PAPERGRID RESEARCH
Original PDF benchmarks, with reproducible evidence.
PaperGrid publishes small, deterministic browser studies so its claims can be inspected instead of taken on trust. Each study includes the exact method, machine-readable results and limits on what the evidence proves.
These tests use synthetic files rather than private documents. They measure specific production behaviors; they are not market-wide rankings and should not be read as universal performance guarantees.
MERGE INTEGRITY
Does merging preserve document structure?
Three deterministic source PDFs were combined into a six-page output in an isolated browser session. All eight defined checks passed: reopening, page count, selected file order, selectable marker text, rotations, page sizes, a custom crop box and rendering of every page.
The study did not test encryption, signatures, forms, embedded files, malformed PDFs or every font configuration.
CROSS-FILE PAGE ORDER
Does an interleaved page plan survive export?
A separate six-page browser run deliberately crossed three source boundaries in the order C2, A1, B2, C1, A2, B1. All eight checks passed for exact label order, selectable text, rotations, dimensions, one crop box, reopening and rendering.
The study tests a deterministic page-copy plan; separate product tests cover the visual organizer controls, mobile width, CSP and offline operation.
COMPRESSION
When does browser compression help?
Three deterministic fixtures tested image-heavy, mixed and already compact text-first PDFs. Image-heavy candidates were 88.1% to 94.1% smaller and mixed candidates were 84.5% to 92.3% smaller. Every candidate for the text-first fixture was equal to or larger than its source and was rejected by the byte-size safeguard.
This is a small synthetic study, not an average saving and not a competitor comparison.
VERIFIABLE PRIVACY
Can local processing be checked?
PaperGrid documents a repeatable network-panel procedure and a returning-user offline test. Its merger also uses same-origin scripts and connections under a browser-enforced Content Security Policy. These controls reduce exposure; they do not make every browser, extension or device trustworthy.
HOW TO CITE THE RESULTS
Link to the study and keep its boundary.
When referencing a result, link to the human-readable study and identify the fixture, browser date and measured outcome. Use the JSON distribution when exact values are needed. Do not generalize a fixture result into an average for arbitrary documents.
TEST THE PRODUCT