BuildDocket
Live · NYC Open Data ·

Show your work

Every number we publish, and where it came from.

Nothing on this page is typed in. Each figure is read at page load from the same database that builds the morning digest, and each carries its own data vintage — so if a nightly run fails, you see stale numbers stamped stale rather than fresh numbers that are not.

City records ingested
Datasets read
Trade divisions scored

The feeds

Fourteen datasets, refreshed nightly.

Record counts are what is actually in our database right now, not what the City's portal advertises.

Dataset

The divisions

What is open right now, by trade.

Two numbers matter and they are different: the pool of violations open today, and the flow of new ones each month. A big pool with a thin flow is a market you work through once; a steady flow is a subscription that keeps earning.

A live sample

Real rows, scored, contacts withheld.

Straight out of the ranking, with the resolved contact masked — that part arrives in the digest.

AddressDivisionDeadlineValue ContactScore

Method

How a city record becomes a lead.

1

Ingest

Nightly pulls with a watermark per dataset, so a feed that stops publishing shows as stale instead of quietly repeating itself.

2

Keep only deadlines

Records without a correction date, filing date or unpaid balance are dropped. Most of what the City publishes is paperwork, and paperwork does not make anybody hire a contractor.

3

Segment by trade

The order text decides the division, not the permit type — a leak over a ceiling is roofing or plumbing depending on what the inspector described.

4

Resolve the contact

Owner of record, or the managing agent registered with HPD where one is on file, cross-checked against the tax roll. Each carries a confidence band; we would rather show you a low band than invent a name.

5

Score and rank

Deadline proximity, violation class, declared value and how reachable the contact is. You get the top of the list, not the whole database.

Honesty section

What this data cannot do.

Declared values are not quotes. They are the valuations filed with the City on work at these buildings. Useful as a scale, useless as an estimate.

Not every contact resolves. Some buildings are held by entities with no registered agent. Those leads carry a low confidence band and a route in rather than a phone number — we tell you how to reach them, we do not pretend to have found a number.

Two divisions have no money math. Mold and lead have too few declared values to publish a median honestly, so we do not publish one. They are still delivered; the arithmetic on those pages would be a guess.

New York City only, today. The method transfers to other cities with open violation data. It has not been done yet, and we are not going to sell it before it is.