ATS Resume ToolsPARSE LAB
Menu

Protocol R-01 / v1.0

PDF reading-order test

Measure whether visible columns extract in the intended sequence. This is a protocol, not a completed comparison.

Status: protocol readyResults: pending
Last reviewed 2026-08-23

Research question

When the words stay the same, how does page layout change extracted order?

The test compares a one-column baseline with controlled two-column layouts. Every file uses the same text, typography, margins, and export software.

Document set

Four layouts

A / BASELINE

Single column

Expected top-to-bottom reading order.

B / SIDEBAR

Narrow skills sidebar

Primary work history remains in the wide column.

C / BALANCED

Balanced columns

Work entries continue in both columns.

D / POSITIONED

Positioned blocks

Text boxes imitate the two-column page.

Markers

Use tokens that expose sequence

Each content block begins and ends with a unique marker. Expected order is stored before the test. The plain-text output can then be compared without guessing where a sentence belonged.

[[WORK_01_START]]
Data Engineer | Northstar Labs | 2023-present
...
[[WORK_01_END]]

Run matrix

Named extractors only

Record each PDF extraction library, parser API, or ATS environment separately. A result should never be labeled “ATS” without naming the software that produced it.

  • Extractor or system name
  • Version or visible release date
  • PDF generator and version
  • Input file hash
  • Raw output retained
  • Repeat run after any failure

Scoring

Report order errors, not a universal pass percentage

Recorded measures
MeasureDefinitionFailure example
Block orderUnique block markers appear in the expected sequence.Skills sidebar appears between a job title and its bullets.
Block integrityText between start and end markers stays together.A bullet from column two enters the first job.
Character fidelityLiteral characters match the manifest.Ligature or accented name is substituted.
Field adjacencyTitle, employer, and date remain neighboring content.Date moves to another entry.
References

Sources used on this page

  1. What you may be missing when you search PDF documentsPDF Association

    Demonstrates how PDF content order can differ from the order visible on the page.

  2. PDF format familyLibrary of Congress

    Notes that scanned-image PDFs do not necessarily support text indexing.

  3. Recognize text in scanned PDFs with AcrobatAdobe Help Center

    Explains that OCR adds a searchable text layer to a scanned PDF.