Employment and Landlord Verification: Building a Workflow That Catches a Fake Reference
Fraud rings no longer just forge documents, they staff the phone number you call to verify them. Here is how to build a verification workflow that confirms income and rental history at the source.
Verification Is the Control That Actually Matters
Document review catches careless fraud. Verification catches organized fraud. The difference matters because the population of rental applicants submitting fabricated materials is no longer dominated by someone who edited a PDF at the kitchen table. In a National Multifamily Housing Council pulse survey of 75 apartment owners, developers, and managers conducted between November 2023 and January 2024, 84.3 percent of respondents reported seeing applicants falsify or fabricate pay stubs, employment references, or other income documentation, and 70.7 percent reported an increase in fraudulent applications and payments over the prior twelve months.
A well-built fake pay stub can be internally consistent, arithmetically correct, and rendered by the same payroll template a real employer uses. Auditing the artifact only gets a landlord so far. The purpose of verification is to move the question off the document entirely and onto an independent source, so that the test becomes whether an employer and a prior landlord who exist independently of the applicant confirm what the file claims.
This is also the part of screening most often done badly, because it feels like a formality. A verification workflow that consists of calling the number printed on the application is not a control. It is a step that produces a record of having made a call, which is a different and much less useful thing.
The Manufactured Employer
The most important shift in rental application fraud over the past few years is that sophisticated operations no longer forge only the paperwork. They construct the entity behind it. A shell company gets a domain name, a single-page website with a plausible services description, a business profile with a street address, a social profile with a handful of connections, and a phone number that routes to a person prepared to confirm employment on request.
Every element of that setup is cheap and fast to assemble, and each one is designed to satisfy exactly the checks a leasing agent performs under time pressure. The verification call, which for decades functioned as the backstop for a questionable pay stub, becomes worthless when the applicant controls the line being called. Worse, it produces false confidence, because the file now contains a note recording that employment was verified by phone.
The defense is not to stop calling. It is to stop treating the applicant as a source of the contact information. Every channel used to verify should be discovered independently of the application packet, and the confirmation should ideally come from a system of record rather than from a person answering a phone.
Signals That an Employer Was Manufactured for This Application
None of these are conclusive on their own, and small legitimate businesses will trip several of them, which is why they inform how hard to look rather than deciding the application. A domain registered within the last few months for a company the applicant says they have worked at for years. A business address that resolves to a mail drop, a residential unit, or a coworking suite with no signage. A website with no named staff, no history, and stock photography throughout. A main phone line that is a mobile number, or that is answered without a company greeting. No listing with the relevant state business registry. And the pattern most worth noting, an employer that appears on more than one application in your portfolio within a short window when nothing else connects the applicants.
Building the Employment Verification Workflow
Start by ranking the evidence. Data pulled directly from a payroll provider or from a bank account the applicant connects is the strongest, because it originates from a system the applicant does not control. A written verification returned from a corporate human resources or payroll department through a channel you located yourself is next. A phone confirmation on an independently discovered main line follows that. A phone confirmation on the number printed on the application is the weakest form of evidence and should never stand alone.
The independent channel is the core discipline. Look up the employer through a state business registry, an industry licensing database, or a search that does not begin at the application, then call the main line and ask to be transferred to human resources or payroll rather than to a named individual the applicant supplied. Ask for confirmation of hire date, current status, position, and whether compensation is consistent with what was represented, and note that many larger employers will confirm only dates and title as a matter of policy, which is a legitimate answer rather than a red flag.
Direct-source income verification through payroll connections or bank linking has become common enough to be worth offering, and it resolves several problems at once. It confirms income without depending on the applicant's documents, it handles gig and self-employed applicants whose income never appears on a pay stub, and it usually returns in minutes rather than days. Offer it as one accepted method alongside conventional documentation rather than as a mandate, keep the accepted methods identical for every applicant, and be prepared for applicants who decline to connect an account for reasons that have nothing to do with fraud.
Whatever the method, the standard for what counts as verified should be written down before the applications arrive. Decide in advance what happens when an employer will not confirm, what tolerance applies when verified income differs from the stated figure, and what documentation an applicant with no traditional employer may substitute. Deciding those questions case by case, with a unit sitting vacant, is how inconsistent treatment enters a screening process.
Verifying the Rental History, Not Just the Reference
Landlord references have the same structural weakness as employer references, with an added complication. The most common fake landlord reference is not an elaborate forgery, it is a friend or relative with a prepared story and a phone. And the second most common problem is not fraud at all, it is a current landlord with an incentive to describe a difficult tenant in generous terms in order to see them leave.
The first move is to verify the property before verifying the person. County assessor and recorder records are public in most jurisdictions and usually searchable by address, and they establish who actually owns the unit the applicant claims to have rented. When the name on the reference does not match, that is worth a question. It is not automatically a lie, because rentals are frequently held in an LLC or a trust and managed by a company whose name appears nowhere in the deed, and an applicant may know only the manager. The point is to reconcile the discrepancy rather than to score it.
Then structure the call. Open-ended questions expose coached references faster than yes or no questions, because a coached reference has rehearsed the answers to the obvious prompts. Asking what the monthly rent was, when the tenancy started and ended, whether the deposit was returned in full, and how maintenance requests were handled produces specifics that a real landlord recalls without effort and a rehearsed one often gets wrong. Comparing those answers to the application catches inconsistencies that no document review would surface.
Where the rental history allows, reach past the current landlord to the one before. A prior landlord has no unit to fill and no departing tenant to encourage, which makes them the more candid source. Note also that verification confirms the tenancy an applicant disclosed, and it cannot reveal an address the applicant chose to omit, which is why address history from a screening report belongs alongside reference calls rather than being replaced by them.
Consistency Is What Makes the Workflow Defensible
A verification process that intensifies when something about an applicant feels off is a fair housing problem waiting to be documented by someone else. Feelings about applicants are not evenly distributed, and a process that applies extra scrutiny selectively will produce a pattern that is difficult to explain later, whatever the intent behind any individual decision.
The workable approach is a baseline every applicant receives and defined triggers that escalate. Every applicant gets employer confirmation through an independently located channel and confirmation of the stated rental history. Escalation to deeper review is tied to written triggers that describe facts rather than impressions, such as a mismatch between the deposit amounts on a bank statement and the pay stub, an employer that cannot be located in a business registry, or a reference whose name does not reconcile with ownership records. When the trigger is a documented fact, the escalation is explainable.
Record the results with the same discipline. For each verification, log the date, the channel used, how that channel was located, who was reached, and what was confirmed. That log is what allows a decision to be reconstructed months later, and it is also what distinguishes a denial based on unverifiable income from a denial that merely followed an uneasy feeling.
What to Do With What the Workflow Finds
Give the applicant a chance to respond before deciding. Verification failures have innocent explanations more often than the fraud statistics suggest. Employers change payroll vendors, human resources departments decline to confirm anything beyond dates as a matter of policy, small businesses go quiet during busy weeks, and a prior landlord may have sold the building and moved on. A short, specific request for a corrected contact or an alternative form of proof resolves a meaningful share of failed verifications and costs almost nothing.
When verification genuinely fails, deny on the criterion rather than on an accusation. The defensible basis is that income or rental history could not be verified to the written standard applied to every applicant. Asserting that an applicant committed fraud invites a dispute over proof that a landlord has no obligation to take on, and there is little practical benefit to making the stronger claim.
One legal distinction is worth keeping straight. Information a landlord gathers through their own verification calls is generally not a consumer report, so a denial resting only on that firsthand work does not by itself trigger the FCRA adverse action requirements. If a third party is engaged to perform the verification, or if any part of the decision rests on a screening report, the adverse action obligations apply, and the safer practice when a decision draws on both is to send the notice. State and local law adds requirements in some jurisdictions, so confirm what applies where the property is located.
Finally, treat automation as leverage rather than as a substitute for judgment. Automated document analysis, such as the fraud checks in the TenantFort screening pipeline, is very good at flagging manipulated files and internal inconsistencies at a speed no manual review matches. It cannot tell you whether the company on the pay stub exists. That question is answered by a workflow that goes to independent sources, and the two layers together are considerably stronger than either alone.
Frequently Asked Questions
Do I have to send an adverse action notice if I deny an applicant because employment verification failed?
If the denial rests entirely on verification you performed yourself, that information is generally not a consumer report and the FCRA adverse action requirement is not triggered by it. If a third party performed the verification, or if a screening report contributed to the decision in any part, the adverse action notice is required. When a decision draws on both sources, the safer practice is to send the notice, and some state and local laws impose their own notice requirements regardless.
What should I do when an employer refuses to verify anything?
Treat it as a normal outcome rather than a red flag on its own. Many employers confirm only dates of employment and job title as a matter of policy, and some route all requests to a third-party verification service. Ask the applicant for an alternative form of proof that meets your written standard, such as recent bank statements showing deposits, a tax return, or a direct payroll or bank connection, and apply the same substitution rule to every applicant in that position.
Is it a mistake to call the current landlord instead of a prior one?
It is not a mistake, but it should not be the only call. A current landlord who wants a difficult tenant to move has an incentive to give a favorable reference, so their answer carries less weight than it appears to. Where the rental history allows it, contact the landlord before the current one as well, since that person has nothing at stake in the outcome.
Can I require applicants to connect a bank account for income verification?
Landlords generally have latitude to set the forms of income proof they accept, provided the requirement is applied consistently to every applicant. Requiring bank linking as the only acceptable method tends to be a poor choice in practice, because applicants decline for privacy reasons that have nothing to do with fraud, and it can fall harder on people without conventional banking relationships. Offering it as one accepted option alongside documentary proof gets most of the fraud benefit without narrowing the applicant pool.
Does automated fraud detection make manual verification unnecessary?
No, the two catch different things. Automated analysis examines the document itself, including metadata, template reuse, arithmetic, and signs of editing, and it does this faster and more consistently than manual review. Verification tests whether the employer and the prior landlord exist independently of the applicant, which no amount of document analysis can establish. Fraud that clears both layers is meaningfully harder to execute than fraud that only has to clear one.