PDF security blog

Due Diligence Data Rooms: Altered Financial Statement PDFs

HTPBE Team··10 min read
Due Diligence Data Rooms: Altered Financial Statement PDFs

This article is a snapshot – content was accurate as of November 2026 (code examples tested against the API as of October 2026). The product evolves actively; specific counts, examples, and detection rules may have changed since publication – see the changelog for the current state.

A data room hands you copies. The management accounts, the audited statements, the bank statements and the tax filings arrive as PDFs that someone on the sell side exported, gathered and uploaded. Your model, QoE work and credit memo all start from those files, and few people ask whether a given file is still the one the source system produced.

A structural file check addresses part of that question for each copy: whether the file carries evidence of being changed after it was created. It does not tell you whether the numbers are right, and it does not prove a file is the one the source system produced. It tells you which files show a detected modification and which it can’t confirm as unchanged, so you know which ones to chase for an original before you rely on them.

This article is for the deal-team associate, corp-dev analyst, transaction adviser or lender’s credit analyst deciding which files in a folder deserve a second request.

Why data-room PDFs deserve a file check

Short answer: everything else in diligence tests the content of the documents. The usual process rarely tests the files themselves. A file check fills that gap cheaply and early, before analysis is built on a document you would want to replace.

Diligence reviewers read figures, reconcile them and question management. That work assumes the PDF in the room reflects what the company’s systems produced. An edit made after export can sit inside a perfectly plausible set of numbers. A file check reads the PDF’s structure and history instead of its figures, so it can raise a question the content review would not.

What a deal team receives, and who controls the files

A typical financial data room holds some mix of:

The sell side controls these files until they land in the room; the platform controls who can open, download and print them. Access permissions decide who gets a copy; they say nothing about whether the copy was edited before it was uploaded. A file check looks at whether the copy was edited.

Copies in a data room may also have been processed after they left the source system, by the platform or by whoever prepared the upload, and that processing is often entirely legitimate. A file check’s result describes the copy you hold, not the file the source system produced. A finding may reflect how the copy was handled rather than anything the target company did, and a result on the copy does not carry over to the original in either direction. When a document matters, ask for the original export directly from the source system or issuer, and check that file too.

What structural analysis can and cannot show

HTPBE? is a PDF tamper detection tool. It reads how the file was built and what evidence it carries of changes after creation, not what the figures on the page say.

It can show that a copy carries evidence of being changed after it was created, and name what kind of finding that is. It cannot show:

  • That the figures are true. A file check reads the file, not the figures, so it cannot tell you whether the numbers were right in the first place: content that was false from the start carries no evidence of a later edit. intact means no modification was detected, not that the document is genuine.
  • That the issuer stands behind it. HTPBE? does not contact the auditor, the bank or the tax authority, and does not check any issuer’s records.
  • Who changed a file, or why. modified describes the file, not intent.
  • That a scan is genuine. A scan or photo can’t prove where the paper came from, so scanned or photographed documents are never confirmed intact.

One more note on management accounts. Many are built in a spreadsheet or word processor and exported to PDF. A file exported like that is not confirmed intact: without other evidence it comes back inconclusive, and modified when the file shows evidence of editing. An inconclusive result here is routine and says nothing bad about the seller. For these documents the stronger follow-up is reconciliation to the ledger, source-system access or a confirmation.

What a result looks like, and what to do with each verdict

Every check returns one of three verdicts. Agree on what each one triggers before the first file goes through.

VerdictWhat it means for this copyWhat the deal team does
intactNo modification was detected, and the copy is of a kind the check can assessCarry on with normal diligence. It is not a certificate for the document or its figures.
modifiedThe copy shows evidence of a modification after it was createdRequest the source original from the system or issuer, and route the file to the deal lead for human follow-up.
inconclusiveNo modification was found, but the check can’t confirm the copy is unchangedRequest the original, or confirm the content with the issuer or the source system. Not a negative finding about the seller.

A verdict routes a file to people. It should never be the sole basis for walking away, repricing or accusing the seller. Asking for an original export is an ordinary diligence request. If the original comes back and the question remains, that becomes a conversation between the deal lead, counsel and the advisers, with both results on file.

The loss model weighs one missed document against the price of checking it, using your own numbers.

A practical data-room workflow

  1. Check on receipt. Run the financial PDFs as they land in the room, before the model is built on them.
  2. Triage by verdict. The follow-up list is every modified and inconclusive file that matters to the thesis or the credit decision.
  3. Request originals. Ask for fresh exports from the source system (the accounting system, the bank portal) or the issuer, and check them.
  4. Escalate to humans. Anything still unresolved goes to the deal lead and the relevant adviser, with the results attached. The decision stays with them.
  5. Record. Keep the verdict and the named findings with the document index, so the diligence file shows what was checked, when, and what was asked for in response.

The web upload suits a handful of files, but read the confidentiality note below first. If your NDA does not allow a third party to keep copies, or you run many deals, use the API: POST /api/v1/analyze takes an HTTPS URL for a PDF up to 10 MB and returns a check ID, and GET /api/v1/result/{id} returns the full result. For deal documents, use a short-lived, unguessable link that expires soon after the check has run (the private-bucket guide shows how), and pass a neutral original_filename so the stored name carries no deal codename. The file’s own metadata, such as its title and author fields, is stored with the result too, so check it before submitting. An excerpt:

{
  "status": "modified",
  "modification_markers": ["HTPBE_DATES_DISAGREE"]
}

modification_markers lists named public IDs for what was found, which is what you attach to the diligence record. GET /api/v1/checks lists past checks, filterable by status and date, which helps when you need to report on a whole room. There is no data-room integration: you submit files the same way as any other PDF.

On confidentiality, read your NDA and the data room’s terms before sending anything out of the room. On our side, files submitted through the API are fetched, analysed once and discarded, and are not used to improve the algorithm. Files uploaded through the website may be retained after analysis and used to test and improve the algorithm, kept without a link to your account; a deletion request removes a retained copy, but a copy already included in an evaluation set may no longer be individually retrievable. Whichever route you use, the result of each check, including the filename and the file’s technical metadata, is kept for up to three years. Details are in the privacy policy.

Where the check fits next to QoE, audit and legal review

A file check is one narrow control. It complements the work you already do and replaces none of it:

  • Quality of earnings tests whether the reported earnings hold up. A file check says nothing about earnings.
  • The audit tests the numbers in the statements the auditor was given. ‘Our auditors cover this’ is true for the figures; whether the copy in your data room matches what was audited is a different question. A file check can flag a copy that was changed after it was created; it cannot confirm the copy is the version the auditor signed.
  • Source-system access, such as read access to the accounting system or the bank portal, and confirmations from banks or tax authorities, are stronger evidence than any copy. Where you can get them, use them. A file check helps decide where to push for them.
  • Legal review reads agreements for what they say. For contracts in the room, contract tamper detection covers the file side.

Lenders will find the same pattern on the lending page; the proof-of-funds article covers one adjacent document class in depth.

Frequently asked questions

Does an intact result mean the financial statements are correct?

No. HTPBE? checks the PDF file, not the figures in it. intact means no modification was detected in that copy, not that the document is genuine or the numbers are right. A file check cannot tell whether the figures were right in the first place. Whether the numbers hold up is a question for quality of earnings, the audit and source-system access.

Should a modified result change the deal?

Not on its own. modified means the copy you hold shows evidence of a change after it was created. It does not say who made the change or why, and the change may come from legitimate processing of the copy. Request the original export from the source system or issuer and route the file to the deal team. A verdict should never be the sole basis for walking away, repricing or an accusation.

Can we run confidential deal documents through HTPBE?

Check your NDA and the data room’s confidentiality terms first. Files submitted through the API are fetched, analysed once and discarded, and are not used to improve the algorithm. Files uploaded through the website may be retained and used to test and improve the algorithm, kept without a link to your account, and a copy already included in an evaluation set may no longer be individually retrievable. If the NDA does not allow a third party to keep copies, the API is the route that does not keep the file. Either way, the result of each check, including the filename and the file’s technical metadata, is kept for up to three years. The privacy policy has the details.

Try it on a folder you already have

Take a handful of financial PDFs from a closed deal, run them, and see where the verdicts fall before you write the step into your diligence checklist. Create an account for 5 welcome credits; after that, a single deal team can pay per check (8 credits for $5, 25 for $15, plus VAT where applicable), and teams running many deals can use a plan or the API. See pricing for details.

Share this article

Found this article helpful? Share it with others to spread knowledge about PDF security and fraud detection.

https://htpbe.tech/blog/data-room-due-diligence-financial-statement-pdfs

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.