OriginalContentScore: How Google Measures Content Originality

What is OriginalContentScore?

OriginalContentScore measures how much original material is in a content — information that is only found there and that an AI couldn't reconstruct from other sources. The higher the score, the more original material the content contains. This is the Experience dimension of E-E-A-T: does the content reflect first-hand knowledge, or is it merely a rephrasing of existing information?

The Google API leak confirmed the existence of an OriginalContentScore field. The field name itself — 'OriginalContentScore' — is unusually descriptive for a Google system field. Most field names in the leak are cryptic (chard, tofu, keto). The descriptive name suggests this signal was designed to be understandable, not just functional.

OriginalContentScore is the primary Experience signal (E: primary). This means it is the main system Google uses to evaluate whether content demonstrates genuine first-hand experience. A page that reports original research, shares new data, or describes a personal experience will score higher than a page that summarizes existing information.

Originality vs. duplication

OriginalContentScore distinguishes between content that is original (new information) and content that is duplicated or rephrased (existing information presented differently). A page that quotes a government report verbatim has low originality. A page that analyzes the report, adds context, and draws new conclusions has high originality.

The score is not binary — it is a spectrum. A news article that breaks a story has very high originality. A follow-up article that adds one new quote has moderate originality. A roundup article that links to five existing articles has low originality. Google can distinguish these because it has indexed the source material and can compare.

This has implications for content strategy: simply rephrasing existing content — even with different words — does not increase OriginalContentScore. The system measures information novelty, not linguistic variation. A 2000-word article that says the same thing as a 500-word article it links to has low originality despite being longer.

OriginalContentScore in the ranking architecture

OriginalContentScore is fed by IS-Calibration (System 12) — the human rater system. Raters evaluate whether content demonstrates original research, first-hand experience, or novel analysis. OriginalContentScore learns to predict what a rater would say about a new page's originality.

The system feeds into NSR (System 7), the normalized site rank. This means a site with consistently high originality scores will have a higher NSR value, which feeds into Q* (System 6), the site-level quality score. Originality is thus a site-level quality factor, not just a page-level one.

OriginalContentScore operates at the document level (Reach: Doc) — it evaluates individual pages, not entire sites. However, the aggregation into NSR means the site-level effect is real: a site where most pages have high originality will rank better overall than a site where most pages are rephrased content.

Implications for SEO practitioners

The most important implication: content that merely rephrases existing sources will not score well on OriginalContentScore. To increase originality, content must add new information — original data, new analysis, first-hand experience, or expert commentary that isn't available elsewhere.

For data-driven sites, originality comes from the data itself: proprietary models, unique measurements, location-specific analysis that no other site produces. This is 'original material' in the strictest sense — information that exists nowhere else.

The system's connection to the Experience dimension means that first-hand experience signals matter. Content that describes personal experience ('I tested this product for 30 days') has higher originality than content that summarizes others' experiences ('Reviewers say this product is good'). This is why Google added the 'E' to E-E-A-T in December 2022.

AI-generated content faces a specific challenge with OriginalContentScore: if an AI generates content by synthesizing existing sources (which is what LLMs do), the output has low originality by definition. The AI is rephrasing, not creating. This doesn't mean AI content is penalized — but it means AI content that adds no new information will score low on originality, which affects the Experience dimension.