Lesson 0008 Β· The quality signals behind every ranking
E-E-A-T
Google's name for "is this content trustworthy?" It's not a score, not a tag, not a ranking factor β and that's exactly why a builder can win here: turn the fuzzy guideline into signals a machine can check.
Recap from Lesson 0007: that got your page found. But found β believed. Every prior lesson made the page eligible and discoverable; this one is about whether engines (and AI answer engines) have any reason to trust what it says.
E-E-A-T = Experience, Expertise, Authoritativeness, Trust β the framework Googleβs quality raters use to judge content. Google is blunt about its centre: βOf these aspects, trust is most important. The others contribute to trust.β[1] And about the trap most people fall into next.
The myth, and the builderβs opening
There is no βE-E-A-T score,β no E-E-A-T meta tag, and it is not a direct ranking factor. Straight from Google:
βWhile E-E-A-T itself isnβt a specific ranking factor, using a mix of factors that can identify content with good E-E-A-T is useful.β[1]
Read that as a spec. Googleβs systems approximate E-E-A-T with a mix of factors β and many of those factors are concrete, machine-readable things on your page. You canβt compute the score, but you can audit the mix. Thatβs the tool.
eeat_audit.py on a page and get a checklist of the detectable trust signals β byline, authorβs linked identity, publish/updated dates, publisher, outbound sourcing, transparency links β grouped by the E-E-A-T letter each supports. Explicitly not a score; a gap-finder. Four letters, one centre
The higher the stakes, the higher the bar: Google holds YMYL pages β βYour Money or Your Life,β anything affecting health, finances, safety, or societal well-being β to far stricter E-E-A-T.[2]
From fuzzy guideline β detectable signal
This is the whole builder move: each abstract quality maps to something in your HTML or JSON-LD that a crawler can confirm in milliseconds.
| Quality | Machine-detectable proxy | Lives in |
|---|---|---|
| EΒ·E authorship | named author + author entity | meta[author], rel=author, Article author |
| A authority | author identity linked off-site | Person.sameAs β real profile |
| A publisher | org behind the content | Article publisher |
| T transparency | publish + updated dates | datePublished, dateModified |
| T accountability | about / contact reachable | site nav / footer links |
| T sourcing | cites outside references | outbound <a> to sources |
That last row is also an AEO lever: the GEO research found that adding citations, quotations, and statistics measurably lifts how often AI engines cite a page.[3] Trust signals and AI-citation signals are the same signals.
The tool: audit the mix
# NOT a score. Each row is one factor in Google's "mix" β present or missing.
author_named = meta[author] or rel=author or Article.author
author_linked = Person.sameAs -> an authoritative profile # authority
trust = datePublished & dateModified & publisher
& cites_outbound_sources & about/contact & https # the centre- Self-check (offline):
python3 tools/eeat_audit.py --demo - Audit a real article (yours or a competitorβs):
python3 tools/eeat_audit.py https://site.com/blog/post - Fix one MISS with the highest leverage β usually
author+sameAs+dateModifiedin your Article JSON-LD (you already have the schema tool from Lesson 0003).
$ python3 tools/eeat_audit.py https://thin-blog.example/post E-E-A-T proxy-signal audit (not a score β the detectable factors) ββββββββββββββββββββββββββββββββββββββββββββββ [FAIL] EΒ·E Β· named author (byline / meta / schema) [FAIL] EΒ·E Β· author detailed in Article schema [FAIL] A Β· author identity linked off-site (sameAs) [FAIL] A Β· publisher declared in schema [FAIL] T Β· served over HTTPS [FAIL] T Β· publish date present (datePublished) [PASS] T Β· cites outbound sources (2 external links) [FAIL] T Β· transparency link (about / contact) ββββββββββββββββββββββββββββββββββββββββββββββ VERDICT: 1/9 trust signals present. Anonymous / thin β engines have little to trust.
sameAs pointing at a profile that doesnβt establish any real expertise fools no one β Authoritativeness is mostly earned off your page (who links to and cites the author). The audit finds the missing markup so engines can read the trust youβve actually built; it canβt manufacture trust you havenβt. And on YMYL topics, faking it is actively risky.[2] Add the signals because theyβre true, then make them machine-readable. Ceiling to know: eeat_audit.py checks presence of signals, not their quality β it confirms an author is named, not that the author is genuinely expert. Itβs a completeness gate on the parts a machine can verify, deliberately silent on the parts only a human rater can judge.
Retrieval practice Β· no peeking
Earn the trust
Answer from memory β that effort is what makes it stick. One try each; pick before you read the others.
Google's own E-E-A-T explainer: the definition, "trust is most important," and the "not a specific ranking factor / mix of factors" line. Go deeper with the <a href="https://services.google.com/fh/files/misc/hsw-sqrg.pdf">Search Quality Rater Guidelines (PDF)</a> for E-E-A-T + YMYL, and the GEO sourcing finding in <a href="https://arxiv.org/abs/2311.09735">arXiv:2311.09735</a>.