Usage-based pricing simply means you pay only for the data you consume. In Google’s case, it’s charging customers $1 for every 100-megabyte increment. According to Google’s blog, you can buy a one-gigabyte plan for $10 a month or a two-gigabyte plan for $20 a month—but if you only use 1.4 gigabytes of data, then Google will only charge $14 for it. If you go over your plan, Google will charge you an another dollar for every additional 100 megabytes. You’re still on the hook for Project Fi’s $20 monthly basic services charge—which includes unlimited voice, text and Wi-Fi access—but when it comes to wireless (3G and 4G) data, you pay only for what you use.

It’s a simple concept—but it’s one that’s failed to gain traction in the U.S. Carriers have been loathe to assign a specific dollar amount to individual megabytes because it turns their most valuable product into a pure commodity. If data were metered, we’d buy it like we buy gasoline. Some people would be willing to pay extra for the high-octane version of mobile data—faster speeds, better coverage—but for most people buying data would become the equivalent of shopping around for the cheapest per-megabyte prices.

In the carriers’ defense, there’s also consumer resistance to the idea of usage-based pricing. In the voice world, tracking your consumption was pretty intuitive: For every minute you talked, you used up a plan minute. But in the data world there’s a wild variance in the amount of data different apps consume. In a metered pricing model, a normally moderate data user could find his bill doubled in a month he watched a few extra YouTube videos. Many consumers aren’t prepared for that kind of bill shock.

That’s why mobile carriers have leaned heavily on the data bucket—selling a set amount of gigabytes each month that can be consumed by a single device or shared among a family of gadgets. Buying a 10-gigabyte plan gives a customer who averages five gigabytes of data traffic a month a lot of headroom. But it also means that customer is buying far more data over the course of a year than he could ever use.

Google is challenging that fundamental pricing assumption in the U.S. mobile industry. Project Fi still makes you sign up for a nominal data bucket from one gigabyte to 10 gigabytes a month, tacitly acknowledging that customers want to plan their monthly charges. But the specific plan you’re on is meaningless. Project Fi bills you on a straight up meter.

That would keep in line with Google’s plans to experiment with wireless services, as Google senior vice president Sunder Pichai outlined at Mobile World Congress in Barcelona last month. According to Pichai, Google wants to use its still unnamed mobile service to add a connectivity component to its ongoing work with mobile hardware and software, and apparently that means tinkering with how the data feeding its phones and apps is sold.

Google certainly isn’t the first company to test out usage-based pricing models, but most consumers have probably never heard of the carriers that do. Upstart operators like Ting and Karma offer data on a meter, and along with a host of other smaller carriers like Republic Wireless and FreedomPop have been quietly challenging accepted mobile billing practices. What Google has in common with Ting and Karma is that it’s a mobile virtual network operator (MVNO), a network-less carrier that buys minutes and megabytes from an established national operator and then repackages and resells them to its own customers. What Google doesn’t share in common with Ting and Karma is its enormous spotlight.

Using its influence, Google has the opportunity to educate the tech-savvy consumers who will likely gravitate to its service on the benefits of the metered plan. It’s selling moderately priced data with the most flexible pricing plans in the biz. That might just be enough to convince an influential early adopter audience that there is a better way to pay for mobile broadband than the data bucket.

The only problem with that plan is Google is hemmed in by its own carrier partners. Google is buying access from Sprint (S) and T-Mobile (TMUS), so it’s limited by the wholesale rates they charge for minutes and megabytes. Google does have more flexibility in that it’s launching Project Fi concurrently with a 1 million-strong Wi-Fi-hotspot network, presumably crowdsourced from businesses joining Google’s rumored business wireless program. That means it will be able to implement a Wi-Fi first model where its customers connect to free access points whenever possible, giving them many more opportunities to surf, stream and download off of pricier 4G networks. But it needs the carriers’ cellular networks to fill in the sizable gaps between Wi-Fi hotspots.

The carriers won’t necessarily work against Google—T-Mobile and Sprint weren’t forced to partner with Google, and in fact carriers have allowed their virtual partners to flourish in the U.S. for a decade. MVNO TracFone owns not a lick of network infrastructure but it has more than 20 million mobile customers, making it the fifth-largest operator behind the big four. But Sprint and T-Mobile probably look at Project Fi a little differently.

Google isn’t going after budget prepaid customers are foreign nationals. It’s targeting its Nexus 6-focused service would target the hard-core data users that are the bread and butter of every nationwide operator. Google has stuck to its “experiment” story and has said that it has no intention of taking the big incumbent operators head on. Frankly Google probably wants none of the hassle and heartache that comes from being a nationally competitive carrier (can you imagine a Google call center?). But if Google’s fledgling mobile service proves popular, Sprint and T-Mobile might rethink that relationship.

That said, there are already signs that the U.S. mobile industry is heading in the direction Google wants to take it. AT&T (T) and T-Mobile have both launched data rollover programs, which allow you to carry over a portion of your unused data into the future billing cycles. Those rollover plans aren’t quite the same as simply billing customers for what they consume, but they indicate carriers are more willing to let consumers actually use the data they buy every month.