Microsoft’s usage data shows FCC overstates broadband availability

22 March 2019 by Steve Blum
, , , ,

Microsoft oregon analysis 5dec2018

Microsoft is the latest organisation to tell the Federal Communications Commission that its broadband availability data is wrong. Earlier this month, an Internet advocacy group uncovered an egregious outbreak of map spam that skewed the FCC’s broadband analysis in several states, leading to a premature declaration of deployment victory (h/t to Wendy Davis at Digital News Daily for digging out the story). Last week, Microsoft presented its own analysis at the FCC, based on Internet usage data it collected itself, and came to the same conclusion…

The Commission’s broadband availability data, which underpins FCC Form 477 and the Commission’s annual Section 706 report, appears to overstate the extent to which broadband is actually available throughout the nation. For example, in some areas the Commission’s broadband availability data suggests that Internet Service Providers (“ISP”) have reported significant broadband availability (25 Mbps down/3 Mbps up) while Microsoft’s usage data indicates that only a small percentage of consumers actually access the Internet at broadband speeds in those areas.

Microsoft originally provided its data to the FCC in December, but it didn’t seem sink in. For example, the FCC claims that 91% of Oregonians can get broadband service a minimum of 25 Mbps download and 3 Mbps upload speeds. That’s the standard for usable broadband service adopted by both the FCC and the federal agriculture department. But Microsoft, which can see how fast its customers’ connection are, says the real figure is 60%.

Nationally, the discrepancy is even bigger. The FCC did its victory dance based on data that seemed to show that only 25 million people in the U.S. lacked access to that minimum broadband service level. Microsoft’s analysis indicates that 163 million U.S. residents “do not use the Internet at broadband speeds”. Availability and actual usage are two different metrics, but those differences cannot, by themselves, account for the 138 million person gap. Service providers might claim to offer a particular level of service in a given census block, but that doesn’t mean they’re offering it to everyone who lives there or that everyone can afford it.