AT&T tiering, femtocells and holistic traffic management

AT&T has finally seen the light and recognised that "true unlimited" mobile data plans are unrealistic, moving instead to tiered offerings. Most of the rest of the world outside the US has long had tiered / capped services for both smartphones and laptop modems, so this starts to bring North America in line with practices elsewhere.

However, I'm not too convinced by the details. It's 200MB/month for $15, or 2GB/month for $25, plus additional overage charges of $15 for 200MB or $10 for 1GB on the respective plans.

The 200MB / 2GB looks too much like a cynical "Goldilocks" fee structure to me. Too Little, or Too Much. But not "Just Right". A good proportion of smartphone users have monthly usage in the general range 200-500MB - as indicated by AT&T's rather disingenuous comment that "currently, 65 percent of AT&T smartphone customers use less than 200 MB of data per month on average."

Let's think about that last statement. Firstly, we know that some smartphone users - notably corporate BlackBerry users - are relatively low-usage, and bring down the average. And they will be on separate BlackBerry / corporate plans anyway. Secondly, that means that if they use <200MB *on average* then it is likely that there will be variation about that average. In other words, a good proportion who are (say) hovering between 150-250MB/month will incur overage fees on a regular basis, assuming no "rollover" of unused allowances.

The other standout is the pricing of the packages is decidely non-linear. If the incremental cost of 1GB of data is given as $10, then that then points to a single-digit $ base cost per GB. So the overage charge on the 200MB package is clearly being made at an astonishing mark-up.

The clear message is that "normal" consumers are being pointed towards the $25 plan, with only exceptionally low-end smartphone users benefiting from the low-rate option.

The other detail missing from the press release is the apparent fact that femtocell traffic ("Microcell" in AT&T parlance) is *included* in counting towards the quota, but WiFi traffic is *excluded*.

[Hat-tip to competitor / peer Peter Jarich via Twitter for the Microcell anecdote. However, please note that I haven't been able to source this independently, so what follow may need to be edited if AT&T issues a contrary clarification]

This goes to the heart of some of themes in my recent research paper on Mobile Traffic Management, and the need for holistic thinking within operators. Given that the RAN generally costs much more than the core network for most operators, there should clearly be differential (or zero-rated) pricing for traffic using femtocell offload. Either that, or there should be a mechanism for customers to charge AT&T for using THE USER'S broadband pipes for backhaul.

It is critical that any policy management and charging infrastruture is capable of discerning bearer type (which could also be UMA WiFi tunneled via the core on some other networks). Otherwise it makes a total mockery of the concept that policy is intended to align pricing with the underlying costs of service delivery.

It also makes a mockery of the femtocell concept as a mass proposition, if the end-user has to pay more than using their own WiFi. If I was a femto vendor today, I'd be spitting feathers about this, as it completely undermines the positioning vs. WiFi as an offload tool.

I also know that many vendors claim it is feasible to distinguish between femto and macro traffic in their DPI / policy products because I've been asking this specific question to many of them the last month or so. And let's face it, it's pretty obvious if traffic is coming through the carrier's femto gateway - if the operator can be bothered to do the integration, and has a rating/charging system up to the job of differentiating it on the quota and bill.

My guess is that the RAN offload/femto project at AT&T has been disconnected from the tiering/policy initiative. This is not the first example of one isolated aspect of traffic management being disconnected from others. Nor will it likely be the last, given the proliferation of techniques and technologies being deployed. Many will have unfortunate side-effects and unintended consequences - as I discussed regarding video compression / optimisation recently.

It's possible that AT&T recognises the issue and will fix it in time - but at the very least it ought to recognise the issue explicitly.

If you want to know more about the range of mobile broadband traffic management options - and the need for a holistic approach to avoid outcomes like this, you need to read my recent research paper. Details are here, and it's priced from just $350.