Tuesday, January 22, 2013

What are the Right Metrics to Measure Service Provider Success?

“The financial community does not measure our industry correctly,” says Norman Fekrat, former VP and partner at IBM Global Business Services. And there will be consequences once analysts finally figure out that the current metrics, from the legacy voice business, do not accurately describe the actual financial results being generated by telcos globally.

The number of subscribers once was a meaningful metric. Because “subscribers” was useful, so was the concept of “churn,” reflecting a service provider’s ability to keep its customers.

These days, “revenue generating units” are reported by many service providers, because that simply makes more sense. Average revenue per user likewise made perfect sense in a long era where “subscribers” and “lines” were accurate and useful ways to measure business health.

Fekrat argues that the current metrics actually do not capture financial performance in ways that will matter as all services wind up as IP-bandwidth-based apps. In the era to come, where the fundamental network resource consumed by any app is “gigabytes,” profit will have to be measured, per service or application, in relationship to use of the network.

In a voice-centric business model, additional usage actually did not really affect “cost,” in terms of use of the network resource. That means the profit of a video entertainment service would have to be evaluated not only in terms of revenue, but also in terms of consumption of network resources.

The same would hold for voice, messaging, web surfing or any other application using the network. Part of the reason for Fekrat’s concern is that, just to keep profit margins where they currently are, assuming growing consumption of bandwidth, cost per gigabyte has to decline about 70 percent to 90 percent every three to four years.

Some of that cost reduction might already be happening, at least for buyers able to buy in some volume. Fekrat assumes wholesale capacity prices of about $4 to $5 per gigabyte. Some buyers or sellers might argue prices, on some routes, already are in the two cents to three cents per gigabyte range.

In other words, if the cost per gigabyte per service argument is valid, at least for the capacity part of the business, costs might already be falling fast enough to make operating, capital and other overhead costs more significant than network costs, at least where the core networks are concerned. Access networks might be a different matter, since traditional cost analysis might attribute as much as 90 percent of end-to-end cost to the access networks on either side of any session.

And Fekrat has one benchmark in mind: service provider network costs must, over time, match those of Google. That’s a very tall order, but wise advice, if you assume that, in a competitive market, over the long term, the lowest cost network wins.

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...