Tables and text boxes are layout containers, and containers are where resume content goes to disappear. Parsers read a document's main text flow; content inside a table cell or floating text box sits outside that flow, so it extracts out of order, detached from its context, or not at all — and the file looks perfect to you the whole time.
A word processor document has a main story — the stream of text a parser walks from top to bottom. Tables store text cell by cell, and extractors read them in cell order: left to right, row by row, or sometimes column by column. A two-column table pairing jobs with dates can extract as all the jobs followed by all the dates, orphaning every date from its role. The pairing you see on screen exists only visually.
Text boxes are worse: they are anchored floating objects, positioned outside the main flow entirely. Extraction libraries differ — some append text-box content at the end of the document, some interleave it unpredictably, some skip it. A sidebar built from text boxes holding your skills and certifications can simply not exist in the parsed record. Because rendering is flawless, nothing warns you.
Almost nobody inserts a table deliberately. Containers arrive through templates: the elegant Word template whose entire layout is one invisible borderless table, the design-tool export where every text block is a floating object, the two-column look built from a table with hidden gridlines. Borderless tables are still tables; the parser does not care that the lines are invisible.
Detection takes one minute. In Word, click into any part of your resume and watch for the table-layout tab to appear in the ribbon, or turn on formatting marks to reveal anchors. In any exported PDF, run the copy-paste test: select everything, paste into a plain-text editor, and check that reading order survives and nothing is missing. Out-of-order or absent content in the paste is exactly what the ATS will store.
Everything tables are used for on resumes has a flow-safe equivalent. Job title on one line, employer and date on the next — or the date pushed right with a right-aligned tab stop, which lives safely inside the text flow. Skills in a plain list, comma-separated or in short lines. Section headings as ordinary styled paragraphs.
If you love the two-column aesthetic, keep it for the version humans see by choice — a portfolio PDF, a printout. The uploaded version should be a single flow a parser can walk without guessing. Rebuild once on a clean skeleton, verify with a parser-based checker, and the whole category of silent container loss stops applying to you.
Yes, the same problem as a visible one. Parsers read table structure, not border styling, so a borderless layout table still extracts cell by cell in an order you do not control. Rebuild the layout with ordinary paragraphs and tab stops instead.
In Word, click each block of content — floating boxes show a selection frame with an anchor symbol. Or run the copy-paste test on the exported PDF: text-box content typically pastes at the wrong position or not at all. Anything missing from the paste is invisible to parsers too.
Safer than tabling your work history, since skills are order-independent single words — but still unnecessary risk, because some extractors mangle or skip cells entirely. A comma-separated plain-text list gives parsers and recruiters identical information with zero failure modes.
The free scorer parses your actual file and flags content that extracts out of order or goes missing — the exact failures tables and text boxes cause.
