AI Risk Scoring Explained: What Every Landlord Needs to Know Before Approving a Tenant

You have a stack of applications. You've pulled credit, criminal, and eviction reports on your top three candidates. Each report is 8 to 15 pages long. Each one tells you something different. One candidate has a great credit score but a short rental history. Another has a lower credit score but 7 years of documented on-time rental payments. The third has solid income and credit but an eviction from 5 years ago that they say was a dispute with a prior landlord, not a non-payment situation.

How do you decide? This is the exact problem AI risk scoring solves — and it's why landlords who use it make measurably better tenant selection decisions than those who rely on gut instinct and raw report data.

What a Risk Score Is (and Isn't)

An AI risk score is a synthesized assessment of an applicant's likelihood to be a reliable tenant, expressed as a number (typically 0–100) and accompanied by a plain-English explanation. It's not a replacement for human judgment. It's not a definitive verdict. It's a translation layer that takes complex, multi-source data and makes it accessible to a landlord who isn't a data scientist or a professional underwriter.

What the risk score is NOT: a credit score. A credit score (FICO) measures creditworthiness in a general financial context — primarily designed for lenders, not landlords. It doesn't weight rental history, which is arguably the most predictive factor for tenant reliability. A high FICO score doesn't guarantee a good tenant, and a lower FICO score doesn't mean a bad one. The AI risk score incorporates credit as one signal among many, with appropriate weighting for the rental context specifically.

The Data Inputs Behind the Score

A comprehensive AI risk score for tenant screening draws on multiple data sources simultaneously. Credit report data includes FICO score, payment history (on-time vs. late), number and type of derogatory accounts, collections, charge-offs, bankruptcies, and debt-to-income ratios. Rental-specific data includes eviction filings and judgments (searched by jurisdiction for every address the applicant has listed), SSN trace confirming identity and surfacing any unreported prior addresses, and rental history data where available.

Criminal background data includes county-level records, state records, federal records, sex offender registry, and watch lists. Income verification confirms that the stated income is accurate and that the rent-to-income ratio is within an acceptable range. All of these data points are weighted based on their predictive value for tenant reliability specifically — not financial reliability generally.

How the Score Is Calculated

The AI model underlying a risk score has been trained on large datasets of tenancy outcomes — meaning it's learned which combinations of factors correlate with on-time payment, lease completion, and positive tenancy outcomes, versus which correlate with late payments, evictions, and property damage. This training allows it to weight factors appropriately for the rental context.

For example: a history of on-time rent payments is more predictive of future on-time rent payments than a general credit score. An eviction from 2 years ago for non-payment is more concerning than an eviction from 8 years ago for a lease violation. A bankruptcy that was discharged 4 years ago and has since resulted in rebuilt credit is different from a bankruptcy filed last year. The AI can distinguish between these nuances. A human reading a raw credit report often cannot — or doesn't have time to.

Interpreting the Score: What the Numbers Mean

• 85–100: Strong application. Low risk of payment issues or lease violations. Approve with confidence.
• 70–84: Good application with minor risk factors. Review the explanation and assess whether any flags are dealbreakers for your specific situation.
• 55–69: Moderate risk. The explanation matters here — some 65-score applications are much better than others. Read the detail carefully and consider whether additional conditions (larger deposit, co-signer) mitigate the risk.
• 40–54: Elevated risk. Proceed with caution. If you approve, understand what you're accepting.
• Below 40: High risk. Unless there are extraordinary mitigating circumstances clearly explained in the report, this application should be declined.

These ranges are guidelines, not rules. Your specific situation matters. A 62-score applicant for a unit that's been vacant for 45 days and is priced at top market may be worth approving. The same score for a unit that's been vacant one week with multiple strong applicants is not.

Consistency and Fair Housing

One of the most important benefits of AI risk scoring is consistency. When every applicant is evaluated by the same model, against the same criteria, with the same weighting, you eliminate the human bias that can creep into manual screening decisions — conscious or unconscious. This is both an ethical imperative and a legal one.

Fair Housing law requires that landlords apply the same criteria to all applicants. If you approve a 62-score applicant from one demographic and decline a 65-score applicant from another, you have a problem. AI scoring, when combined with documented, consistent screening criteria, provides a defensible record that your decisions were made on objective criteria applied uniformly.