You can buy the numbers or you can build the list yourself, and the choice is mostly about whose time you are spending. What it is not about is quality. Both routes end at the same place, which is a name from the county tax roll matched to a phone number from a data broker’s file, and neither route involves anyone picking up that phone to confirm the person on the roll still answers it.
Hold onto that, because it changes how you read every vendor’s pitch and every free method you will find. There is no such thing as a verified circle prospecting list until you have called it. What you are buying, or building, is a set of educated guesses, and the work of turning guesses into a farm list happens on the first pass through the dialer, no matter where the numbers came from.
What a farm phone number actually is
Start with the address, because every farm and every circle starts there. The county assessor or recorder holds a record for each parcel: the owner’s name, the mailing address, the situs address, the last sale date and price. That record is public, it is usually free or close to it, and it is the most reliable piece of data you will touch in this whole process. If the mailing address matches the situs, the owner probably lives there. If it does not, you are looking at an absentee owner and the person who answers the door is a tenant.
The tax roll does not contain a phone number. The phone comes from somewhere else entirely: a data aggregator that has assembled, from credit header files, utility connections, carrier records, directory listings, marketing databases and old survey forms, a file that says this name at this address has been associated with these phone numbers. A vendor takes the owner name and mailing address from the roll, looks it up in that file, and appends whatever comes back. That is the entire operation. It is called skip tracing when it is done one record at a time and match-and-append when it is done to a whole subdivision, and the two are the same thing at different volumes.
The match has a confidence score, which most vendors show you and most agents ignore. A high-confidence match on a common name at a long-held address is usually right. A low-confidence match on a house that sold eighteen months ago is often the previous owner. The score is the only quality signal you get before the call, so keep it in the export.
The buy route
Buying means one of three things, and they carry different price tags.
The first is a neighborhood or geographic search built into a dialer or lead platform, where you draw a boundary on a map and the tool returns the homes inside it with whatever phones its data partner carries. It is fast, it is usually bundled into a monthly subscription, and the data is refreshed on the vendor’s schedule rather than yours. Ask, before you rely on it, when the phone file was last updated and whether the export shows the match confidence and the line type.
The second is per-record skip tracing. You upload a list of names and addresses, most often straight from the assessor export, and pay per row for the append. Prices advertised in the market run from a few cents to around half a dollar per record depending on volume and on how many fields come back, and vendors quote match rates in the seventies and eighties. Read that number carefully. A match rate is the share of addresses that received a phone. It is not the share of phones that reach the owner, and the second figure is always lower than the first.
The third is a title company farm package. In many markets a title representative will pull a subdivision for you, with the phones their data provider carries, as a courtesy. It is a legitimate starting point and it carries every caveat above, plus one more: you do not control the refresh, so treat the phones as a snapshot from whenever the package was built.
The build route
Building means pulling the assessor data yourself and doing the phone work by hand, which is free in dollars and expensive in hours.
Most county assessor and appraiser sites let you search by subdivision, street or map book page and export the results. Some counties charge a small fee for a bulk file. Some make it tedious enough that a title rep or the MLS tax module is the easier door to the same data. Either way, the output is the owner roll for your boundary, and it is worth having in that raw form regardless of which route you take for the phones, because it is the master record the farm is built on.
Then the phones. Manual lookups through free directory sites work for perhaps a third of the roll, they take a few minutes per household, and the numbers they return skew toward landlines that older owners have kept. A thousand-home farm at three minutes a lookup is fifty hours, which is why almost everyone who starts on the build route ends up sending the assessor export to a per-record vendor for the append and keeping the build route for the address side only. That hybrid is the sensible default: build the roll, buy the match.
| Source | What you get | What it costs | Where it goes wrong |
|---|---|---|---|
| County assessor or recorder | Owner name, mailing and situs address, sale history. No phones | Free to a small bulk fee | Ownership lags a sale by weeks. No phone at all |
| Dialer or platform neighborhood search | Homes in a drawn boundary with appended phones | Bundled into a subscription | Refresh schedule is the vendor’s. Confidence and line type sometimes hidden |
| Per-record skip trace | Phones and often emails appended to your own roll | A few cents to about fifty cents per row | Match rate is quoted, accuracy is not. Old owners on recently sold homes |
| Title company farm package | The roll plus whatever phones the title data carries | Usually free as a courtesy | A snapshot from whenever it was built |
| Manual directory lookups | A phone for some households | Hours, not dollars | Covers a fraction of the roll and skews to landlines |
The compliance work nobody sold you
Here is where the ranking guides stop, and where the actual risk lives, because a list of homeowners who have never contacted you is cold telemarketing in its purest form and the rules apply to every number on it.
The registry scrub is yours to do, on a 31-day clock. Vendors will tell you the list is “DNC scrubbed.” Sometimes it is. It does not matter, because the safe harbor for an accidental call to a registered number under 47 CFR 64.1200(c)(2)(i) belongs to the caller who has written procedures, trained staff, an internal do-not-call list, and a version of the National Registry obtained no more than 31 days before the call. A vendor’s scrub from six weeks ago does none of that for you. Access to the registry is free for up to five area codes and, from October 2025, $82 per area code beyond that, which for a single farm is usually nothing. The registry may be used only to prevent calls, never to build a list, and the FTC says so in plain words.
Nearly every number on the list is a cell phone. The CDC’s most recent wireless survey puts 78.7 percent of adults in wireless-only households as of the second half of 2024, with under one percent in landline-only homes (NCHS, Wireless Substitution). Whatever your vendor labels the line type, plan the farm as a cell list. That matters because 47 CFR 64.1200(a)(1)(iii) prohibits autodialed calls and artificial or prerecorded voice calls to a cellular number without prior express consent, and a homeowner on a farm list has given you none. In practice that rules out prerecorded messages and voicemail drops to the farm entirely. Whether a particular dialing mode counts as an autodialer under the statute is a question for your attorney, because the case law has moved more than once; the prerecorded and artificial voice restriction has not moved, and it is the one agents trip on.
Numbers get reassigned, and the file you bought does not know. Carriers report permanently disconnected numbers each month to the FCC’s Reassigned Numbers Database, and callers can query a number against it before dialing. The formal safe harbor in paragraph (m) is written for callers who had consent from the previous holder of the number, so it does not protect a cold farm call the same way. Use the database anyway, as hygiene. A number that was disconnected after your vendor’s file was built now belongs to a stranger who has no connection to the address you are calling about, and the query is cheap next to the complaint.
Internal do-not-call requests are kept for five years. When a neighbor says “don’t call me again,” the number goes on your internal list before the next dial and stays there, and paragraph (d)(6) says that request is valid for five years. On a farm you will be calling the same street next quarter and the quarter after that, which is exactly the point made in farming versus circle prospecting: a missed request in a farm is an error you will repeat on a schedule.
The first pass is verification, whatever you paid
Because no source verifies, the first time through the list is not really prospecting. It is a data pass that happens to produce a few conversations. Set it up that way on purpose.
Every dial gets an outcome that says something about the number, and the outcomes that matter most on the first pass are the boring ones: wrong number, disconnected, tenant not owner, do not call, and confirmed owner. A confirmed owner who declined the note is still a win for the list, because you now know the number is right. Expect the wrong-number and disconnected outcomes to run somewhere between a fifth and a third of a purchased farm list on the first cycle, higher on a subdivision with recent sales and lower on a stable retirement community, and treat anything much better than that as a sign your vendor’s confidence scores were strong rather than as the norm.
Then prune. Disconnected numbers come out. Wrong parties come out, and if the address is worth it, go back to the vendor for a second match on that owner. Tenants get a note on the record and a decision: for a circle call, a tenant is a dead end, and for a farm you may want the absentee owner’s mailing address anyway. Do-not-call requests go where they go. The list you are left with after one full cycle is the first list that deserves the word verified, and by the second call, which is the one that actually converts, you are dialing a farm rather than a guess.
Run the first pass in the multi-line column of the list allocation, where ring time on unverified numbers is the main cost, and move the pruned list to single-line with the record on screen for every cycle after. If you are combining the farm roll with other sources, the list-building piece covers how to merge them without losing the do-not-call flags in the process.
So, buy or build
Buy the match when the farm is large or the calendar is short, and build the roll yourself in every case, because the assessor data is the part that is actually reliable and it is nearly free. Spend the money on the append and keep the confidence score. Do your own registry scrub inside the 31 days, plan for a cell list, run the reassigned-number check on anything older than a month, and treat the first cycle as the cost of finding out which of your numbers are real.
The farm you chose by the turnover math is only as good as the numbers attached to it, and the circle prospecting hub is built around the assumption that those numbers improve every cycle. They will, if you record what each call taught you about them. Nobody sells that. You earn it, one wrong number at a time.