Before finding a permanent home in the testing kit, a phone has to pass initiation through extensive field testing trials. If the phone can run the testing software, stay connected to the master phone, and deliver data, it stays in the running.

RootMetrics selects higher-end phones for each carrier, like the HTC Rezound for Verizon, the Samsung Galaxy S II Skyrocket for AT&T, and the Samsung Galaxy Attain 4G for Metro PCS. Why? Because these handsets should capture the best network results across the board for all networks, which will paint a clearer picture of the network’s capability. If the test results are poor, chances are it’s because of the network and not due to complications with a sub-par phone.

As such, RootMetrics is auditioning a handset to replace Sprint’s Samsung Epic 4G, which supports 4G over WiMax, with an LTE-ready phone. RootMetrics wouldn’t say which phone is the frontrunner, but they’ll buy 12 units of the winner off the shelf to add to their testing kits.

3. Network testing runs on gas Scouts may be chosen for their technical aptitude and troubleshooting skills, said my RootMetrics guides, Josh Doehne, senior marketing manager, and Tom Ingarson, field operations manager, but they spend between 60 and 70 percent of their 8-hour shift in the car. That’s a lot of driving, around 150 miles per shift and collectively about 200,000 miles a year.

RootMetrics’s house-written desktop mapping software randomly generates the testing route, which scouts then download onto the Garmin. Even though itineraries are selected at random, they’re based on the city’s urban, suburban, and rural geography. Moreover, the route-generator abides by statistical definitions of validity, which means that RootMetrics is making sure it’s conducting enough data and voice network tests in areas where people live, drive, and shop.

“We’re doing enough sampling out there,” says RootMetrics’ president and CEO, Bill Moore, “that we’re properly characterizing the entire marketplace.”

Phone tests are mostly automated; the solo scouts start the data-collection app at the beginning of their shift and turn it off at the end. If all goes well, they won’t have to disturb the phones, except to carry them indoors for more testing and then back out to the rental car. However, in the rare event something goes wrong, scouts are trained to troubleshoot.

“Don’t the scouts get bored?” I asked my guides. Ingarson, who started off scouting and still goes on training and monitoring rides, says the time flies by. Between keeping an eye on the routes and phones, and logging stops as part of a note taking protocol, there’s plenty to do.

4. Scouts are all around you See a man or woman clutching an oversize briefcase and walking the aisles of your local supermarket, and you may be looking at a RootMetrics scout. They could be in the car next to you on the road, portfolio open on the passenger seat, eyes glued to the Garmin. As much as it’s possible to blend in with a large black briefcase at their side, they do.

While scouts spend the majority of the time behind the wheel, RootMetrics tests indoor locations, too. In the course of each shift, there are 10 waypoints that a scout must hit along his or her route, and for each of these 10 are 3 indoor locations that a scout can use to test the portfolio phones indoors.

Scouts are encouraged to takes tests indoors at each waypoint, but use their discretion. Indoor tests last 20 minutes, or four testing cycles, during which scouts anonymously walk the aisles of supermarkets, convenience stores, and book shops, sipping coffee, scanning nutrition labels, and thumbing magazines while the master phone and its worker phones probe the radio waves.

Out of RootMetrics’ growing pool of 65 employees, 13 are scouts. However, no more than 7 are currently in the field at a time, and no one tests more than 7 consecutive days. Schedules vary tremendously, and larger markets require more scouting power.

Here’s how it works. Ever the statisticians, RootMetrics defines markets by U.S. census boundaries, so San Francisco also includes a larger circle of cities to the north, east, and south (go west and you hit the ocean.) Based on population size, RootMetrics calculates that San Francisco is a seven-day market, which one scout can handle on his or her own (Doehne says scouts are evenly gendered.)

There are the small accidents that come with steering a vehicle on the road for a third of a day — the bumps, scrapes, and fender benders, but Ingarson insisted that there’s never any violence, and scouts are told to avoid or vacate any situation that feels unsafe, no matter what the Garmin says.

Sometimes it’s the scouts who appear shady. Every scout has gotten pulled over by law enforcement at least once for their strange-seeming behavior, Ingarson told me. It’s almost a rite of passage. When his Garmin led him to a naval base in Baltimore in the dead of night, suspicious base cops grilled Ingarson for 5 or 10 minutes, but a quick phone call to support staff back in Bellevue quickly cleared things up.

While chats with the local authorities are the height of most confrontations a scout sees while testing carrier networks, Ingarson hinted that scouts get an eyeful of humanity’s more shadowy side. “You see a lot of stuff,” he said.

Where do the numbers go? It takes about a month for RootMetrics analysts to crunch the numbers and compile them into meaningful reports. Free reports appear online every six months for consumers, and the numbers also feed into free apps that you can download for Android and iPhone. In addition to drawing from RootMetrics’ database, the apps also combine crowdsourced results from phone owners’ real-time tests.

Revenue comes from syndicated reports that the device-makers and carriers buy. The four major national carriers are all RootMetrics clients, and use the tests to compare their strengths and weaknesses against their competitors. “We offer carriers the opportunity to see what they get in terms of network performance and signal strength,” said RootMetrics’ Doehne.

“T-Mobile has a long-standing relationship with RootMetrics,” a T-Mobile spokesperson told CNET in an e-mail, “and has worked with them for many years to validate benchmarking methodologies and develop strategies to measure customer experience.”

As I mentioned above, the company also partners with third parties, like CNET, to provide interactive coverage maps.

In the future, RootMetrics plans to move into device-specific tests, perhaps for individual tablets, and may license their testing software directly to phone-makers and carriers so they can perform their own in-house testing using some of RootMetrics’ methods.

Testing is huge More than ever, a carrier’s quality — especially 4G data speed — is a critical factor in choosing one provider over another. CNET’s cell phone editors do our best to test call quality and network speeds in a variety of situations for each phone, but we can’t be everywhere at once and even with time and transportation on our side, testing is limited to a small portion of our respective cities.

The two urban hubs where CNET editors test, San Francisco and New York, are especially challenging. Our cities are riddled with radio-blocking concrete and hills (S.F.), too-few towers, unexpected dead zones, and heavy congestion. We’re often the last to see a new data network technology like LTE, which can make it difficult when our readers ask how fast Sprint’s new 4G LTE network operates (it hasn’t launched in either spot at the time of this story) or T-Mobile’s speeds compared to AT&T’s and Verizon’s in the midwest.

Companies like RootMetrics, although they’re not the only ones in the game, offer a useful service to consumers seeking independent information about carrier performance in their cities and towns.

Smartphones Unlocked is a monthly column that dives deep into the inner workings of your trusty smartphone.