We've had this discussion so often, I'm not going to type up the whole thing of access, time-based vs. physical, pipes and buckets, transient nature of data, etc. it's all on this board.
Out of interest, these other cellular providers, what are their business rules when it comes to rollover and order of consumption? This article is about MTN, so how does it work for Cell C and 8ta?
I don't use Vodacom since 2006 because the company breached my contract and despite acknowledging this persisted in f@cking around, granted it was a measly student with a weekender account and sought blackberry bolt on is a perfect example of grand corporate stupidity - - the goodwill destroyed is nothing in comparison to the business lost to MTN - but it is clear to me that they do take the lead in premier 3G internet services and come dead last in providing cheap consumer grade internet. IIRC they introduced roll on policies earlier than the other players but this doesn't have
The MNOs and Telkom are largely to blame for the construction of an equivalency between paying a service provider and an expectation of a quantity of data in megabytes to use - the idea of buying data which should suggest that the buyer owns the data to use how and when they want.
The bottom line is that instead of aggressively from the beginning offering internet access with a tiered capping/throttling policy with soft caps leading to throttling and so on they moved towards selling data at ludicrous prices R2 per meg for goodness sake has always been highway robbery and in bundles. By starting ADSL with a 3gig cap Telkom set up a broadband market which prizes using the data allocated, fortunately this move has almost been reversed with TI going to an uncapped proposition accross the board.
Consumers are demonstrably able to use under 150 megs on an "uncapped" device product with few exceptions and even higher demand smartphones does not in and of itself lead to usage of more than 2gigs for the bulk of consumers a month but because data packages are sold and have been so since inception in the way in which they are it is impossible to argue that the transient nature of data justifies an arbitrary time window system that is devised from a billing perspective and not from a physical capacity perspective.
When somebody buys a railway ticket for a specific date there concern is exclusively about the price and whether it is an acceptable price for their purposes, they don't give a damn about how the railway operator and railway industry works or the costs involved; on the other hand if the operator sells tickets for a certain number of trips to be used within a certain period of time the consumer may do the maths and almost invariably will wish that the period within which to use the ticket was longer in choosing which bulk ticket to purchase. SA mobile data industry largely works in the second equation with data bundles but because you've set the ground to equate purchasing a bundle with an expectation to use the entire bundle (rather than being able to use the entire bundle) you are stuck and it simply doesn't suffice to argue that there are technical reasons - because the consumer shouldn't have to care, the consumer buys the product or service not the underlying system. If a railway ticket is too expensive its because Transnet is full of crap and that is how people think.
Moreover - and probably more importantly - I don't believe that the guys working out products for retail actually understand the situation and every time a product is released which focuses the size of a data bundle you simply reinforce the idea of buying data itself which simply doesn't map the oft expressed technical realities.
The bottom line is that you'd have a lot less complaints about things like roll over if Vodacom and MTN aggressively pushed its data products along the lines of being the recommended high usage on a particular product and establish a soft cap at the advertized bundle size at which point consumers can elect either to (a) pay an out of bundle rate; (b) experience throttling which grows more severe until 25% over usage is a top (at which point access is cut); (c) purchasing a once off bolt up which expires at the same point as the contract and is priced accordingly. There are obvious (and probably lots of not-so-obvious reasons) why such an approach would not work but my point stands.