Skip to content

Screening September 28, 2026

Great credit score, terrible tenant: the signals that actually matter

A credit score predicts whether someone repays credit. It was never built to predict whether they pay rent on the first, keep the unit, or stay past month eight. Four signals that do more work, and how to write each one into criteria you can defend.

A credit score is a prediction of credit behavior. The CFPB’s own definition says so: a prediction of how likely a person is to pay a loan back on time, built from what is in their credit reports. Rent is usually not in a credit report. Landlords are not required to furnish payment data to the bureaus and mostly do not, which means an applicant can have a long record of late rent and still carry a clean score.

So stop treating the score as the answer and start treating it as one of five inputs. Concretely: add two lines to your criteria this week, a minimum verified tenancy length and a minimum number of landlord references you will actually reach, and weight them the same as the score.

Here is what each signal is really measuring.

What the score does measure, and it is not nothing

Bill-paying history, unpaid debt, how many accounts and of what type, how long they have been open, credit utilization, and recent applications. Collections, foreclosure, and bankruptcy show up too, with timing.

That is a genuinely useful picture of financial pressure. A score in the low 500s driven by active collections tells you the applicant is already being pursued for money, and rent competes with that. What the score cannot tell you is whether this person is the kind of tenant who calls about a leak on day one or the kind who lets it run for a month.

One more thing worth knowing before you set a cutoff: nobody has one score. Different models and different data produce different numbers for the same person on the same day, and most of them land somewhere in a 300 to 850 range. A cutoff of 680 is not a precise instrument. It is a rough line, and it should be written down as one.

Signal 1: rent payment history, verified with the landlord

This is the closest thing to a direct measurement, and it is the one most landlords skip because it takes phone calls.

Ask the current landlord how many times rent arrived after the grace period in the last twelve months. Ask the prior landlord the same thing. Two independent answers that agree are worth more than any score, and two that disagree are worth investigating.

The catch is that the current landlord has an incentive to give a glowing reference to a tenant they want gone. The prior landlord has none. That is why the list of questions matters more than the fact of calling, and why the prior landlord goes on the list at all: questions to ask a tenant’s previous landlord covers the script and the verification step that keeps a friend from answering the phone.

Signal 2: income stability, not income size

Two applicants both earn four times the rent. One has been at the same employer for six years. The other started ten weeks ago after a seven-month gap. Same income multiple, very different risk, and only one of those differences shows up in the number your criteria are measuring.

Write the criterion to capture both: a minimum multiple of rent, and a minimum length of current income with a defined path for applicants who have recently changed jobs, are self-employed, or are retired. The path matters. A rule that only recognizes W-2 employment quietly excludes self-employed applicants, retirees, and people on disability benefits, which is a fair housing problem hiding inside an income rule.

California adds a hard requirement on top: where a government rent subsidy is involved, the income standard has to be based on the portion of the rent the tenant actually pays, not the full rent (Cal. Gov. Code § 12955(o)). How to verify tenant income and employment goes through the documents in order of how hard each is to fake.

Signal 3: the address timeline, read as a timeline

Take the addresses on the application and the addresses on the consumer report and line them up by date. You are not looking for a number of moves. You are looking for gaps and contradictions.

A gap of eight months with no address is a question, not a decline. Sometimes the answer is a deployment, a hospital stay, or a stint back with family, and it takes one sentence to clear. Sometimes the answer is a tenancy the applicant would rather you not call. An address on the report that is missing from the application is the same kind of question.

Ask it the same way every time, write the answer down, and move on. The value here is not suspicion. It is that a timeline with no unexplained holes is a much stronger file than a high score with three.

Signal 4: how the applicant behaves during the application

This one needs care, because it is the signal most easily contaminated by preference, and preference is where fair housing complaints come from.

Keep it to things you can record as facts, and record them for everyone:

  • Was the application complete on the first submission
  • Were requested documents provided within the window you gave
  • Did the stated employer, income, and address match what came back on verification
  • Were the same answers given twice when asked twice

Notice what is not on that list. Tone, dress, accent, how the person seemed, whether they were easy to talk to. Those are not signals. They are the raw material of a disparate treatment claim, and a decision that rests on them is indefensible in exactly the moment it needs defending.

A discrepancy between what the application says and what verification returns is different. It is a fact, it applies to every applicant equally, and it is the single most predictive thing in the file.

What to do with a great score and a bad signal

Decline on the written criterion, not the vibe. If the file shows verified late payments in excess of what your criteria allow, that is the reason, and it is the reason you record.

Then send the notice. If any part of the decision drew on the consumer report, the Fair Credit Reporting Act requires an adverse action notice naming the agency, saying the agency did not make the decision, and telling the applicant about the free copy and the dispute right (15 U.S.C. § 1681m(a)). Adverse action notices covers the contents.

And if you find yourself wanting to approve someone whose file fails a criterion, that is a signal too, about the criterion. Rewrite it for everyone, date it, and apply it going forward. Do not make an exception for one applicant. That is not flexibility; it is the pattern that a complaint is built out of.

The score belongs in the file. It just does not belong at the top of it. Screening in LeaseRooster keeps the report, the verifications, and the written criterion that decided it on the same application record, which is the only form in which any of this is useful later.

This is general information, not legal advice. Rules change and local ordinances vary. Confirm current requirements with your city, county, or an attorney before you act.

Common questions

What to look for when screening tenants?

Verified rent payment history from the current and prior landlord, income stability rather than income size, an address timeline with no unexplained gaps, and whether the application matched what verification returned. The credit report and score sit alongside those as one input, not as the decision.

Can a tenant with a good credit score still be a bad tenant?

Yes, and it is common, because most landlords do not report rent payments to the credit bureaus. A credit score measures how a person handles credit accounts. It contains little or no information about whether they paid rent on time, kept the unit in condition, or gave proper notice at move-out.

Get more posts about screening tenants

New posts most weeks. No spam, unsubscribe anytime.

The report is one input. Keep the rest.

Verifications, references, and the written criterion that decided the application stay attached to the applicant, so the whole picture is still there when it matters.