SiftServe scoring methodology – CORE-EEAT Content Benchmarks
· Rafi
Key takeaways
This page covers: What is CORE-EEAT; How to read the scores; Dimension and total scores; Contextual Clarity (agent experience, GEO); Organization (agent experience, GEO); Referenceability (agent experience, GEO); Exclusivity (agent experience, GEO); Experience (human experience, SEO).
- CORE-EEAT is an 80-item content-quality rubric: eight dimensions of ten checks each, half scored for AI agents (GEO), half for human readers and search (SEO).
- This post is the reference for anyone reading a SiftServe audit report, and for anyone who wants to grade their own content by the same standard. It lists every check in all eight dimensions, the scoring formulas, and the per-content-type weights.
- Every SiftServe page is scored twice, once for the original and once for the AI-readable sifted copy, so the improvement is a measured before-and-after pair, not a claim.
- Weights differ by content type. A landing page is graded on different priorities than a how-to guide or an FAQ page.
- The rubric is public. It builds on the open-source CORE-EEAT content benchmark, and every check is listed below so you can grade your own pages against it.
What is CORE-EEAT
CORE-EEAT is the scoring standard behind every audit score SiftServe shows: 80 checks across eight dimensions, each dimension scored 0–100. The first four dimensions (Contextual Clarity, Organization, Referenceability, Exclusivity) measure how well AI agents can find, parse, and cite a page; together they form the GEO score. The last four (Experience, Expertise, Authority, Trust) are Google's E-E-A-T framing of credibility for human readers and search; together they form the SEO score.
The rubric builds on the open-source CORE-EEAT content benchmark by Aaron He Zhu, with the agent-experience checks grounded in the Princeton GEO study (Aggarwal et al., KDD 2024) and the human-experience checks grounded in Google's Search Quality Rater Guidelines.
This post is the reference for anyone reading a SiftServe audit report, and for anyone who wants to grade their own content by the same standard. It lists every check in all eight dimensions, the scoring formulas, and the per-content-type weights. It does not cover the capture stats (page weight, tokens, coverage); those have their own methodology post.
How to read the scores
Every page is checked twice using the same 80-item standard. The first check is for the original page, and the second is for the sifted copy that AI agents read. We compare the agent-experience dimensions, which focus on what AI crawlers and assistants can find and understand, between the original and sifted versions. The human-experience dimensions, which cover credibility signals for readers and search ranking, apply to the original page, since human visitors continue to see it while the sifted copy is shown to AI crawlers.
Our own homepage is the standing example: on the agent-experience view, the original page scores 65 and the published sifted copy (v10) scores 76. Both numbers, along with how they were produced, are in the capture stats post.
A few checks are vetoes rather than points. An undisclosed affiliate relationship, for example, fails the page outright regardless of its other 79 answers.
Dimension and total scores
GEO Score = (C + O + R + E) / 4SEO Score = (Exp + Ept + A + T) / 4Total Score = (GEO Score + SEO Score) / 2
Weighted scoring by content type
Different page types earn trust differently, so the total is also computed as a weighted sum: Weighted Score = Σ (dimension_score × weight)
| Dim / Page type | Product review | How-to guide | Comparison | Landing page | Blog post | FAQ page | Alternative | Best-of | Testimonial |
|---|---|---|---|---|---|---|---|---|---|
| C | 10% | 20% | 10% | 20% | 25% | 25% | 10% | 10% | 10% |
| O | 10% | 20% | 20% | 10% | 10% | 25% | 15% | 25% | 5% |
| R | 15% | 10% | 25% | 5% | 10% | 15% | 25% | 20% | 15% |
| E | 20% | 5% | 10% | 5% | 20% | 5% | 5% | 15% | 10% |
| Exp | 20% | 5% | 5% | 5% | 10% | 5% | 15% | 5% | 30% |
| Ept | 5% | 20% | 15% | 5% | 10% | 10% | 5% | 10% | 5% |
| A | 5% | 5% | 5% | 25% | 5% | 5% | 5% | 5% | 5% |
| T | 15% | 15% | 10% | 25% | 10% | 10% | 20% | 10% | 20% |
Read down a column and the logic shows: a landing page lives or dies on Authority and Trust (25% each), a blog post on Contextual Clarity (25%), a testimonial page on first-hand Experience (30%).
Contextual Clarity (agent experience, GEO)
| Check | Experience | What good looks like |
|---|---|---|
| Intent Alignment | Agent | Title promise = content delivery |
| Direct Answer | Agent | Core answer in first 150 words |
| Query Coverage | Agent | Covers ≥3 query variants (synonyms, long-tail) |
| Definition First | Agent | Key terms defined on first use |
| Topic Scope | Agent | Explicitly states what is and isn't covered |
| Audience Targeting | Agent | States "this article is for…" |
| Semantic Coherence | Agent | Logical flow between paragraphs, no jumps |
| Use Case Mapping | Agent | Decision framework: when to choose A vs B |
| FAQ Coverage | Agent | Structured FAQ covering long-tail follow-ups |
| Semantic Closure | Agent | Conclusion answers the opening question + next steps |
References: GEO: Generative Engine Optimization (Aggarwal et al., KDD 2024) · Google: creating helpful, reliable, people-first content
Organization (agent experience, GEO)
| Check | Experience | What good looks like |
|---|---|---|
| Heading Hierarchy | Agent | H1→H2→H3, no level skipping |
| Summary Box | Agent | Has TL;DR or Key Takeaways section |
| Data Tables | Agent | Comparisons and specs presented in tables |
| List Formatting | Agent | Parallel items use bullet or numbered lists |
| Schema Markup | Agent | Appropriate JSON-LD (Article/FAQ/HowTo/etc.) |
| Section Chunking | Agent | Each section has single topic; paragraphs 3–5 sentences |
| Visual Hierarchy | Human | Key concepts bolded or highlighted |
| Anchor Navigation | Agent | Table of contents with jump links |
| Information Density | Agent | No filler; consistent terminology throughout |
| Multimedia Structure | Human | Images/videos have captions and carry information |
References: The RefinedWeb dataset: filtering web data for LLM training (NeurIPS 2023) · GEO: Generative Engine Optimization (Aggarwal et al., KDD 2024)
Referenceability (agent experience, GEO)
| Check | Experience | What good looks like |
|---|---|---|
| Data Precision | Agent | ≥5 precise numbers with units (%, $, ms) |
| Citation Density | Agent | ≥1 external citation per 500 words |
| Source Hierarchy | Agent | Primary sources first; ≥3 Tier 1–2 sources |
| Evidence-Claim Mapping | Agent | Every claim backed by evidence immediately after |
| Methodology Transparency | Agent | Sample size, steps, and criteria documented |
| Timestamp & Versioning | Agent | Last updated <1 year; version changes noted |
| Entity Precision | Agent | Full names for people/orgs/products; no "a company" |
| Internal Link Graph | Human | Descriptive anchor texts forming topic clusters |
| HTML Semantics | Agent | Uses <article>, <figure>, <time>, <cite> |
| Content Consistency | Agent | Data self-consistent; no broken links (404) |
References: GEO: Generative Engine Optimization (Aggarwal et al., KDD 2024) · Google: how structured data works
Exclusivity (agent experience, GEO)
| Check | Experience | What good looks like |
|---|---|---|
| Original Data | Agent | First-party surveys, experiments, or statistics |
| Novel Framework | Agent | Named, citable original framework or model |
| Primary Research | Agent | Original experiments/surveys with documented process |
| Contrarian View | Agent | Challenges consensus with evidence |
| Proprietary Visuals | Human | ≥2 original infographics, charts, or diagrams |
| Gap Filling | Agent | Covers questions competitors don't |
| Practical Tools | Human | Downloadable templates, checklists, or calculators |
| Depth Advantage | Agent | Deeper than competing content on same topic |
| Synthesis Value | Agent | Cross-domain knowledge combination (A+B=C) |
| Forward Insights | Agent | Data-backed predictions and trend analysis |
References: Google: creating helpful, reliable, people-first content · GEO: Generative Engine Optimization (Aggarwal et al., KDD 2024)
Experience (human experience, SEO)
| Check | Experience | What good looks like |
|---|---|---|
| First-Person Narrative | Human | Contains "I tested" or "We found" + action verbs |
| Sensory Details | Human | ≥10 sensory words (smooth, heavy, bright) |
| Process Documentation | Agent | Step-by-step process with timeline |
| Tangible Proof | Human | ≥2 original photos/screenshots with timestamps |
| Usage Duration | Human | States "after X months of use…" |
| Problems Encountered | Agent | Shares ≥2 real problems + solutions |
| Before/After Comparison | Human | Shows change, improvement, or difference |
| Quantified Metrics | Agent | Measurable experience data (time, cost, success rate) |
| Repeated Testing | Human | Multiple tests or long-term tracking |
| Limitations Acknowledged | Agent | States "we only tested X scenario" |
References: Google: E-E-A-T in the Search Quality Rater Guidelines · Google Search Quality Rater Guidelines (PDF)
Expertise (human experience, SEO)
| Check | Experience | What good looks like |
|---|---|---|
| Author Identity | Human | Byline + avatar + bio (>30 words) |
| Credentials Display | Human | Relevant degrees, certs, years of experience |
| Professional Vocabulary | Agent | Accurate industry jargon, no misuse |
| Technical Depth | Agent | Parameters, thresholds, examples are actionable |
| Methodology Rigor | Agent | Analysis method is reproducible |
| Edge Case Awareness | Agent | Discusses ≥2 exceptions or "when this doesn't apply" |
| Historical Context | Human | Shows knowledge of the field's evolution |
| Reasoning Transparency | Agent | "We chose A over B because…" with tradeoffs |
| Cross-domain Integration | Agent | Connects knowledge across fields |
| Editorial Process | Human | "Reviewed by" or "Fact-checked by" labels |
References: Google Search Quality Rater Guidelines (PDF)
Authority (human experience, SEO)
| Check | Experience | What good looks like |
|---|---|---|
| Backlink Profile | Human | Cited by authoritative sites (.edu, .gov, leaders) |
| Media Mentions | Human | "Featured in" with media logos |
| Industry Awards | Human | Displays relevant industry awards or recognition |
| Publishing Record | Human | Conference talks, publications, patents |
| Brand Recognition | Agent | Brand has search volume |
| Social Proof | Human | Authentic user testimonials with real details |
| Knowledge Graph Presence | Agent | Has Wikipedia entry or Google Knowledge Panel |
| Entity Consistency | Agent | Brand/author info consistent across the web |
| Partnership Signals | Human | Shows partnerships with authoritative organizations |
| Community Standing | Human | Active and influential in professional communities |
References: Google: E-E-A-T in the Search Quality Rater Guidelines · Google Search Quality Rater Guidelines (PDF)
Trust (human experience, SEO)
| Check | Experience | What good looks like |
|---|---|---|
| Legal Compliance | Human | Privacy Policy + Terms of Service present |
| Contact Transparency | Human | Physical address or ≥2 contact methods |
| Security Standards | Human | Site-wide HTTPS, no security warnings |
| Disclosure Statements | Agent | Affiliate links disclosed (veto if missing) |
| Editorial Policy | Human | Content standards and review process published |
| Correction & Update Policy | Agent | Has corrections page or changelog |
| Ad Experience | Human | Ads <30% of page; no intrusive popups |
| Risk Disclaimers | Agent | YMYL topics have necessary disclaimers |
| Review Authenticity | Agent | Reviews show authenticity signals |
| Customer Support | Human | Clear return policy, complaint channels, response SLA |
References: Google Search Quality Rater Guidelines (PDF)
Further reading: how AI systems read your pages
Primary sources on how AI crawlers fetch and select content, and how generative engines choose what to cite:
Where the score fits
CORE-EEAT is one half of what a SiftServe audit reports; the other half is the capture stats (page weight, tokens, coverage), measured by their own published pipeline. Together they answer two questions: how much of a page an agent can read, and how well the page scores once read. If you want both run on your own site, before and after sifting, become a research partner.