Reference
Backflow records availability index
Every entry below comes from a records request that was filed and answered, or from an open dataset that was pulled and read. It records what each jurisdiction actually released, what it charged, which fields arrived, and where the request stopped. The unsuccessful ones are the useful half.
Compiled from requests filed 12 June 2026 and their responses through 17 July 2026. Fees, portals and retention practice change; treat every line as a dated observation, not a standing rule.
Key Two tests
What "available" has to mean
A jurisdiction is only useful to a testing contractor if it clears both tests below. Most of the entries in this index fail one of them, and several fail both. That is the whole point of publishing it.
Test one — is there a compliance state?
A list of devices is an inventory. A work list needs a due date or a status field the agency itself maintains. Several utilities publish the first and hold nothing resembling the second.
Test two — is there a private market?
Where the utility owns and tests the assemblies itself, there is no outside work to find, however good the data is. Who performs the test is a question about the jurisdiction, not about the file.
§01 The index
Twelve jurisdictions, and what each one did
Jurisdiction names link to the agency's own records channel or open-data page, so any line here can be checked at the source. Nothing in this table is redistributed data — it describes what was released, not what it contained.
| Jurisdiction | Compliance state | Outcome | Verified |
|---|---|---|---|
| Lubbock, Texas | Yes — a maintained Current/Overdue field |
Released in full at no charge, after agreeing to the city's standard personal-information redactions. 39,909 device rows across 31,749 addresses. | 2026-06-23 |
| Columbus, Ohio | Yes — a maintained Status field with a test-due date |
Released at no charge through the city's request portal. 37,658 device rows across 27,474 addresses. | 2026-06-17 |
| Gainesville, Florida | Yes — the utility's own overdue flag against a stated interval | Released at no charge as a raw export, narrowed to commercial devices at the utility's request. Accuracy expressly disclaimed by the utility. | 2026-06-23 |
| New York City | Partial — violation and hearing status, not device-level test dates | Published continuously as open data. No request needed. Covers cited buildings rather than an assembly inventory. | 2026-07-17 |
| Clearwater, Florida | No — the agency states it keeps none | Released at no charge, 14,943 device rows over 31 columns. The next-test and compliance columns contain the city's written statement that no responsive record exists. | 2026-06-18 |
| Port St. Lucie, Florida | No — export described as device attributes only | Records located at no charge, retrieval declined. The city describes an export with no test or due dates, and states it performs its own testing. | 2026-06-22 |
| Greensboro, North Carolina | Stale — renewal dates long past, unmaintained | Was an open dataset, 13,663 device rows. The dataset owner confirmed it is no longer maintained and asked for its removal from the portal. Not used. | 2026-06-15 |
| Lakeland, Florida | Not reached | A special service charge of $185.40 survived narrowing, so the request was withdrawn. The city separately reports owning and maintaining about 99% of the preventers in its territory. | 2026-06-23 |
| Medford, Oregon | Not reached | Estimated at $112, revised to $56 after narrowing, against a stated one-hour minimum that would not go lower. Withdrawn. | 2026-06-22 |
| Salem, Oregon | Not reached | A special service invoice of $29.16 was issued before production. Withdrawn. | 2026-06-17 |
| Frisco, Texas | Not reached | Records located, cost letter of $9.72 issued. Left unpaid; the city's terms withdraw an unpaid request automatically after sixty days. | 2026-06-19 |
| Texas, statewide tester roster | Not applicable | Refused. The state environmental agency withholds licence-holder contact details under 2023 legislation; only a limited group search and an opt-in business-contact file remain available. | 2026-06-15 |
Scroll the table sideways for the remaining columns.
Requests filed in several further jurisdictions had no settled outcome on the compile date and are left out rather than listed as pending. Four of the twelve above released a usable file; three of those four hold a real compliance state.
§02 Fields released
What the four delivered files actually contain
Column names as the agency supplied them. This is the part nobody publishes, and it is the part that decides whether a file can be worked. Row counts are as delivered on the dates above.
| Jurisdiction | Columns supplied | Rows |
|---|---|---|
| Lubbock, Texas |
Address, Manufacturer, Model No,
Size, Last Inspection, Next Inspection,
Current/Overdue
|
39,909 |
| Columbus, Ohio |
Site Address, Type, Size,
Last Test, Test Due, Status
|
37,658 |
| Clearwater, Florida |
31 columns covering asset number, set date, service type, series, size, type,
manufacturer, the three check positions, bypass detail, and a fully parsed
address. Next Test Date and Current Compliance Status
are present as headings but carry the refusal sentence in place of a value.
|
14,943 |
| Gainesville, Florida |
Device Serial #, Device Type, Device Model,
location name, full address, Set Test Month,
LastTestDate, overdue flag
|
4,847 |
Scroll the table sideways for the remaining columns.
The Clearwater result, stated exactly
All 14,943 rows arrived, free, within six days. In the two columns that would make the file workable, the city wrote a sentence instead of a date: that no responsive record exists for a next test date, because testing occurs year-round and is not scheduled in advance, and that compliance status is not maintained as a standalone record.
A separate check on that file found the most-recent-test-date column clustering on a handful of values across rows with different set dates, which is what an export-time stamp looks like rather than a real test history. Both findings are why Clearwater is marked as holding no compliance state despite a generous response.
§03 Failure modes
Five ways a jurisdiction turns out not to have records
Each of these was met at least once in the set above. Anyone filing their own requests will meet them in roughly this order.
- The fee wallLakeland, Medford, Salem, Frisco
- A special service or staff-time charge attaches before production. Narrowing the scope sometimes halves it and sometimes does nothing, because several agencies assert a minimum billable increment that no narrowing can go below. Four of the twelve ended here.
- No compliance stateClearwater, Port St. Lucie
- The agency holds a device inventory and nothing that says when the next test is due or whether the last one passed. Often because testing runs continuously rather than on a scheduled cycle, so no due date is ever computed.
- The utility tests its ownLakeland, Port St. Lucie
- The city owns and maintains the assemblies, so no outside contractor is hired and the list describes work that is already covered. This disqualifies a territory regardless of how good the data is, and it is worth asking before filing anything.
- Statutory withholdingTexas statewide roster
- Some categories are closed by statute rather than by discretion, which means there is no appeal and no narrowing that helps. Certification and licence rosters are the common case; some jurisdictions also treat utility asset locations as protected infrastructure information.
- Publisher withdrawalGreensboro
- An open dataset exists right up until the publisher notices it is being used for commercial solicitation. Here the owner confirmed the file was unmaintained and asked for it to be taken down. An open portal is not a durable supply, which is why every pull on this site is dated.
§04 Use and limits
How far this index can be trusted
It is a record of twelve specific interactions on specific dates. It is not a survey, a ranking, or a statement about how any agency will behave next time.
- Fee schedules, portals and staffing change. A charge quoted in June 2026 says nothing certain about the same request filed later.
- Outcomes reflect one particular scope: device records with a compliance state, address level, no account-holder details. A different scope gets a different answer.
- Where a request was withdrawn over a fee, that is a decision made here, not a refusal by the agency. Those records remain obtainable by anyone willing to pay.
- No agency data is republished on this page. Row counts and column names describe files; they are not the files.
- BackflowLeads is not a government agency, is not affiliated with or endorsed by any agency, and has no non-public access to any system named here.
Corrections
If you work for one of these agencies and a line is wrong, out of date, or describes a practice that has since changed, send the correction and it will be amended with the date it changed.
Not a government document
§05 Next
Where this leads
The index says which jurisdictions hold something worth having. What a compiled file looks like, how it is checked and what it cannot tell you are covered separately.