FCC broadband data vs. ground-truth location counts, explained
If you've pulled a BDC filing for a county and then driven it, you already know the two numbers rarely match. The fabric says 340 broadband serviceable locations in a given hex. Your field team counts 280, or 410, and now you're explaining the gap to a state broadband office that wants a clean unserved/underserved list by Friday.
Two different measurement methods produce two different answers to how many locations sit in that hex, and reconciling the gap is most of the work before a priority list goes out.
What the location fabric counts
The CostQuest fabric that underlies the National Broadband Map is a parcel-and-structure model. It starts from address points, parcel boundaries, and building footprints, then infers a broadband serviceable location wherever those layers suggest a habitable or business structure sits. It's rebuilt on a biannual cycle, and ISPs report service availability against it.
That inference step is where the trouble starts. A mobile home added last spring may not have a permit record yet. A parcel split into three buildable lots still shows as one address point. A farm with two houses and a bunkhouse on one tax parcel might register as a single location, or three, depending on how the county's assessor data structured it. None of this is the fabric vendor doing a bad job. It's a national model built from county-level inputs of wildly uneven quality, refreshed on a schedule that can't keep pace with what's getting built in a fast-growing exurb or torn down in a shrinking one.
Where ground-truth counts diverge, and why it matters for the list
Field verification, a windshield survey, a locate crew, or a high-res imagery pass, counts the structures standing on the ground, independent of what a parcel database implies should be there. The common splits:
New construction outruns the fabric. A subdivision permitted eighteen months ago can have forty occupied homes and still show as six scattered BSLs if the fabric hasn't absorbed the plat update.
Seasonal and vacant structures get counted inconsistently. A fabric built from tax assessor data may include a hunting cabin as a year-round BSL, or drop it entirely if it's classified non-residential.
Multi-unit and accessory dwellings undercount. ADUs, converted garages, and informal mobile home placements rarely have their own address point, so the fabric collapses them into the parent structure's single location.
Demolished or never-built locations overcount. Parcels platted a decade ago for a subdivision that stalled still generate BSLs nobody will ever connect.
For a challenge process filing, a few of these per hex is manageable with narrative evidence. For a statewide rural build priority ranking or a subsidy bid where you're allocating a fixed pool of funds across hundreds of hexes, that same drift compounds into a ranking that no longer reflects where people live.
Reconciling the two numbers
Reconciling the two counts means giving the fabric a check against an independent count built from the ground up, at a cadence tight enough to catch what's changed since the last BDC cycle. That means an independent settlement count, not another parcel-inference model layered on the same stale assessor data, and not a population grid that smooths a cluster of twelve new homes into a county-level average.
A settlement layer built from an annual high-res satellite pass does this by mapping every settlement cluster that's visible on the ground, with cluster size and location attached, so you can line it up against the fabric's BSL count hex by hex and see exactly where the two diverge before you submit a challenge or finalize a build priority list.
That comparison won't resolve every discrepancy on its own, since a satellite pass can't tell you whether a structure is occupied or how many units are inside a multi-family building. But it closes the biggest gap: knowing which hexes have new construction the fabric hasn't caught up to yet, and which have fabric-only locations with nothing built on the ground.
If your next allocation round depends on getting that gap right, it's worth seeing what an annual settlement count looks like against your target region before you file.