Quality Rater Guidelines(QRG)

Long definition

Google has employed external human quality raters since at least 2003, and the QRG is the document those raters use to evaluate test searches. Google began publishing the QRG officially in 2015 (after years of leaks). The current version is hosted at services.google.com/fh/files/misc/hsw-sqrg.pdf and is updated 2-3 times a year, typically with incremental changes plus occasional major reframings.

How the QRG fits into ranking:

1. Google's engineers propose changes to ranking algorithms or run experiments.
2. Quality raters use the QRG to evaluate the experiment's search results — judging page quality, query-result fit, and overall usefulness.
3. The aggregate ratings inform whether the experiment ships. Raters do not directly affect live rankings; they evaluate what the algorithm produces.
4. Over time, ratings serve as labeled training data that ranking models learn to predict.

The QRG defines several core concepts now ubiquitous in SEO:

• E-E-A-T — Experience, Expertise, Authoritativeness, Trustworthiness.
• YMYL — Your Money or Your Life topic classification.
• Page Quality (PQ) ratings — Lowest, Low, Medium, High, Highest.
• Needs Met ratings — Fails to Meet, Slightly Meets, Moderately Meets, Highly Meets, Fully Meets — the rater's judgment of how well a result satisfies the query intent.
• Lowest quality red flags — deceptive design, malicious behavior, harmful content, content created with no effort.
• Reputation research — raters are explicitly instructed to look up the website and its authors elsewhere on the web.

December 2022 was a notable QRG update — Experience was added to E-A-T, making it E-E-A-T. The update aligned the guidelines with the Helpful Content System's emphasis on first-hand experience. Subsequent updates have refined the relationship between page quality and E-E-A-T.

The strategic use for SEO: read the QRG before writing important content. It tells you, from Google's own pen, what a good page looks like. Most "best practices" articles paraphrase the QRG indirectly; reading the source removes the telephone-game distortion.

Common misconceptions

• "Raters directly demote my site." They don't. Raters evaluate experiments and quality samples. Their ratings are aggregated training signals. No individual rater's judgment lowers your specific site's rankings.
• "The QRG is the Google algorithm written down." It's the target the algorithm is trained against. The actual algorithm is a complex stack of ML models, classical ranking signals, and policy systems. The QRG describes what good output looks like, not the mechanics of producing it.
• "Reading the QRG lets me reverse-engineer rankings." It tells you what Google considers high-quality. The "how" is opaque — many ML signals correlate with QRG criteria without being explicit implementations of them.
• "The QRG is leaked or unofficial." It's published officially by Google. The PDF link above is the current source of truth. Older copies leaked before official publication began in 2015 and shaped early SEO understanding of E-A-T.