Skip to content
PDF privacy guide · architecture and evidence

Local vs server processing: what really happens to your PDF

Two PDF tools may look almost identical while handling the document in very different ways. One may perform the operation inside your browser; another may upload the file to a server and return the result.

The essential difference

With local processing, the document operation runs on the user's device. With server processing, the document must reach remote infrastructure so the operation can run there. Hybrid architectures also exist, so the real workflow matters more than the appearance of the website.

What happens in a local workflow

  1. The application and required components load in the browser.
  2. The user selects the PDF.
  3. The document is read through browser-local APIs or memory.
  4. The transformation runs on the device.
  5. The result is generated and offered locally.

What happens in a server workflow

  1. The user selects the PDF.
  2. The browser transfers data to remote infrastructure.
  3. The server receives the file or its contents.
  4. The operation runs remotely.
  5. The resulting document is returned to the browser.

Hybrid models also exist

A service can process some operations locally and others on a server. Behaviour may even change with file size, feature, browser or software version. A test should therefore always describe the exact operation and conditions tested.

Local and server processing compared

Aspect Local processing Server processing
Document required outside device No for the tested local operation Yes for the remote operation
Internet after application load May not be required for the tested flow Normally required to send and receive the document
Large file upload Not required for local processing Common signal of a remote workflow
Computing resources Device CPU and memory Provider infrastructure
Document exposure Stays on the device during local processing Reaches provider infrastructure

How to distinguish them with a reproducible test

You do not have to rely only on marketing statements. You can observe the actual behaviour.

  1. Use a synthetic PDF with no personal information.
  2. Fully load the tool first.
  3. Disconnect before selecting the document.
  4. Check whether the operation completes and produces a valid result.
  5. Repeat the same operation online and open browser developer tools.
  6. Inspect POST and PUT, multipart/form-data, application/pdf, large requests and upload endpoints.
  7. Separate small telemetry events from actual document transfer.

What an offline test actually proves

If an operation completes after preloading the application, disconnecting, and only then selecting the document, that strongly supports the conclusion that the tested operation does not require a server in that flow. It does not prove that every feature is local or that the application can cold-start offline.

HTTPS does not determine where the PDF is processed

HTTPS encrypts communication between browser and server. It protects a transfer, but it does not prevent a transfer from occurring.

Telemetry is not the same as document transfer

A page can send performance metrics, analytics or navigation events without sending the PDF. To determine whether the document leaves the device, inspect request content, size, type and destination.

Why this matters for privacy

When a document is processed locally, the provider does not need to receive it for that operation. Remote processing requires the file to cross the network and reach infrastructure outside the device. That changes the exposure model and raises questions about storage, retention, jurisdiction and access.

What we verified in PDFPrivado

In PDFPrivado's published audit, all 14 tested tools completed their workflow after the tool was preloaded, the connection was disabled, and synthetic documents were selected afterwards. In the equivalent online control we observed no document transfer during processing. A fully offline cold start was not tested.

See methodology and evidence in Privacy Lab →

Check it yourself

Our practical guide explains how to detect transfers, interpret POST and PUT requests, use canary PDFs and avoid incorrect conclusions.

How to tell if a PDF tool uploads your documents → Back to the Guides Center

Frequently asked questions

Does local processing mean there is never an Internet connection?

No. The application may need Internet access to load code, update components or send telemetry. The key question is whether the document must be transferred for the operation.

Does a blob: result prove everything happened locally?

Not by itself. You must also check whether an upload or remote processing request happened earlier.

Does working offline prove files are never uploaded?

No. It proves the behaviour of the tested workflow under the defined conditions.

Does remote processing mean a tool is unsafe?

Not necessarily. It means the document is transferred for processing outside the device, which is a different privacy model.

Published and verified: September 6, 2026