Comparison

Manual PDF Review vs Automated Tamper Detection

Built for fraud ops at lending, insurance & compliance teams

A reviewer reads what the page says. An automated check examines how the file was built. They answer different questions, so they miss different things. This page shows where each one stops, and how teams put them in order so neither is asked to do the other’s job.

~3 sec
per document
63
forensic checks per document
1,500+
docs / month on Growth
Plans
subscriptions or one-off packs – see pricing

Two different questions

A reviewer judges the content. A file check judges the file.

Manual review is a judgement about the document as a reader sees it: do the figures add up, does the layout match what this issuer normally sends, does the story fit the applicant. A good reviewer catches things no software can, because they know the context.

What a reviewer cannot do is see inside the file. A PDF that has been opened in an editor, changed and saved again can look identical to the original on screen. The page renders the same, the fonts look right and the totals still add up. Whether the file was changed after it was issued is a property of its construction, not of its appearance, and a checklist has nothing to look at.

The reverse also holds. A file check says nothing about whether a salary is plausible for the job title, or whether an invoice is for goods that were ever delivered. It reports on the file, and a document can be structurally unremarkable and still describe something untrue.

What a checklist cannot see

  1. 1Whether the file was re-saved after it was issued
  2. 2Whether a signed document was changed after signing
  3. 3Whether pages come from more than one source file
  4. 4Whether the file’s origin matches the issuer it claims to be from
  5. 5Anything that renders correctly but was edited

What this looks like

Where manual review and a file check each earn their place

Three real fraud mechanics we catch at the structural PDF layer.

01

A bank statement with one changed balance

The edit is clean: same font, same alignment, totals recalculated. A reviewer comparing the page against a checklist has no reason to object. A file check looks at whether the statement is still the file the bank’s system produced, and flags it as modified if it is not.

02

An invoice with swapped bank details

The supplier’s name, logo and amounts are all correct, and only the account number differs. A reviewer who has never seen this supplier’s previous invoices cannot tell. A file check does not need previous invoices, because it examines the document in front of it. Calling the supplier on a number you already hold is still the right second step.

03

A pay stub that is internally consistent but implausible

The file is fine, but the stated salary does not fit the role, the city or the employer’s pay structure. No file check can catch that. This is the reviewer’s job, and it is why an INTACT result is not an approval.

04

A scan or photo saved as PDF

Scanners and phones produce files with little origin information, so a file check often cannot give a firm answer and returns INCONCLUSIVE rather than guessing. This is where the reviewer’s attention is most useful: the file check has told you that it cannot vouch for this one.

05

A contract signed electronically

A visual check cannot tell whether anything changed after the signature was applied. A file check reports whether the signed document is still the one that was signed.

06

A genuine document that was re-saved

A customer annotates a statement, merges two attachments into one file or prints a page to PDF before sending it. The content is true and the file has still been through another tool. A file check reports that the file changed; the reviewer decides, with the source copy in hand, whether the change matters. This is why MODIFIED opens a conversation rather than closing a case.

07

A document that is both implausible and edited

Agreement between the two layers is the strongest signal either can give. When the figures do not fit the case and the file also reports a modification, the reviewer has two independent reasons to ask the issuer for a fresh copy instead of one.

63
Forensic checks per document
~3 sec
Median analysis time, end to end

How they fit together

Put the file check first, the reviewer second

Each layer covers what the other cannot reach.

Manual review alone

Judges what the page shows

  • Catches implausible figures and inconsistent stories
  • Brings case history and issuer knowledge
  • Cannot see whether the file was changed after issuance
  • A well-made edit looks the same as the original

Strong on context, blind to construction.

Manual review plus a file check

The check examines the file, the reviewer examines the case

  • Every file gets the same structural examination
  • MODIFIED and INCONCLUSIVE arrive with named findings
  • The reviewer’s attention goes where it is needed
  • Content, plausibility and sign-off stay with people

Neither layer is asked to do the other’s job.

Results in under 3 seconds30 to 1,500+ documents/month

What HTPBE? checks

Detection capabilities

Deterministic structural signals. No probabilistic scores, no model training.

Step 1: run the file check on arrival

Submit every incoming PDF before anyone opens it. The result is one of three verdicts – INTACT, MODIFIED or INCONCLUSIVE – with named findings attached, so the reviewer starts with context instead of a blank file.

Step 2: route by verdict, not by volume

MODIFIED goes to a person together with a request for a fresh copy from the source. INCONCLUSIVE is handled according to what the document claims to be: a bank portal download that comes back INCONCLUSIVE deserves a question, a scan the sender openly scanned does not. INTACT continues to the normal content review.

Step 3: keep the content review exactly as it is

The reviewer still checks arithmetic, plausibility, consistency with the case file and the facts that only context supplies. The file check adds a question the checklist never asked. It does not remove any question the checklist already asks.

Step 4: confirm with the source when it matters

For documents that carry real money or a legal decision, neither layer replaces asking the issuer. A direct request to the bank, employer or supplier settles what no review of the file can.

What to write on the case file

Record the verdict, the named findings, who reviewed the case and what the issuer said when asked. A decision that rests on two documented layers, a file result and a human judgement, can be explained later to a customer, an auditor or a regulator without anyone reconstructing it from memory.

Learn from the cases where the layers disagree

Cases where the reviewer was uneasy and the file came back INTACT, or where the file came back MODIFIED and the reviewer saw nothing, show what each layer adds in your own queue. Review them periodically and adjust the checklist, not the file check: the checklist is yours to change.

A finding is a reason to look harder

A MODIFIED verdict reports that the file changed after it was issued. It does not say why, and it does not say anyone acted in bad faith: people re-save, merge and annotate legitimate documents. The reviewer decides what the finding means for the case.

Accountability stays with the person

Approval and rejection decisions are made by people who can explain them. The file check supplies evidence for that decision and is not wired to make it automatically.

Share with engineering

Wire this into your intake pipeline in under a day

Two API calls – one POST to submit the PDF, one GET to retrieve the verdict. Forward this page to your engineering team; the full API reference, quotas, and copy-paste examples in cURL, JavaScript, Python, PHP, Go, and Ruby are one click away.

Pricing

Self-serve plans, no sales call

All plans include the same forensic checks. Pick the quota that matches your monthly document volume.

manual

Starter

30 credits/mo

Manual spot-checks and integration testing

most common

Growth

350 credits/mo

Active document processing pipelines

high volume

Pro

1,500 credits/mo

High-volume automation and API integrations

Enterprise (unlimited, on-premise available) — see full pricing

API key on signup. Free test environment on every plan. No card required.

Customer Stories

Teams that stopped document fraud

Compliance, finance, and risk teams use HTPBE? to catch manipulated PDFs before they become costly mistakes.

Caught an invoice where the total had been changed by less than a thousand dollars. Without this I would have approved it without a second look.

Sarah M.

AP Manager

United States

We had three applicants in the same week with bank statements that looked completely fine. Two of them were flagged as modified. You simply cannot see this by reading the document – it is in the file structure.

Lars V.

Risk Analyst, Online Lending

Netherlands

Salary slips were coming with altered figures. We identified two problematic files before the placement was finalised.

Priya K.

HR Operations Lead

India

Since we started checking documents this way, we stopped two applications early in the process that would have been very difficult to reverse later.

Julien R.

Fraud Analyst, Fintech

France

Some applicants were sending PDFs that looked authentic but had been edited in ways not visible to the eye. We now ask for checked originals when something is flagged. Already saved us from a few bad decisions.

Marta S.

Compliance Coordinator

Spain

One invoice was caught because there was a mismatch between the document dates and structure. That particular case would have cost us significantly.

Tariq A.

Finance Manager

United Arab Emirates

FAQ

Frequently asked questions

Can a person tell by eye that a PDF has been edited?

Sometimes, when the edit is careless: mismatched fonts, uneven spacing, a blurred patch, numbers that do not add up. A careful edit leaves none of those, and the page looks like the original. Whether the file itself was changed is a different question from whether the page looks right, and only examining the file can answer it.

What should a manual PDF review checklist include?

Arithmetic and totals, dates that fit the period, names and addresses that match the case file, a layout consistent with what this issuer normally sends, and a plausibility check on amounts. Add a request to the source for anything high-value. A checklist covers the content of the document well. It cannot cover how the file was produced, which is the gap an automated file check fills.

Can the PDF authenticity checker detect all types of modifications?

No. HTPBE? looks for evidence that an existing PDF was changed after it was created, so content that was false from the start can show none, and password-protected files cannot be analysed. Treat it as one layer in a fraud check.

How does HTPBE? determine whether a PDF was modified?

HTPBE? looks for evidence inside the file that it was changed after it was created, reports each finding by name, and returns Intact, Modified or Inconclusive. It reads the file, not the text of the document.

Does an INTACT result mean the document is genuine?

No. INTACT means the file shows no evidence of modification after it was issued. It says nothing about whether the content is true: a document can be created clean with invented figures. Content review and source confirmation stay in the process for that reason.

What does a reviewer do with MODIFIED or INCONCLUSIVE?

MODIFIED is a reason to ask for a fresh copy straight from the source and to look harder at the case. INCONCLUSIVE means the file’s origin gives nothing to check integrity against, which is common for scans, photos and consumer software. It is a routing signal rather than a pass or a rejection: weigh it against what the document claims to be.

Do we have to change our review process to add a file check?

No. The file check sits in front of the existing queue and runs on each incoming PDF, through the web checker for occasional use or through the API for a pipeline. Your reviewers keep their checklist and their sign-off, and receive the verdict and findings with each file. Plans and per-check pricing are on the pricing page.

Does the file check read the content of the document?

No. The analysis is structural. It examines how the file was produced and whether it still matches its own history, not what the text says. That is why content review and plausibility checks remain the reviewer’s work.

Which documents should get a human look even when the result is INTACT?

Anything that carries real money or a legal decision, anything from a sender you have not dealt with before, and anything whose content does not fit the rest of the case. INTACT removes one question from the list. It does not shorten the list of questions about the case itself.

What if the reviewer and the file check disagree?

Treat it as information. A reviewer who is uneasy about an INTACT file has usually noticed something about the content, which a file check cannot see, so ask the issuer. A MODIFIED file that looks fine to the reviewer is the case a checklist was never designed to catch, so ask for a fresh copy from the source. Neither layer overrules the other.

Is a file check a substitute for training reviewers?

No. Reviewers still need to know the usual layout of the documents they handle, the tactics used against their process and the questions to ask an issuer. A short guide such as how to spot fake pay stubs or how to spot fake invoices covers the content side, and the file check covers what those guides cannot.

How is this different from comparing a metadata viewer with a file check?

A metadata viewer is a tool for reading the properties stored in a file. A human review is a process around the document. They answer different questions again, which is why they have separate pages: what a viewer shows and what a file-history check adds is covered on its own, and PDF tamper detection describes the check itself.

Secure your workflow

Create your account – check PDFs on the web or with an API key, both ready on signup.
No sales call. Cancel any time.