Coverage: what is live, what is not, and what each source withholds

EarlyFloor sells one metro. This page exists so you can find out that your metro is not covered before you give anyone an email address, and so you can see which limits are ours and which belong to the permitting authority.

EVERY FIELD LIST BELOW RE-CHECKED AGAINST THE LIVE ENDPOINT ON 2026-07-30

The matrix

Metro / authorityStatusPermits publishedContractor phoneTrade scope in textPermit status field
Miami-Dade County
miamidade_permit_data
LIVE — this is the feed yes, daily yes — ContractorPhone no trade code; free text only no status field exists on this layer
City of Miami (one of 34 municipalities)
Building_Permits_Since_2014
LIVE — the declared-scope file yes, daily none — 41 fields, not one is a phone yes — WorkItems names WOOD/LAMINATE/TILE/VINYL yes — BuildingPermitStatusDescription, IsPermitFinal
Orlando (City of Orlando)
ryhf-m453
read by the pipeline, not sold yes, daily yes — contractor_phone_number, populated on 746 of the 1,362 permits issued since 2026-08-01 (checked 2026-08-19). This cell used to read “none”: our field map never requested the column, and we wrote our own omission down as a fact about the city. Corrected. Orlando is still not sold — see below for the real reason. no yes — application_status
Tampa Bay not built no source evaluated yetunknownunknownunknown
Broward not built no permit source evaluated yetunknownunknownunknown

Miami-Dade is 34 municipalities, and the two sources do not overlap much

This is the most important limit on the page and it is structural, not a formatting quirk.

How far apart, measured. Over the 14 dated files on disk (2026-08-06 to 2026-08-19): of 91 distinct phone-feed permits, 0 carried the county's own declared floor scope; of 26 distinct declared-scope permits, 0 carried a phone. The two sets did not overlap on a single row. This is not a tuning problem, it is structural — the phone comes from the county layer, whose permit text never names a floor covering (its categories read "MBLD INTERIOR REMODEL", "BLDG RENOVATION/ADD", and the like), and the floor scope comes from the city layer, which publishes no phone on any of its 41 fields. So a single row cannot, today, be both flooring-confirmed and callable. The phone feed is a remodel-permit prospecting list; the declared-scope file is a small, name-only, mostly-commercial confirmation set. They are sold together but they are not the same rows, and neither one is "a homeowner who wants floors and a number to call".

Closing that gap means adding the municipal portals that run their own permitting — Miami Beach, Hialeah, Coral Gables, Doral among them. None of those are built. If someone tells you EarlyFloor covers "Miami", that is what it means and does not mean.

Why Orlando is read but not sold

Corrected 2026-08-19. This section used to say the Orlando dataset "publishes no phone number on any record — verified against its own live schema, not inferred". That was wrong, and it was wrong in a way that flattered us: it turned our own missing configuration into a fact about the city. The City of Orlando dataset ryhf-m453 does publish contractor_phone_number, and on 2026-08-19 it carried a number on 746 of the 1,362 permits issued since 2026-08-01. What was true is that our field map never asked for that column, so the pipeline read an empty string on every Orlando row. The map was fixed the day this was found. Confirm the schema yourself — the key names are in the raw record:

one Orlando permit record, raw

Orlando is still not sold, and here is the real reason. With the phone read, the 2026-08-19 run produced 2 Orlando flooring rows. One commercial and one whose occupancy that source does not resolve; neither is residential. Earlier that day this paragraph said 4 — and two of those four were fire-alarm sub-permits, a fire-protection contractor filing against somebody else's tenant improvement, which is not work a flooring crew can quote. Those two rows were themselves a gate failure: the sub-permit check only understood Miami's permit-number convention (trade letter at the end), while Orlando puts the discipline in front (FIR/ELE/MEC/PLM), so every Orlando sub-permit walked through it. Fixed the same day; sub-permits no longer ship in any Orlando file. Re-measured 2026-09-17, and it has got worse, not better: the eight nightly Orlando pulls from 2026-09-09 to 2026-09-16 each produced 0 rows, and the last run that produced a row of any kind was 2026-09-08, with one. Two commercial rows a run was already not a feed worth putting in front of anyone; zero is not a feed at all. So Orlando stays in the pipeline, stays off the sample page, and the download endpoint refuses it by name if anyone asks for it. This figure is hand-written on this page and therefore dated — the nightly files in out/permits/orlando/ are the record, not this sentence.

If your metro is not on this list

There is nothing to sell you and no waiting list that implies otherwise — and since 2026-08-21 there is nothing to sell you in Miami-Dade either: EarlyFloor is published as a free showcase and has no paid feed at any price. The form on the home page has a metro field; putting yours in it records demand, and that is all it does. It does not reserve anything, it does not schedule anything, and no date is promised.

Ground truth: EarlyFloor has never had a paying subscriber in any metro, and as of 2026-08-21 it is not going to be sold in one. Nothing on this page describes a customer base, because there is not one.

See the Miami rows that do exist →