Thursday, December 03, 2009

Limited Unlimited

[an atypical business-related post with due credit to the Telco 2.0 Blog, an excellent resource on the modern economics of telco]

I've witnessed this point argued again and again: at my current employer, my previous employer, and the one before that. Telcos are greedy; telcos sell "unlimited data" but deliver otherwise; telcos will lose the broadband market to a player who can offer better value; and so on. Definitely I'm not defending misleading marketing of broadband services, but claiming that telcos are on a mission to exploit consumers falls short of the mark.

It's really not as simple as that.

The problem the telcos have is this: their costs building and running a packet-switched network are fundamentally a function of data volume on that network, (# of bits transported) * (cost per bit). As a first-order approximation, this linear function works pretty well; in this simplified model the cost per bit is an aggregation of OpEx and CapEx (amortized). CapEx includes licensing wireless spectrum, building out towers and backhaul and so on.

Telco costs are a function of data transport volume, more or less.

So you might be tempted to think that a "cost-plus" pricing model is the answer: charge the consumer per transported bit (ie. a metered data plan) at a rate of X * (cost per bit), where X>1. In general these pricing plans don't work well because on the consumption end what users end up paying doesn't correlate with the value they're getting. Is watching that YouTube video on my iPhone 100x more valuable to me than downloading that email from my boss? No, but it might cost me 100x more. Consumers can't make sense of that, and metered data plans haven't worked for the industry.

Consumer value is not a function of data transport volume.

So the telcos move to a subscription-based pricing model, otherwise known as flat-rate. Here they charge you a monthly fee for access to the network but typically give you "unlimited data" along with that. This is easy for consumers to understand, avoids the problem of data value being independent of data volume, and everything works just fine... EXCEPT that now the telco's revenue is a function of (# of subscribers) while their costs are a function of (# of bits transported).

In a subscription-based world, telco data costs and revenues are decoupled.

Having your revenues decoupled from your costs is a dangerous situation to be in, because you could easily find yourself paying out more money than you're taking in. So you try a few things:

  • you place a cap on the "unlimited data", to control how far revenues can deviate from costs and keep your margins positive. Providers are in general moving this way.
  • you start throttling traffic, for the same reason
  • you start looking at ad-supported models, inserting X ads per bit transported, to couple revenues to costs again
  • you start looking for a way to make your costs independent of traffic but instead a function of the number of "unlimited data" subscribers

After the launch of iPlayer in the UK (iPlayer is the BBC's streaming media product), one major ISP revealed that their costs of carrying streaming traffic trebled within weeks. Unless they can pass those costs to consumers, or fund them some other way, or prevent them entirely, they'd be in trouble. Bad news when your costs treble and your revenues don't.


julianb said...

Interesting stuff. My old employer used to wrestle with the same problem and we tried (and failed) to market technology to telcos to allow them to charge per service (e.g. per email, per video, per VoIP call etc) by doing "layer 7" analysis on IP traffic. Never worked, the flexibility and dynamism of TCP/IP-based services meant that it was hard to scale such a solution.

Regardless, I feel that Telcos are a little marginalised by Net Neutrality. As you mention, they need to continue to invest in their infrastructure and need a means of differentiating themselves. I think QoS will become more of a differentiator over time and consumers will need to start paying more to get acceptable quality for the IP-based services they'll come to rely upon.

Alternatively, they could just acquire NBC.

Jean said...
This comment has been removed by the author.
Isaac said...

Layer 7 analysis is conceptually tempting but ultimately doomed too, I feel. Even at the application layer you can't easily value the traffic and price it accordingly.

fwiw I'm a perfectly happy Comcast customer. I never have to think about my broadband consumption (including watching HD Netflix on TiVo) and haven't come up against any limitations... yet.

Nick Telford-Reed said...

Great summary, Isaac. Not sure I've seen it boiled down so nicely.
The root of this thorny issue lies in the complex relationship between consumer value and information density. In the old, offline world, consumers do pay for data volumes. We just don't think of it that way. But a newspaper is cheaper than a book. Colour magazines full of glossy pictures are dearer than b&w print. Streaming media are so expensive that you have to pay £8 per film to download (through special technology called the cinema). TV is an exception, and the UK in particularly the dear old beeb proves that which sufficient volume and arguably a market dominant position allows the unlimited data model to thrive.
I still believe micropayments are key to fixing this is the online world but that just be me getting stuck in the late nineties!

Chief said...
This comment has been removed by the author.
Matt said...

I'm not convinced you can't train end users about the 'cost' of their behavior. Electricity is metered and yet there are the same order of magnitude differences between different activities and appliances as you'd find between online applications. Consumers know that heating or cooling their living space outweighs all other electricity consumption.

Similarly in the online space, video is the salient part of consumption. It really boils down to video vs everything else, and that's a straightforward thing to understand.

If you make users pay, they'll educate themselves quickly.

FJGLYNN said...

I think you’re right on here. Especially in the mobile space. Thankfully I’m grandfathered in on my unlimited data plan for my iPhone -- although I wish I could tether to it (thank you AT&T....). And it will be interesting to see what happens on the wires from a flat rate perspective once Apple fires up that data center and dishing out lots of video streams.....