permits.llc
New Jersey Permit Data

38% of New Jersey's Permit File Is Finished Work

By the permits.llc team · Last reviewed September 8, 2026 · Optimal window: Weeks 1–4

New Jersey publishes two different records under one heading. 396,315 of our 640,741 New Jersey rows are construction permits, which mark a job starting, and 244,426 are certificates, which mark one finishing. Reading the file without separating them is the most expensive mistake available in New Jersey permit data, and almost everyone makes it.

TL;DR

  • New Jersey's 640,741 records are 396,315 permits and 244,426 certificates.
  • Certificates mark completion. 110,965 addresses appear only as a certificate.
  • The date pair is not broken. On certificate rows it measures build duration.
  • Statewide completion clock: 18 days at Q1, 41 at the median, 92 at Q3.

What is the difference between a New Jersey permit and a certificate?

A permit authorises work to start. A certificate records that the work finished.

The Department of Community Affairs writes the state Uniform Construction Code at N.J.A.C. 5:23, and municipal construction officials administer it. When a project passes its final inspection, the construction official issues a certificate of occupancy. For work that required a construction permit but does not create or change an occupancy, a boiler replacement or a service upgrade for instance, the official issues a certificate of approval instead. N.J.A.C. 5:23-2.23 governs both, along with the temporary certificate of occupancy that lets a substantially complete building be used while punch-list items remain.

Each of those certificates enters the public record as its own row. It is not an update to the permit. It is a separate record, with its own number, its own dates and its own declared cost.

That is why the file splits the way it does. 237,718 of the certificate rows sit under a building permit type and 6,708 under demolition, which is the closeout side of the same work covered in the 45-day rebuild clock on New Jersey teardowns.

How big is the split, and where does it show up?

Statewide it is 38.1% certificates. County by county it swings by a factor of nearly three.

CountyPermitsCertificatesCertificate shareMedian days to certificate
Monmouth41,67625,21937.7%44
Bergen43,80222,96534.4%38
Morris25,25522,46547.1%41
Middlesex40,79321,18234.2%41
Ocean39,94620,41333.8%45
Somerset15,70717,98053.4%41
Camden18,26313,97243.3%38
Burlington24,12311,59632.5%35
Hunterdon7,7256,47245.6%59
Hudson9,1152,17319.3%61

Somerset County files more certificates than permits. Hudson files four permits for every certificate. Same state, same code, same forms.

The gap is a reporting decision rather than a construction one. Some municipal offices push their closeout records to the state feed as diligently as their permit records, and some do not. It follows that a county-level volume ranking built on the mixed file is partly a ranking of local clerical practice, a hazard that also runs through the block and lot buried in every New Jersey description field.

Why does New Jersey's date pair look broken?

Because it was being read across two record types at once. Here is the arithmetic.

Every one of the 396,315 permit rows carries an identical application and decision date. Not most of them. All of them. On a permit row New Jersey writes one event into both fields, so the gap is zero by construction and can never be anything else.

Add the 2,324 certificate rows that happen to carry the same date in both fields and the total is 398,639. That is exactly the count of identical-date records in the file, to the record.

The county pattern falls out the same way:

CountyRecords with identical datesRecords that are permit rows
Hudson81.1%80.7%
Mercer74.9%74.8%
Essex70.4%70.2%
Morris53.1%52.9%
Warren49.6%49.3%
Somerset46.8%46.6%

Six counties, six matches inside half a percentage point. The variation that looks like towns recording dates differently is the certificate share, and nothing else. Our own earlier piece on New Jersey's 20 business day permit rule reported the bimodal shape correctly and read the cause as municipal filing habit. The record-type split is the better explanation, and it is a happier one, because it means the second mode is measuring something real.

What does the certificate date actually measure?

Construction duration. Permit issued to final inspection passed, in calendar days.

Across the 244,426 certificate rows the median is 41 days. The first quartile is 18 and the third is 92. The 90th percentile is 189. In buckets: 97,127 jobs close within 30 days, 82,541 between 31 and 90, 57,948 between 91 days and a year, and 4,120 run longer than a year.

Use group moves that clock more than geography does.

Use groupCertificatesMedian days90th percentileMedian declared cost
Restaurants, night clubs, dance halls1,22764253$8,825
Mercantile buildings1,36163215$10,975
Business uses11,09256217$9,365
Multiple family dwellings, dormitories9,18550238$4,000
International Residential Code195,08640182$5,000
Accessory buildings and structures11,65138211$2,100
One and two family, 5 lodgers or less7,20136199$4,941

No other state file we hold supports this cut. It exists in New Jersey because the closeout record is published as its own row with its own date, which turns an apparent data-quality flaw into the one duration measure the product has.

Which trades should be reading certificates instead of permits?

The ones that arrive after the building official leaves.

A permit is a signal about the homeowner, not the contractor, and the two record types catch that homeowner at opposite ends of the project. A permit row says a decision has been made and money is committed. A certificate row says the space is finished, usable and empty, and that the general contractor has gone.

Flooring, painting, landscaping and hardscape, security and alarm, appliance and fixture supply, interior design and window treatment, moving services and pool and spa work all sell into that second moment. For those trades the permit date is six weeks too early in a house and four months too early in a restaurant. Working the permit file and wondering why the timing never lands is a symptom of reading the wrong row.

Trades that need to arrive first, mechanical and plumbing replacement, envelope work, site and septic, generator and charger installation, want the permit rows and should filter the certificates out. That is a legitimate use of the status field. Doing it without knowing you are doing it is not.

The cost band sorts the certificates further. 81,386 of them declare under $2,500, which is what a service closeout looks like. 28,260 sit between $25,000 and $250,000 and 3,999 clear $250,000. That top band is 1.6% of the certificate file and it is where a property genuinely changed shape, which pairs with the cohort logic in what New Jersey's empty square footage box tells you.

What can this data not tell you?

Quite a lot, and the limits are worth stating plainly.

The status field carries two values, Permit and Certificate, and nothing more. It does not distinguish a certificate of occupancy from a certificate of approval, and it does not flag a temporary certificate. A $700 certificate is almost certainly a certificate of approval and a $400,000 one is almost certainly an occupancy certificate, but that is inference from the cost, not a field.

There are no names. All 640,741 New Jersey records carry zero contractor names and zero owner names, certificates included. A certificate tells you an address finished a job on a date, and finding the person behind it remains your work.

366 certificate rows carry a decision date earlier than their application date, which is impossible and which we exclude from every duration figure above. Six records carry a decision date later than the file itself runs, three of them beyond today, with the furthest at 2027-01-27. On 640,741 records that is noise, but it is there.

The file itself ends on 2026-08-07. That is the most recent New Jersey application date we hold, and it is a real constraint on anything time-sensitive rather than a rounding detail. Coverage figures are published on the New Jersey county coverage pages, and permit-level date alerts for states with a current feed are set up through permit alerts.

How permits.llc fits in

New Jersey is 640,741 records, and the first useful thing you can do with it is cut it in half on a field most buyers never look at. 396,315 starts. 244,426 finishes. Two different sales motions, one download.

Take the New Jersey permit download and sort on the status column before you build anything. Then pull your municipalities from the New Jersey county coverage pages and set your follow-up cadence off the local completion median rather than the statewide one, because 21 days in Salem and 61 in Hudson are not the same business.

Frequently asked questions

Get started

Download the free 2025 New Jersey permit dataset to see the real records, or set up daily alerts for the permits that trigger work in your trade.

Related playbooks