ISP Metrics & Limitations

This page logs known limitations in Broadband Map's internet availability data. Limitations in cell coverage data are covered separately on the Cellular Mapping Limitations page.

Availability

• Most of Broadband Map's underlying data on internet availability is aggregated to the level of resolution 8 H3 cells. These cells are typically about 0.74 km² (0.29 mi²). Rather than showing address-level speeds and availability, the data reflects the providers serving at least one address within a given cell and the maximum speeds reported within it.
• For region-level availability metrics (e.g., availability across a city or an entire state), population-weighted availability is calculated over data from many H3 cells. Since ISPs do not always service every address in every cell, Broadband Map's estimates may be overoptimistic.
• For city-level metrics, Broadband Map uses the United States Census Bureau's boundaries for census places. These boundaries may not match postal boundaries or colloquial impressions of city boundaries.
• Availability data isn't real time. The FCC's Broadband Data Collection (BDC) program—the most comprehensive availability dataset Broadband Map has access to—releases data about every six months, and each release captures availability from roughly six months before it comes out. Accordingly, data is generally between half a year and a full year behind.
• BDC data isn't always reported accurately. Beyond accidental errors, the use of BDC data in subsidy allocations may incentivize ISPs to strategically overreport or underreport availability.
• BDC data reflects whether fixed locations (homes, businesses, and other structures) are covered. For products like Starlink Roam that regularly provide service in areas without fixed structures, true availability is not captured.

Max Speeds

• Broadband Map generally shows internet service providers' maximum advertised speeds, which may not match actual speeds.
• With certain technologies (e.g., fixed wireless and satellite), it's genuinely difficult for providers to estimate speeds. Further, characteristics of a specific setup (e.g., antenna angle, satellite dish position, obstacle presence, building material) can significantly affect speeds delivered.
• Most, but not all, speed data comes from ISPs' reports to the FCC. In some cases, potentially when speeds are reported in an overly conservative manner to the FCC, Broadband Map may use alternative sources for speeds. Alternative sources of speed data may not have the same geographic granularity as FCC-derived data.
• Broadband Map is still figuring out how to juggle multiple speed-related data sources. Non-FCC sources have their own issues, including the possibility of being overly optimistic to reflect ISPs' marketing goals.
• Broadband Map doesn't have a totally proceduralized approach for deciding when use of alternative data is appropriate. ISPs that take the time to engage with Broadband Map may influence Broadband Map's decisions about whether to use alternative data. Risk of bias may be heightened when ISPs that have financial relationships with Broadband Map request the use of alternative speed data.
• Speeds reported to the FCC reflect the highest advertised speed at any address within a given H3 cell. Not every address within a cell necessarily has access to the best speeds in that cell.
• The FCC does not precisely collect speed data for services falling short of 25/3 Mbps. These services are bucketed as either short of 10/1 Mbps or at least 10/1 Mbps but short of 25/3 Mbps.

Technology

• The technology an ISP reports doesn't always align with the technology the ISP actually offers in a given area. Discrepancies may be the result of honest mistakes, shady marketing (pretending a cable service is a fiber service), or genuine edge cases where a technology is difficult to categorize cleanly.
• Broadband Map often flags suspicious technology categorization, and availability maps usually offer a settings menu option to override ISP-reported technologies with Broadband Map's predictions. However, Broadband Map's flags and overrides are imperfect and incomplete.

Public API Data

• Without careful handling, speeds reported for services falling in the low-speed buckets may be misinterpreted.
• As of June 2026, the public API does not provide users with adequate options for electing between raw FCC data vs. Broadband Map's override-associated data.