Methodology
How the numbers get built.
Three layers, never blended: the FCC filing per market, the county layer, and RCLX’s own live address checks. What each one counts, where it comes from, and the test a number has to pass before it publishes.
Updated September 23, 2026 · Market data rebuilt September 23, 2026
Three layers
Each number has one source.
Every figure on the site names its layer and its date. We never multiply one layer by another.
Layer 1 · FCC filing
Who has wires in town.
The FCC Broadband Data Collection summary as of December 31, 2025, per market. Residential units only: fiber from any provider, cable, cable homes that can’t advertise 100 Mbps up, and homes with nothing wired at 25/3 or better. It measures the competition. It never speaks to what’s sellable.
See it on any market pageLayer 2 · County layer
Where the footprint reaches.
Fiber-only filings from the FCC BDC J25 (29 April 2026) release, read by county, built May 11, 2026. Counties the federal file hadn’t caught yet come from published service-area pages and carry a status, not a number.
County list, state by stateLayer 3 · Live checks
What a rep can sell.
Real home addresses from the county address register, run through the live availability check. Each one comes back orderable, on record but not open yet, or no plant. Every rate carries its range.
The measured tableWhy three? The federal filing lags the network. A build newer than the filing reads as nothing there, even on streets reps sell every day. The live check closes that gap. It can’t replace the filing, though: it says nothing about cable.
The publish test
A number earns its spot.
One function decides it, in code, for every market. The page reads its answer and never re-decides.
Sixty checks first.
Under 60 valid checks, nothing publishes. Not a rate, not a zero.
A rate needs a shape.
At least 10 orderable addresses, and a 95% interval no wider than 15 points. A lucky streak doesn’t clear it.
A zero costs more.
It prints only as a ceiling at or under 5%, and only when outside evidence agrees: no completed RCLX order there, no matching fiber in the FCC filing, and a passed control.
Everything else waits.
The market keeps its page and its FCC picture. The measured line stays off until the sample settles.
The control is the check on the checker. We re-run addresses where we’ve already closed sales. If those don’t come back orderable, no zero from that region publishes. In states where we have no sales yet, the control is borrowed from one that has them, and the data says so.
Samples are spread across the whole county address register, so a small sample can’t sit in one corner of town. Samples drawn before September 2026 are weighted by area before they publish.
What stays private
Shape, never the ZIP list.
Which ZIP codes hold the plant is the list an agency takes to request turf. We don’t publish it. Where the sample is deep enough, a market page says how concentrated the plant is. That’s it.
The market file is public at /data/city-stats.json, with both market layers side by side and each one dated.
Providers report their own coverage to the FCC and those reports can be challenged. Advertised speeds are ceilings, not what a given house gets. Nothing on this page is a promise of territory, earnings, or service at any specific address.