For RPM Living regional managers · a two-document test

One of these pay stubs is fake. Find it.

93% of property managers hit application fraud last year. Your site teams are still squinting at PDFs.
Click the document you’d reject.
Both passed a leasing agent’s eyeball test.

DOCUMENT A
EARNINGS STATEMENT
Northwind Logistics LLC · 4410 Harbor Blvd · Period 06/01–06/15/2026
EmployeeA. RIVERA · ID 20831
Gross pay (semi-monthly)$2,445.00
Federal tax−$262.10
State tax−$97.80
FICA + Medicare−$187.04
NET PAY$1,898.06
YTD gross$26,895.00
Generated by employer payroll system · doc ref NW-88213
DOCUMENT B
EARNINGS STATEMENT
Northwind Logistics LLC · 4410 Harbor Blvd · Period 06/01–06/15/2026
EmployeeD. OKAFOR · ID 20790
Gross pay (semi-monthly)$4,890.00Font weight on this figure differs from the rest of the table — the amount was edited after the document was generated.
Federal tax−$264.30
State tax−$98.10
FICA + Medicare−$188.15
NET PAY$4,339.45Withholding matches a ~$2,400 gross, not $4,890 — the taxes were never recalculated when the pay was inflated.
YTD gross$29,340.00YTD ÷ 11 pay periods ≈ $2,667 — the history doesn’t support the claimed rate.
Generated by employer payroll system · doc ref NW-88214File metadata shows a browser-based PDF editor commonly used for document fraud — not a payroll system.

  • LAYER 1Forensic analysis: the gross-pay figure was modified after generation — pixel-level edit artifacts.
  • LAYER 2Font analysis: the edited figure doesn’t match the payroll template’s type.
  • MATHWithholding and YTD totals contradict the claimed rate — fraudsters raise one number and forget the four that depend on it.
  • LAYER 5Bank cross-reference: actual payroll deposits are $1,890 twice a month, not $4,339.
See all five detection layers ↓
93%
of property managers hit application fraud last year — up 40% year over year
$16B
estimated annual rental fraud losses industry-wide
1 in 28
applications carries fraud signals, across 500,000+ screened

Sources: NMHC pulse survey · Industry loss estimates · Two Dots production data, 2026

The detection stack

Five layers, because good fraud beats any single check

Each layer catches a class of fraud the others can’t see. All five run on every document at every property — including the one staffed by the agent who started last week.

LAYER 01

Forensic document analysis

Every pixel of every PDF, checked for post-generation edits: amounts, names, dates, and the fingerprints of browser PDF editors. This is where the “perfect” paystub fails first.

catches: Photoshopped statements · generator-template stubs · edited amounts

LAYER 02

Font pattern analysis

Real ADP documents have exact typographic signatures. Lookalikes from stub generators don’t — invisible at reading distance, unmistakable to the model.

catches: ADP-template fakes that pass visual review

LAYER 03

Synthetic identity scoring

Sentilink risk scoring at the SSN level. A CPN attached to a manufactured identity can carry a spotless 720 score — a clean credit report is exactly what this fraud is built to produce.

catches: CPN fraud · synthetic identities — invisible to credit checks

LAYER 04

Institution-specific rules

100+ false-positive filters tuned across 50+ financial institutions — so one bank’s formatting quirks don’t flag honest applicants, and forgers mimicking that bank get measured against its real conventions.

catches: bank-specific forgery tells · keeps honest applicants out of the fraud pile

LAYER 05

Live bank cross-reference

The document says $4,890 twice a month; the linked account says otherwise. Direct-source verification is the layer no forged PDF survives — and matching documents get marked VERIFIED. Proof in both directions.

catches: everything the paper trail claims that the money trail contradicts

The invoice for a miss

What one miss costs and who takes the owner’s call

A fraudulent tenancy surfaces as a missed payment around day 45, becomes an eviction filing by day 90, and lands as a five-figure write-off. The owner doesn’t call the leasing agent who approved the file — they call you.

Screening quality is portfolio reputation, and it’s being tested 4–5 times per hundred applications.

Cost of one fraudulent tenancy
Unpaid rent (4 months)$6,400
Legal + eviction proceedings$3,500–$10,000
Unit turn + damages$3,000–$5,000
Staff time, lost velocity$2,000+
Typical all-in (NAA est.)$20,000–$25,000
The counter-record

What the file looks like when Evie keeps it

Evie runs all five layers on every application and attaches the evidence to every decision — roughly 1 in 28 applications gets caught, each one an eviction that never happens. The same audit trail defends your approvals when an owner, auditor, or fair housing complaint asks how a decision was made.

Operators on Evie: 60–70% bad debt reduction · ~$2M saved annually per 10,000 units
Trusted by operators nationwide
BH Management RAM Partners Fogelman Pretium Corcoran Timberland Partners Van Metre FirstKey Homes Cerberus Moss & Company MG Properties
The offer

See what Evie finds in your applications

Bring last quarter’s approvals. In 15 minutes we’ll show you the fraud signals a five-layer screen would have flagged — and what each one would have cost.

Request a fraud audit