[lpolovets]

One particularly nasty thing about automatically throttling the top 5% is that it is completely divorced from costs and level of usage. Over time, if heavy users try using less data, that will just depress the amount of usage required to be the in top 5% -- but 5% of customers will still be affected by throttling.

On a side note, I wonder how the cost of supplying customer service to address the complaints of the top 5% compares to the cost of just letting people consume as much data as they like. I would guess even a few minutes of a CS rep's time is more valuable than letting someone download an extra gig of data per month.

[click170]

That is why I made as much noise as possible about the (IMO) abusive throttling practices of my previous ISP. Don't take this cr*p lying down, call them and tell them youre unhappy, ask them to disable it on your connection (they won't) and when they refuse, demand to speak to the supervisor. Rinse and repeat. Make it cost more to throttle than it does not to.

[abs314159]

It's more underhanded than nasty but you're absolutely right that this didn't mean what they portray and most people thought.

Regardless of whether a given user moderated his data usage voluntarily, because he was throttled, or switched plans, the vast majority of the top 5% likely leaves the pool every month. This means that six months after the change was introduced roughly one quarter subscribers will be removed.

Why would AT&T do this? To avoid the negative publicity associated with eliminating a plan so many people valued while maximizing profitability. This was quite cleverly executed: Offer cheaper limited plans to save customers money, which has the effect of offering new customers a more competitive price while existing customers will happily pay the higher price as insurance against potential overage fees. Then, by removing the highest data users, they could maximize revenue from them while continuing to collect the higher fees from the lower data users. Then once unlimited becomes a less attractive offer than the $25 plan, eliminate it to avoid revenue leakage. The strategy might not be optimal but its superior to the simpler alternatives in the short run.

Of course, the reason is works is that the majority of customers are sufficiently credulous to purchase an implicit insurance contract that AT&T never had to honor. Their actions are legally defensible since the customers did receive the uncertainty reduction in the present period but most of them likely knew they wouldn't go over and paid only for the option of maintaining that protection in a future period. That said, communication that is intentionally ambiguous so as to benefit from misunderstanding should constitute an unfair trade practice.

Disclaimer: I have no knowledge of the actual thought process behind these decisions.