We process millions of documents through VerifyPDF. The pattern is obvious: bank statements and payslips dominate the workload, accounting for 67.4% of checks in the 28 days ending August 16, 2026.
For lenders, tenant-screening platforms and other teams assessing affordability, these are ordinary business documents. They are also the evidence used to decide whether an applicant’s income and cash flow hold up. Reviewing every file manually creates a bottleneck. My view is simple: automate the consistent checks and reserve human judgement for documents with real warning signals.
Bank statements dominate the workload
The table below covers all documents processed through VerifyPDF during the 28-day period ending August 16, 2026. Each percentage is that document type’s share of the period’s total verification volume. Figures are rounded, so they may not sum to exactly 100%.
| Document type | Share of checks |
|---|---|
| Bank statements | 46.6% |
| Payslips | 20.8% |
| Other documents | 14.2% |
| Government documents | 8.9% |
| Invoices | 2.8% |
| Tax returns | 2.4% |
| Bank letters | 1.2% |
| Employment letters | 1.2% |
| Benefit letters | 0.8% |
| Credit reports | 0.8% |
Bank statements alone represented almost half of the checks. Payslips were a clear second. Together, they accounted for more than two-thirds of verification volume.
That makes sense. The same bank statement can support a loan application, a rental assessment or a due-diligence review. The use case changes, but the pressure does not: someone needs a reliable answer quickly.
If bank statements dominate your intake, start there. Standardize how you collect the original file, run the same checks on every upload and decide which warnings require a person. Our analysis of why fake bank statements are difficult to assess visually explains why visual review is a weak control.
Payslips are the second major workload
Payslips represented 20.8% of checks in the same window. They are compact proof of claimed income, but their formats vary by employer, payroll provider and country. That variation is precisely why checking them by eye is unreliable.
This does not measure payslip fraud incidence. It does show that a verification system focused only on bank statements leaves a fifth of the workload outside the control.
Bank statement and payslip checks should run in parallel while the applicant continues through the workflow. Our guide to income document verification during onboarding explains how to do that without creating another queue.
Do not ignore the long tail
Government documents were 8.9% of volume. Invoices and tax returns added another 5.2%. Bank letters, employment letters, benefit letters and credit reports were smaller again.
No single category rivals bank statements. Combined, however, these documents still reach identity teams, supplier checks, credit decisions and employment reviews. Automating one document type while sending everything else to an inbox does not remove the bottleneck. It moves it.
“Other documents” accounted for 14.2% of checks. That is not noise. Real customers submit regional forms, supporting letters and institution-specific evidence that do not fit a tidy taxonomy.
Give those files an explicit route. Ask for a better original when the file is unusable and send ambiguous cases to a person. Forcing every upload into a confident verdict creates false certainty.
About one in four documents needed more attention
During the same 28-day period, 21.5% of all processed documents were classified as medium risk and 4.9% as high risk. Combined, that is 26.4% of the period’s verification volume.
| VerifyPDF risk band | Share of processed documents |
|---|---|
| Trusted | 16.7% |
| Low risk | 56.2% |
| Medium risk | 21.5% |
| High risk | 4.9% |
| Other or unclassified | 0.6% |
I want to be precise here: this is not a measured fraud rate. A risk band is a triage result, not proof that somebody committed fraud. Medium-risk documents need review. High-risk documents contain stronger warning signals, but the organization receiving the file still makes the decision.
Calling 26.4% of documents fraudulent would be wrong. We do not see the final business decision, a confirmed loss or the outcome of an investigation. What the data does support is a routing decision: trusted and low-risk files follow the normal path, while medium and high-risk files go to a reviewer with the findings attached.
Most checks now arrive through the API
Across VerifyPDF API and dashboard checks in the 90-day period ending August 16, 2026, 79.2% came through the API and 20.8% through the dashboard.
The dashboard is useful for occasional checks and investigations. Once verification becomes routine, clicking through files one at a time is the bottleneck. If a document already enters through an onboarding flow or case-management system, the check should happen there too.
The workflow is simple:
- The applicant uploads the original document through the existing interface.
- The backend sends it to VerifyPDF for analysis.
- VerifyPDF returns the risk band and the warnings behind it.
- A business rule sends the file down the normal path or into review.
If several departments need the same control, our enterprise document verification page shows how one API can support onboarding, finance, claims and HR.
Where this data stops
These are VerifyPDF usage figures, not an estimate for the whole document verification market. The mix reflects our customers, their use cases and the documents they choose to check. A specialist insurer or public-sector team may see something very different.
The selection effect is unavoidable. A document type may rank highly because it is common, because teams consider it risky or both. This data shows where verification demand concentrates. It cannot tell us why every file was submitted or whether the final decision confirmed fraud.
Manual review should be the exception
We have processed enough documents to take a clear position. Checking a small sample misses files that matter. Reading every PDF manually does not scale.
Verify every submitted file consistently. Let trusted and low-risk documents continue through the normal process. Put human judgement where it earns its keep: medium and high-risk cases with the warnings already attached.
If you are planning this across several teams, start with the enterprise deployment model. If you are ready to implement it, go directly to the VerifyPDF API process and see how each result maps to accept, review or reject.