes (much depends on coding, of course). 

One presumes there is some clear point where AT&T might start to "lose money" in terms of licensing fees, incur higher network usage that could affect peering deals, or incur consumer displeasure because monthly caps are breached. On the other hand, AT&T and other ISPs might someday create "video-specific" usage plans that accommodate the higher usage heavy video watching represents. 

Though it is not a real question at the moment, one wonders how to reconcile the cost of bandwidth for video, which currently might be said to "cost" as much as bandwidth used for voice, messaging, web surfing or other applications, but which has quite different quantitative dimensions, and a clear expected consumer price point. 

In other words, including costs of bandwidth, peering, marketing, delivery, licensing and other costs, people expect "unlimited streaming" for a small fixed cost. In that sense, the "cost" of any single video event is deemed to be relatively low, even if "value" might be moderate to high. 

Another way of putting matters is that a consumer would expect unlimited access to 1,000 movies for $5, while a bucket of voice minutes of use might cost an order of magnitude more, even while consuming a vastly-smaller amount of bandwidth. 

The way I used to describe this was that 24 hours of video delivery of an older analog TV system would easily represent the equivalent of scores of DS-3s worth of delivered "data."

Yet where a single local DS-3 might cost $10,000 a month, for a cable TV subscriber the cost of scores of DS-3s used for video would be $35 a month up to $80 a month. 

It used to take as much as 24 MHz of bandwidth to deliver a single, uncompressed full-motion video stream, or about half a DS-3. 

These days, using much better coding, we can squeeze multiple standard-definition TV signals into a couple of megabits per second, or less, and a single HDTV signal into 6 MHz of bandwidth. 

The point is that there is a vast difference between the "value" and the "price" of network resources and bandwidth used to deliver a single TV event, compared to two hours of talking or texting, or web surfing. 

On a revenue-per-bit basis, voice and texting are really high value, for the amount of bandwidth consumed. The revenue-per-bit for entertainment video is frightfully low. How to reconcile those extremes is going to be an issue, some day, if value-based pricing happens. 

Video has for some time been driving bandwith consumption, but without a good relationship between value and revenue, from an ISP perspective, or from a retail buyer perspective, based on prevailing tariffs. 



No comments:

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...