A practical document verification guide for online newsrooms

Press releases, minutes, contracts, notices, internal memos, and screenshots can make reporting more specific, but a document-like appearance does not make a claim true. A source may provide only selected pages, an old draft may circulate as a final decision, or an authentic file may arrive with a misleading explanation. Familiar editing and generation tools also make logos and typography weak substitutes for verification.

The goal is not to declare absolute authenticity for every file. It is to separate what the newsroom has established from what remains unknown and to collect evidence appropriate to the claim being published. This is not legal or forensic advice. It is a repeatable operating workflow for a small online newsroom, from intake through publication and retention.

Preserve the received item before working on it

Save the item as received before opening it for edits or converting it to another format. Record whether it arrived by email attachment, message download, physical delivery, or shared link, along with the time of receipt. Keep the original filename and extension when practical, and make a working copy for highlights and notes.

If the source provides only a screenshot, ask for the original file or a fuller screen recording when appropriate. A screenshot may omit surrounding context, an address, timestamps, revision history, and file properties. For paper records, capture every side, envelope, attachment, and visible page number so missing material can be identified.

The intake record should name the assigned reporter, storage location, access level, and the explanation supplied by the source. Do not expose a confidential source's identity in a filename or folder shared with the document. Preserving an original does not prove authenticity, but it creates a stable reference for comparison and later review.

Define the claim the document needs to support

Before examining details, write one sentence describing what the story expects the document to establish. “Is this real?” is less useful than “Did organization A issue this decision on date B?” A single file may support authorship, dispatch, receipt, or implementation to different degrees.

Separate the document's existence from the truth of its contents. An authentic internal report may contain inaccurate figures, and a signed plan may never have been implemented. A draft can still be relevant to what an organization considered, but it should not be described as a final decision.

In the reporting notes, divide facts directly visible in the document, facts that require independent reporting, and interpretations that remain uncertain. This prevents a headline or lead from extending beyond the evidence.

Check the source and the transmission path separately

One source is the person who provided the file; another is the organization or person that created or held it. Ask the provider when, where, and how it was obtained and whether they saw it directly or received it from someone else. An incomplete answer is not automatically evidence of deception, so record what can and cannot be checked.

Contact the apparent issuer through a published website, switchboard, or other independently obtained channel. Do not rely only on a number supplied with the tip. Ask about the document number, title, department, date, and relevant public personnel without sending the complete file or unnecessary identifying details when source protection matters.

Design questions so one verification route does not contaminate another. In addition to asking whether the organization produced a file, ask about its document format, numbering system, and distribution process at that time. A refusal or failure to respond is not proof of fabrication; describe the request and response status accurately.

Use formatting as a clue, not a verdict

Compare logos, reference numbers, date styles, page numbering, role titles, signature blocks, headers, and footers with public documents from the same organization. Check whether terminology, department names, contact domains, and titles existed at the stated date. Use comparison samples from a nearby period so normal redesigns are not mistaken for anomalies.

Do not call a file false because of one formatting error. Internal material may use a different template, and rushed drafts may contain alignment mistakes. Conversely, a perfectly matched format can be copied from a genuine document while the content is replaced.

List anomalies and consider alternative explanations for each. A note such as “the logo is wrong” is weak. A note explaining that the proportions differ from three documents published in the same month and that the organization denied using that version separates observation from corroboration.

Build a timeline of dates and versions

Creation, approval, dispatch, receipt, and publication dates may all differ. Put dates from the text, filename, email receipt, and file properties in a simple timeline and compare them with the events in the story. Check whether a department name or official title that appeared later is present in a supposedly older record.

When drafts, revisions, and final versions are available, compare their language. A single page may have been replaced, or an attachment list may have changed while the date on the main text remained the same. Creation and modification timestamps can change during copying, scanning, and export, so do not treat them as standalone proof.

If a major conclusion appears in only one version, report who received that version and whether it was used. Distinguish carefully among “was drafted,” “was approved,” and “was implemented.”

Examine technical properties in a reproducible way

PDFs and images may retain information about creation software, timestamps, dimensions, or scanning. Use a documented newsroom method or trusted internal tool, work from a copy, and record what was checked and when. Keep screenshots or text exports of relevant results where newsroom policy permits.

Unexpected technical data does not establish manipulation. Scanning, email delivery, document merging, and cloud export can remove or replace properties. OCR errors may come from text recognition rather than the original document, so compare extracted text with the visible page.

When a high-risk story requires deeper file analysis, do not force a conclusion from general-purpose tools. Consider an appropriately qualified review and publish only the scope that review supports. Technical vocabulary should clarify evidence, not disguise uncertainty.

Corroborate the contents independently

Compare important names, amounts, addresses, schedules, attendees, and project titles with official public records, contemporaneous coverage, and separate interviews. Multiple pages copied from the same originating source are not independent confirmation. Look for evidence created through a different route.

Read the paragraphs and attachments around any proposed quotation. A conditional statement, proposal, or dissenting view can be misrepresented when clipped to one sentence. For tables, inspect units, reference dates, footnotes, and totals. If the newsroom recalculates a figure, retain the formula and location of the source values.

Offer people named in the document a meaningful chance to address specific claims. If sharing the full file would expose a source or unrelated private information, provide only the relevant claim or passage needed for a response. Record separately whether a response concerns authenticity, interpretation, or the underlying events.

Remove sensitive information from the publication copy

Information needed for verification is not necessarily needed by readers. Check for identification numbers, account details, personal contacts, signature images, exact home addresses, access links, and codes in images and PDFs. Placing a black shape over text may not remove selectable content underneath, so test the publication copy itself.

After redaction, try copying, searching, zooming, and downloading the file. Filenames, metadata, annotation layers, and hidden pages can also retain identifying information. If the story does not require the complete document, a limited quotation or newsroom-created table may be safer and clearer.

Keep the verification original and the public copy unmistakably separate. Record who prepared the public version and who performed the final check to reduce the risk of attaching the wrong file during a rushed publication.

Match article language to the level of confirmation

If the issuing organization confirms a record, the story can say so. If formatting and external facts align but the issuer does not confirm it, explain the acquisition and the checks performed rather than stating that authenticity has been established. Verification of several passages should not be described as verification of the entire file.

Labels such as “exclusive,” “internal document,” and “final report” each make a separate claim about access, circulation, or version status. Do not add them by habit. Headlines should also avoid verbs that turn a proposed action or written assertion into a completed event.

When useful, include a short methods note. Explain which organization was contacted, which materials were compared, and what remains unconfirmed. Readers can then judge the strength and limits of the evidence.

Keep a verification log tied to the story

For each document, record an identifier, receipt time, original storage location, source account, confirmation requests, responses, comparison materials, unresolved questions, and publication decision. Link important quotations to page and paragraph references so an editor can find them quickly.

Distinguish observation from inference. Labels such as “observed,” “source statement,” “external confirmation,” and “editorial assessment” prevent a reporter's tentative note from becoming a settled fact through copying. Set access and retention according to the sensitivity of the material and newsroom policy.

If another version or response arrives after publication, append it to the log rather than overwriting the earlier record. Connect any article change to the evidence that prompted it. This makes corrections and follow-up reporting faster and more accountable.

Start with a two-stage review in a small newsroom

The reporter preserves the original and checks provenance, timeline, key contents, and sensitive data using a checklist. An editor then examines the evidence attached to each central claim, whether uncertainty appears in the wording, and whether the public file is safe. Add specialist review when the stakes and unresolved questions require it.

A complicated system is not necessary at the start. Create three shared templates: an intake record, a claim-to-evidence table, and a prepublication redaction check. Apply them to one recent document-based story. The points where the team repeatedly pauses are the next improvements to make.

Strong document reporting does not end with trusting a file's appearance. Preserve the original, narrow the claim, confirm provenance, timeline, and contents through separate routes, and make the article language reflect the actual level of certainty. A recorded workflow helps a small newsroom protect both speed and credibility.