The purpose of a femtocell is, first and foremost, to reduce a carrier's churn rate and increase ARPU (annual revenue per user). A femtocell could allow this by providing you coverage in a currently poor area in your home or even an otherwise badly covered neighborhood. If the coverage is good, you'll use your service more, potentially incurring overage rates (on a standard plan) or by paying a fixed additional amount per month for more minutes or an unlimited home usage plan (AT&T, Sprint).

Happy customers with great indoor coverage don't abruptly switch carriers. In fact, such customers are then more likely to convert the rest of their family to the same carrier, if they were holding out, to gain the benefits of family plan discounts, and potentially cut off their landline.

For AT&T and Verizon, losing a landline isn't a bad deal because landlines are tariffed in such a way that there's not much money to be made from them. (Add-on services, yes; plain POTS, no.) In AT&T and Verizon's home territories, they can be making money selling someone broadband and selling them femtocell service for the same-branded wireless offering. So far, there's no indication that there will be cross-offering subsidies.

The biggest two problems with femtocells are the cost and a required location awareness. Femtocells are being sold by Verizon for $250 (one-time fee) and by Sprint for $99 with a $5 per month usage fee (forever). AT&T's pricing hasn't been set, but it's likely to be more like Sprint's.

Verizon isn't cushioning the blow by offering any calling deals. Both Sprint (currently) and AT&T (future) offer unlimited calling plans when using the femtocell. Sprint adds $10 per month for an individual cellular plan and $20 per month for a family/multi-line plan.

Most calls at home are likely to be made within the evening and weekend rate period, so this may be gravy for Sprint and AT&T regardless. The biggest winners would be parents who have teenagers calling a lot during early-release, early-evening, and vacations; or small businesspeople using cell plans to avoid racking up long-distance charges.

Sprint and Verizon are focused on extending coverage, and both allow other parties to make calls over the femtocells. Sprint lets a subscriber block this out by specifying up to 50 registered phone numbers. Verizon also allows blocking (through what seems like a more cumbersome process than Sprint), but reserves the right to allow one unused call slot to be used by an unregistered user who can't connect to a cell base station. Given that call slots are about 40 Kbps of usage, that's not horrible, but it's a little tacky, isn't it?

T-Mobile's offering, in contrast to these femtocell services, relies entirely on Wi-Fi, but requires a customer to switch to a handset that supports unlicensed mobile access (UMA), which allows seamless call handoffs between a GSM and Wi-Fi radio connection. To the cell network, the calls appear identical, but a lot of fancy footwork has to happen in the phone and on the backend to make UMA work.

T-Mobile charges about $50 for a router with the latest Wi-Fi voice enhancements, and they offer a landline replacement service, too, that can be found in a single router. Handsets aren't more expensive than their non-UMA equivalents, although there are many fewer choices. T-Mobile charges $10 per month for 1 to 4 lines for unlimited hotspot calling, whether from a home or any T-Mobile hotspot or roaming partner hotspot, as well as from any other networks to which a handset can authenticate.

The advantage for T-Mobile has to do with the second femtocell problem: location awareness. Because femtocells use licensed frequencies that are limited by geographic regions, they must also contain GPS technology. GPS receivers work poorly inside. Without a GPS receiver, a femtocell could operate illegally. This also prevents femtocells from being used outside a given country. T-Mobile has no such limitation with Wi-Fi, and I know of some users who have taken their HotSpot@Home gateway overseas while traveling.

Given that femtocells only need find their location once each time they're powered up (assuming they're moved only occasionally), you'd think this wouldn't be too big a problem, until you realize that a 30-foot GPS antenna is included with the femtocells currently being sold.

That makes it a bit of an issue. I wrote today over at Ars Technica about Rosum, a firm that can assist weak GPS reception by using a television-frequency receiver to triangulate fairly precisely indoors, exactly where femtocells are likely to be found.

But this remains a big femtocell issue. Customers that can't get a GPS lock will be told that they can't use a femtocell, and will return the item.

Allen Nogee, an In-Stat analyst, released a report recently in which he speculated that the greatest savings to carriers would be to bundle femtocells in with broadband modems in markets they serve, where millions of customers would suddenly be extending Verizon and AT&T's cell networks, for a potential savings of up to $4 billion in averted network buildout costs.

Many commenters on my description of Nogee's report at Ars Technica said: What's in it for me? And they're right. For carriers to encourage the adoption of femtocells, there has to be a deal that doesn't involve irritaing setup (or a 30-foot antenna run), that's far cheaper than the hundreds of dollars offered now, and that has a real incentive, like unlimited calling at a low rate.

With all that carriers can save from femtocell deployment, they're going to have to give back a lot more to their subscribers than they're currently planning.