High-equity homeowner lists are sold as though equity were a field on a record. It is not. It is the output of a subtraction where neither term is known:
equity = (an estimate of what the house is worth) − (an estimate of what is still owed)
That does not make the number useless. It makes it a number with a confidence interval that nobody prints, and understanding where the uncertainty comes from is what separates a list that performs from one that produces polite confusion on the phone.
The value side: what a valuation model knows
An automated valuation model predicts a sale price from the attributes of a property and the recent sales around it. It is genuinely good at this in the ordinary case, and its failure modes are specific and predictable.
It knows: square footage, bed and bath counts, lot size, year built, and the prices of nearby comparable sales. Where the housing stock is uniform and turnover is frequent, that is most of what determines price, and the model does well.
It does not know:
- Interior condition. A gut renovation and deferred maintenance look identical from the assessment record.
- Anything unpermitted. Finished basements and additions that never saw a permit are invisible.
- Site-specific quality — the view, the busy road, the power line, the school boundary that runs down the middle of the street.
- Recent local events too new to appear in comparable sales.
The practical rule: valuation confidence tracks how ordinary the property is and how many comparable sales are nearby. A tract home in a large subdivision is estimated well. A five-acre property with an outbuilding, in a county with forty sales a year, is not — and no error bar accompanies the number when it reaches your list.
The debt side: what recording does and does not tell you
This half is worse, and it gets much less attention. What is public is the instrument — a mortgage was recorded, for a stated original amount, on a stated date. What is not public is the current balance.
Which produces four distinct sources of error:
| Problem | Effect on the estimate |
|---|---|
| Balance is inferred from an assumed amortisation schedule | Wrong whenever the loan was refinanced, recast, paid ahead, or is interest-only |
| Releases are not always recorded promptly | A paid-off mortgage can sit on the record for years, understating equity |
| Lines of credit have no fixed balance at all | A recorded HELOC limit is a ceiling, not a debt. Counting it as drawn understates equity badly |
| Second positions and private liens vary in coverage | Missing them overstates equity — the opposite direction from the other three |
Three of those four push the estimate the same way. In aggregate, debt modelled from recorded instruments tends to be stale in the direction of showing more debt than exists — which means a strict high-equity filter is more likely to exclude qualified homeowners than to include unqualified ones. That is the less damaging failure, but it is still a failure, and it is invisible.
Why the errors compound
Subtracting two independent estimates produces a result less certain than either. The uncertainty of the difference is larger than the uncertainty of the value alone.
This is why a threshold — “more than 50% equity” — is a much softer line than it appears, and why the records sitting near it are essentially unsorted. Records well past the threshold are reliable; records within a few points of it are a coin toss dressed as a filter. If you are going to use a cutoff, set it further out than feels necessary and accept a smaller list.
The signals that behave better than the score
Because the composite is fragile, the more robust approach is to filter on the underlying facts. Each of these is recorded rather than modelled:
- No open mortgage on record. The strongest single signal available, and it needs no valuation at all. It is also the cleanest to explain to a rep.
- Length of ownership. A long hold on an amortising loan builds equity mechanically, and tenure is a hard date rather than a model. Nationally, tenure skews long — 25.3% of single-family homes have been held five to ten years and 24.5% ten to twenty, with median tenure past twelve years in many states. The distribution is in how long people stay in their homes.
- Purchase price against current estimated value. Appreciation since purchase is a real equity source and uses a recorded number on one side of the comparison.
- Original loan amount against purchase price. A large down payment at acquisition is durable evidence and does not decay.
- No refinance activity in a period of low rates. An owner who did not extract equity when it was cheapest to do so probably still has it.
A filter combining two or three of these is more defensible than any single equity percentage, and it has the practical advantage that a manager can explain why a record is on the list.
Building the list
- Start from the recorded signals above, not from a vendor’s equity band.
- Decide whether you want occupants. A high-equity absentee owner is a different pitch from a high-equity resident — see how to find absentee owners.
- Exclude entity ownership deliberately rather than accidentally, remembering that family trusts often carry the same flag as institutional owners.
- Set the equity threshold further out than the marketing case requires, so the borderline records fall outside rather than inside.
- Check coverage in the counties you actually work. Lien recording quality varies by county, and an equity filter in a poorly covered county is filtering on absence of data rather than absence of debt.
Point five is the one that quietly ruins campaigns. Missing lien data looks exactly like no lien, so the counties with the worst records produce the most apparent high-equity homeowners.
What to tell the floor
Reps should know that equity on a record is an inference. The failure it produces is a specific and avoidable awkwardness — a rep who opens with certainty about someone’s financial position and is wrong has lost the conversation in one sentence, and they will not know why.
The workable posture is to treat equity as a reason the household is on the list rather than a fact to state to them. It informs which offer to lead with; it is not a thing to say out loud. The same discipline applies to any modelled field on a list, which is a large share of what gets sold — see skip tracing accuracy for the same argument applied to contact data, and how to find out who owns a house for what the underlying ownership record does and does not establish.
Frequently asked questions
How is home equity estimated in property data?
By subtracting recorded debt from an estimated value. Both sides are modelled rather than known: the value comes from an automated valuation model, and the debt comes from recorded mortgage instruments whose current balances are not public. The result is a difference between two estimates, which is a weaker number than either one.
Does “no mortgage on record” mean the home is owned free and clear?
Usually, and not always. It can also mean a lien exists that was recorded in a way the data source missed, that the record is in a county with poor coverage, or that the property was bought with financing that was never recorded as a conventional mortgage. It is the single strongest equity signal available, and it is still a signal rather than a fact.
Why do equity estimates differ so much between providers?
Because the two inputs differ. Valuation models disagree with each other routinely, especially on unusual properties and in thin markets, and providers make different assumptions about how much of a recorded mortgage has been paid down. Two reasonable methods can put the same house on opposite sides of a 50% equity threshold.