If you buy a list, you are buying the same rows your competitor bought, from the same vendor, refreshed on the same schedule. The vendor's incentive is coverage, not truth, and their business depends on selling those rows more than once. There is no version of that where you end up with an advantage.
Public record is different in one specific way that matters more than any other: it is timestamped. A list tells you who exists. A filing tells you what just changed and when. The first is a directory. The second is a signal, and a signal has a shelf life you can actually work against.
What New Jersey actually publishes
You do not need exotic sources. The state's ordinary machinery generates most of what matters, at three levels of government, on a predictable cadence.
County. New Jersey has 21 counties, each with an elected Surrogate whose office handles the probate of wills and the administration of estates. Deeds and mortgages are recorded with the county clerk or register. Sheriff's sales — the terminal stage of a foreclosure — are noticed publicly by the county sheriff before they happen.
Municipal. New Jersey is unusually fragmented: 564 municipalities as of this writing, each with its own tax collector, construction office and governing body. Every one of them is required to hold an annual sale of delinquent property taxes under the state's Tax Sale Law. Every one of them issues construction and demolition permits. Every one of them adopts ordinances, and a large share publish their codes online.
State. Business entity registrations, professional licensure, and the Superior Court's handling of foreclosure — New Jersey is a judicial-foreclosure state, so a foreclosure is a court proceeding with a docket, not a private sale.
None of that is a secret. All of it is boring. That combination is exactly why it works.
I am describing the category, not handing over the pipeline. Which of these Auctus actually reads, at what cadence, and how signals are weighted against each other is disclosed under engagement. The method is not a secret; the configuration is the product.
Why timestamped beats complete
Three structural properties make record-derived flow behave differently from a purchased list.
It has a date on it
Every record carries the moment something changed — a filing, a transfer, a delinquency, a permit pulled. That means you can rank by recency rather than by whoever the vendor happened to enrich most recently. Recency is the only thing that reliably predicts whether a conversation is timely, and it is the one field a purchased list almost never has honestly.
It is adversarially complete
A vendor's coverage is a business decision. A statutory filing requirement is not. If the law says a thing must be recorded to have legal effect, then the record is complete in a way no commercial dataset ever is — not because anyone is being generous, but because the alternative is that the transaction doesn't count.
It is bounded by jurisdiction, which is a gift
Public record forces you to pick a geography and go deep, because the plumbing is different in every county. Most people experience that as friction. It is actually the moat: a competitor who wants your coverage has to rebuild your integrations town by town, and there is no shortcut they can buy. Depth in one state beats a shallow national list, and it is not close.
Anyone can buy the same rows you bought. Nobody can buy the four months you spent learning how one county's clerk formats a name.
The four ways this fails
I would rather tell you why people abandon this than pretend it is easy. All four failure modes are real and all four are why most firms quietly go back to buying lists.
- Identity resolution. The record names a party; it does not name a person you can call. "J. Smith," "John Smith Jr." and a trust with a lawyer's address may all be the same owner, or three different ones. Getting this wrong is not a small error — it is the difference between a well-timed call and an offensive one. This is where most homegrown attempts die.
- Staleness. A distress or life-event signal decays. Past its useful window it is not a weak lead, it is an actively bad one: the situation has usually already resolved, and arriving late marks you as someone who bought a list. A pipeline without an explicit expiry rule will keep shipping stale rows forever, because nothing in the system objects.
- Format drift. County systems change without notice — a portal is redesigned, a field is renamed, a search form gains a parameter. Anything that silently returns zero rows instead of failing loudly will do so for weeks before anyone notices the pipeline has been dry. Fail loudly or don't build it.
- Volume mistaken for value. The record will happily give you more rows than you can work. Ranking is the entire job. A pipeline that surfaces everything has moved the bottleneck rather than removed it, and your team will stop opening it by week three.
The part most people skip
A significant share of the most actionable signal in this category comes from someone's bad day — an estate, a delinquency, a foreclosure docket. I think it is worth being direct about that rather than dressing it up in the language of "opportunity."
Two things are true at once. These filings are public by deliberate legislative design, and using them to make a fair, timely, clearly-explained offer to someone who has decisions to make is legitimate work — the alternative is often a worse outcome for them, not a better one. And: the same data used badly is predatory, and the industry has enough operators proving it.
The practical version of that principle is a set of rules the pipeline enforces rather than a paragraph in a values deck. Consent gates on any channel that requires one. Suppression that actually suppresses, permanently, on first request. No pretending to be anything other than a buyer or a broker. No volume tactics on a distress cohort. If a system can't enforce those without a human remembering to, it will eventually not enforce them.
Where to start if you're building this yourself
Take one county, one signal type, and one 90-day window. Resolve every record to a named party and a verified channel by hand — actually by hand, all of them. Two things come out of that week: a realistic conversion rate for that signal, and a precise list of the places automation will break. Both are worth more than a year of reading about data strategy.
Then automate the parts that were tedious rather than the parts that were hard. The tedium is the sourcing and the shape-checking. The hard part — deciding whether this particular signal on this particular property is worth a call — should stay expensive and human for a lot longer than you expect.
County and municipality counts are as of August 2026 and do change — New Jersey's municipal count has fallen through consolidation more than once in the past two decades. Statutory descriptions here are general orientation for operators, not legal advice; verify current requirements before relying on them.