How should I prioritize an off-market property list before calling or mailing every owner?

Property fit is scored first because it is the only component that can remove a record outright. Everything else moves a record up or down the queue, so a parcel that fails the buy box never reaches the stage where somebody buys a phone number for it. Keep the four scores in four spreadsheet columns or four CRM fields, which is what makes a tier assignment explainable two weeks later.

What each of the four scales measures

• Property fit measures the parcel against the written criteria that produced the list, and a 0 removes the record from the campaign.
• Timing evidence measures whether a dated document says the owner is moving, so a modeled score with no filing behind it stays at a 1.[2]
• Contactability measures whether a channel can carry a message to the owner of record this week, and a 0 sends the record to a mail or research queue.[3]
• Evidence quality measures provenance, which keeps a high total from being assembled out of inferences.

Contact data is metered, which is the part of a top-to-bottom dial that shows up on an invoice. PropertyRadar includes 250 phone and email contacts a month on the $119 Solo plan, 500 on Team, and 2,500 on Business, then charges 8, 6, or 4 cents per contact past the ceiling.[1] A 4,000-record list appended in export order burns the Solo allowance sixteen times over before anyone asks which records deserved the call.

A record with no working number and a modeled score of 91 produces no conversation this month at any price. Contactability and evidence quality therefore set the order more than any motivation label does, because reach and provenance can be checked today while a propensity band is a statement about the next six months.[2] A record scoring 12 out of 12 and a record scoring 6 both stay in the same campaign, since the lower one moves to Tier 3 rather than out of it.

Property fit is scored against the criteria already written into the search that produced the list. Running them a second time catches parcels that slipped a loose filter, such as a duplex counted inside a single-family pull. A record scoring 0 leaves the campaign, because a phone number for a property you would never buy costs the same 8 cents as one for a property that fits.[1]

A deed recorded last week may not be searchable yet, which is how an ownership criterion goes stale. Harris County's real property portal tells filers a document can take 1 to 2 business days after processing to appear in a search inquiry, and the same page states the online database is not the official repository of real property records.[4]

A record matching four criteria and a record matching one arrive in the same export with no marking between them. Filter strictness belongs in the fit column for that reason, since absentee ownership, high estimated equity, a vacancy flag, and tax delinquency stack differently depending on which criterion the search treated as required.

Timing evidence scores highest when a dated document exists and lowest when the only input is a model's output. A recorded notice of default, a probate filing, or a listing that expired last month all carry dates somebody can check against a public index. Mecklenburg County's delinquent taxpayer list shows the shape of a 3, since each monthly release states the tax record date it reflects and names the owner shown as of the January 6, 2026 delinquent date.[5]

PropertyRadar describes four forward-looking Likely Transactions signals, Likely to Sell among them, each a 0 to 100 score with bands from Very Low to Very High.[2] They estimate who is most likely to transact in the next six months, with every property rescored monthly, and they enter this column as one input and never as the column itself.[2] CEO Mark Hockridge is quoted on that page calling the product something that is not a crystal ball, so the score arrives carrying no date of its own.[2]

Sorting a whole list on that single axis lifts unreachable records to the top beside reachable ones. The same announcement describes customers filtering by score and sorting high to low to work the best opportunities first.[2] A modeled score sits at a 1 here and earns a 2 only when a dated event appears beside it.

Contactability scores whether a specific channel can carry a message to the owner of record this week. An active mobile tied to the person named on the deed scores 3, a landline or a shared household number scores 2, a deliverable mailing address with no phone scores 1, and a record carrying neither scores 0. PropertyRadar marks the distinction inside its owner profile, where a green checkmark means the number is active and safe to call or text while a yellow triangle covers statuses including opted out, wrong person, disconnected, do not call, and blacklisted.[3]

The do-not-call status appears only once the user has accessed DNC data, and the blacklisted status only once a Blacklist Alliance account is connected.[3] An account with neither configured shows green checkmarks screened against nothing at all. Sellers using the National Do Not Call Registry must synchronize their calling lists with an updated registry at least every 31 days, so a list you scored in March cannot be dialed in June without another suppression pass.[6]

CDC's National Health Interview Survey estimates that 78.7% of adults lived in wireless-only households during the second half of 2024.[7] A landline-only record scores lower on reach for that reason, and roughly 35 million numbers are disconnected and made available for reassignment each year according to the FCC order behind the Reassigned Numbers Database.[8] That churn alone can move a record from a 3 to a 0 between scoring passes.

San Bernardino County's Assessor-Recorder-Clerk states that mailing addresses come first from the Preliminary Change of Ownership Report filed with the recorded document.[9] The address stays in use until the owner signs a request to change it, which is why a record with a deliverable address and no phone stays in the campaign at a 1.[9]

A recorded document carrying an instrument number and a filing date sits at the top of this column. Government data with a published lag sits a step below, and modeled fields such as estimated equity sit below that. A third-party field with no attribution scores 0, since a number nobody can trace to a source cannot be re-checked when the owner disputes it on the phone.

Texas requires a certificate of formation to name each initial manager, or each initial member where the company has no managers.[10] Delaware's LLC Act requires only the entity name and the registered agent's name and address.[11] A person named on a Texas filing scores a 3 here, and a name attached to a Delaware entity by a data provider scores a 1 until a document confirms it.

Estimated equity is modeled from assessor values and recorded loan balances, so it scores a 1 even when the number looks precise. Grading provenance this way keeps a 12 out of 12 record from being built out of fields nobody can re-derive.

Tier 1 gets worked to completion before Tier 2 opens, the discipline most easily lost when a fresh list lands mid-campaign. New records enter the same scoring pass as everything else and take their place by score, so an arrival holding a 6 waits behind an older record holding a 10.

Buying phone numbers for all 4,000 records on the $119 Solo plan costs $300 in overage past the 250 included contacts.[1] Buying Tier 1 first spends the included allowance on records that already carry a dated event and a deliverable address, which turns a monthly allowance into a queue instead of a budget somebody drains in week one.

PropertyRadar rescores every property on its Likely Transactions signals monthly, which is one reason a monthly interval is defensible.[2] A monitored list adds and removes properties as their criteria are re-evaluated and keeps a New Since count that can fire alerts or automations, with a ceiling of 10,000 monitored properties per list that add-ons cannot expand.[13]

A phone status flipping from active to disconnected drops contactability by 3 points while the other columns hold.[8] An ownership change resets all four columns at once, which is why the monthly pass reads county filings before anything a vendor appended.[4]

• Four columns cost more time than one number, and a first pass over a 4,000-record list takes hours that pay back only where the append budget is capped.[1]
• A rubric built on county data inherits county lag, since Harris County tells filers a document can take 1 to 2 business days after processing to appear in a search inquiry.[4]
• The four scores decay at different speeds, since a recorded filing keeps its date while a phone status can flip between scoring passes.[8]
• A green phone flag means less than it looks like when screening is unconfigured, since the do-not-call status appears only after DNC data is accessed.[3]
• No conversion rate belongs in this rubric, because the tiers rank records by what is documented and reachable and nothing here predicts what share of any tier answers or signs.

A single motivation score ranks records against each other and does not say which record to call first. The score blends inputs that carry different confidence and decay at different speeds, so nobody can audit the number when it drops from 74 to 61.

Google's machine learning course defines prediction bias as the difference between the mean of a model's predictions and the mean of the ground-truth labels.[12] A model trained on data where 5% of emails are spam should predict spam about 5% of the time on average, which is a statement about a population.[12] Averages of that kind describe the list, which makes any single high score a weak statement about one owner's plans.

A BiggerPockets thread carries the title of dialing straight down a lead list, and the replies describe buckets by preforeclosure recency and by vacancy duration.[14] Forum practice is evidence that the ordering question is common, and it is not evidence about what converts.