Is E-E-A-T a ranking factor?

Google's documentation, "Creating helpful, reliable, people-first content", says E-E-A-T is not itself a specific ranking factor. It is how quality raters describe a page worth trusting. seodraft treats it as an editorial standard instead: three first-hand pieces per article, each traceable in the finished text.

Why did my draft fail with evidence-unaccepted?

The plan cites a piece that is still pending or one you discarded. Open the evidence bank in Profile, accept or correct it, and run the rules again. A piece the bank has never seen does not block that run: the gate registers it as a proposal so it reaches your tray.

Can my agent add evidence on its own?

It can, and what it adds arrives pending. Anything you said in the conversation, or typed into Profile yourself, is accepted as it lands. That split is the whole reason the bank is a table of rows with states.

Feature · Write

Every article cites three first-hand facts you accepted.

seodraft keeps an evidence bank: numbers you measured, cases you can name, mistakes you made and what they cost. Your agent plans each article against that bank, and a draft that leans on a piece you never accepted fails the rules. The card below is those rules as they run in seodraft's code today.

Create account Create account

7 days free, no card needed. Then $5/month if you subscribe.

What is E-E-A-T?

E-E-A-T stands for Experience, Expertise, Authoritativeness and Trustworthiness. It is the vocabulary Google's quality raters use to describe a page worth trusting, and Google's own documentation, "Creating helpful, reliable, people-first content", states that E-E-A-T is not itself a specific ranking factor.

The first E is the part a tool can act on. Experience is what only your business can report: a figure you measured, a client you can name, a mistake you made. seodraft stores those pieces as rows with a state, and the rules refuse a draft that cannot point at three of them.

What the bank holds

Six kinds of piece: data, example, quote, experience, capture and source: one vocabulary for the bank and for the article plan.
Who put it there: What you said is accepted the moment it lands; what the agent inferred, measured or fetched waits for you in Profile.
Three per article: A plan carrying fewer than three pieces fails as evidence-min, and the skill tells the agent to ask you instead of inventing one.
Traceable in the finished text: A piece longer than 25 characters has to appear in the body, the summary or a FAQ, or the draft fails as evidence-unused.
Set aside, still on file: A piece you discard keeps its row, so the rules can still recognise it in a later plan and refuse it.
A review queue sorted by doubt: A classifier scores how first-hand each proposal reads and puts the weakest ones first. It marks a row and never changes its state.

The rules that read the bank

You said it, the agent typed it: accepted on arrival
The agent inferred, measured or fetched it: pending until you rule
Pieces per article: at least 3
A pending or discarded piece in the plan: evidence-unaccepted, blocks
A piece the bank has never seen: registered as a proposal, reaches your tray

Kinds a piece can be · 6

data
example
quote
experience
capture
source

Rules that read the bank

evidence-min: the plan carries fewer than three pieces.
evidence-unaccepted: the plan leans on a piece that is pending or discarded.
evidence-unused: a piece never reaches the body, the summary or a FAQ.
unsourced-numbers: advisory, the figures in the draft read as nobody's.

The evidence rules as they stand in seodraft's own code on September 24, 2026: which state a piece gets, and how the rules count and trace it. The other editorial rules, which say nothing about evidence, were left out.

How a fact becomes a paragraph

A piece of evidence travels from your conversation to a finished article in three steps.

1
The agent writes it down. It stores what you said as accepted, and anything it inferred, measured or fetched as pending.
2
You accept it, correct it or set it aside. In Profile, an edit is an approval with a correction: your wording is what gets accepted.
3
The rules read the plan against the bank. Three pieces, none of them waiting on you, each one traceable in the article you are about to approve.

What it costs

The evidence bank uses no paid data. The rows are yours, and the three rules that read them run with the rest of the checks before you approve.

The classifier that sorts the review queue and flags unsourced numbers costs around USD 0.0001 for a full article with every check running. It advises and never changes a verdict, so an article that passed yesterday passes today.

Questions about E-E-A-T evidence

Short answers on what Google says, why a draft gets blocked and what your agent can add by itself.

Is E-E-A-T a ranking factor?: Google's documentation, "Creating helpful, reliable, people-first content", says E-E-A-T is not itself a specific ranking factor. It is how quality raters describe a page worth trusting. seodraft treats it as an editorial standard instead: three first-hand pieces per article, each traceable in the finished text.
Why did my draft fail with evidence-unaccepted?: The plan cites a piece that is still pending or one you discarded. Open the evidence bank in Profile, accept or correct it, and run the rules again. A piece the bank has never seen does not block that run: the gate registers it as a proposal so it reaches your tray.
Can my agent add evidence on its own?: It can, and what it adds arrives pending. Anything you said in the conversation, or typed into Profile yourself, is accepted as it lands. That split is the whole reason the bank is a table of rows with states.

Editor · Where you read the draft and the rules verdict before approving.
Content brief · What the agent reads before it writes.
AI slop checker · The same prose rules, run free over any text you paste.