Editorial QA is a release check for truth, usefulness, and working links—not a final spellcheck pass. A page should fail review if its answer is generic, its example has no evidence label, or its Arabic reads like English syntax wearing Arabic words.
Review the article in four passes
1. Task: Can a reader tell what problem the page solves and get the answer before the background? Does each section add a new step, distinction, or piece of proof?
2. Evidence: Highlight changing product instructions, platform features, policies, legal claims, prices, and statistics. Verify them against current primary sources and record the checked date. Label hypothetical examples. Remove claims that lack support.
3. Local language: Have an Arabic editor check the target register, terminology, number and currency formatting, RTL tables, punctuation, and whether the advice actually fits the stated country. Avoid a single “MENA” example that silently mixes market rules.
4. Page behavior: Open every internal and external link, check canonical and language alternates, review structured data against visible content, and test mobile display. A correct-looking metadata block does not compensate for a broken article.
Use a stop-the-line rule
Do not publish if the article has unfilled blanks, repeated paragraphs, unsupported client results, invented expertise, a mismatched title, an inaccessible primary source, or a call to action that goes nowhere. Assign an owner and a correction deadline. For a time-sensitive instruction, add a review trigger rather than a decorative “updated” date.
Keep a small editorial record
Store the intended reader, primary task, distinctive contribution, sources, Arabic reviewer, publication date, and next review condition. This makes later refreshes faster and helps prevent two pages from competing to answer the same task.
AI disclosure: AI tools assisted with drafting. Human review remains necessary for factual claims and market-specific language.