The simple assumption that most analysts use goes something like this: a monthly subscriber who gets billed regularly is better than a prepaid subscriber who has to pay up front every month.

It’s founded on some basic facts, including that prepaid subscribers on average spend less for service, buy cheaper phones, and tend to defect to other carriers more frequently. Postpaid subscribers—the ones getting a regular bill—spend more on service, phones, and stick around a lot longer, often on highly-profitable family plans.

According to the web site catering to prepaid customers on AT&T’s Cricket brand, for example, a 5 GB plan is available for $45 a month with low-end HTC and LG phones being pitched with one-time payments of as little as $10. One year revenue for the carrier? $550.

By contrast, on its main AT&T web site, postpaid customers are offered a 5 GB plan that costs $75 a month for service and are prompted to add a high-end Samsung or Apple phone for another $25 or more charge per month. One year’s revenue for AT&T? $1,200.

Get Data Sheet, Fortune’s technology newsletter.

But AT&T, under constant assault from T-Mobile (TMUS) ever since the two carriers’ merger was blocked at the end of 2011, hasn’t had much success with its postpaid phone subscribers. It lost a net 266,000 in the second quarter, the seventh consecutive quarter of decline according to BTIG Research analyst Walter Piecyk.

Yet on the prepaid side where AT&T (T) does business mainly under the Cricket brand name, AT&T did quite well, adding 365,000 customers in the quarter, according to analyst Craig Moffett at MoffettNathason Research.

The bottom line is that so far this year, AT&T’s postpaid subscribers grew only 1% while prepaid subscriptions increased 21%. That’s disturbing to Wall Street, based on the ruling assumption that postpaid customers are preferable.

New Math

Thus, Stephens has been trying to push some new math on the analysts. In essence, his argument is that the best customers in prepaid are actually a lot better—and more profitable—than the worst customers in postpaid.

The average service revenue AT&T collected from postpaid customers who have left—and who mostly had not upgraded to smartphones yet—was only $35, he said during a conference call with analysts on Thursday afternoon. But the new prepaid customers signing up with Cricket are bringing in “closer to a $41, $42” of average revenue. Additionally, it costs less to acquire a new prepaid customer and less to provide them with customer service, he noted.

“So from that standpoint, the economics are better, and it is being shown in our margins,” Stephens told analysts, pointing out that while total wireless revenue was down slightly, profit margins were at record highs. “We think of branded customers and try not to make distinctions on classification, really make distinctions on economics, and I think it’s showing up in our profitability.”

Still, some leading analysts don’t see the strategy as sustainable over the long term. “Profitability is improving, largely as a consequence of lower churn and upgrade rates that boosted the margins of the two wireless segments,” Moffett wrote on Friday. “But with a fixed cost business like telecom, it is hard to sustain margins forever when revenues are declining.”

To some degree, AT&T’s effort to overhaul its wireless customer base is going on across other parts of its business as well. The telco is shedding expensive-to-maintain cable TV customers at its U-Verse unit while adding less costly satellite TV customers for DirecTV. AT&T is dropping broadband Internet customers who connect via older DSL lines while trying to add fiber optic broadband customers. And it’s trying to move corporate customers from traditional managed networks to cheaper virtualized networks. If all of the transitions succeed, both revenue and profits should grow.

But there’s still a long way to go and plenty of skepticism among analysts over Stephens’ new math.