Skip to content

Documentation

E-E-A-T Analysis

Grade the Experience, Expertise, Authoritativeness, and Trust signals on a public page or a saved draft.

Open E-E-A-T Analysis

E-E-A-T Analysis, the PirateSERP page grader, scores the Experience, Expertise, Authoritativeness, and Trust signals that a page's HTML shows a reader. The graded report needs no API key. The grade is a PirateSERP assessment, not a Google score and not a ranking prediction.

Grade a page

  1. Switch to the project you want from the header.
  2. Enter a page address in From a URL, or choose a Content Editor draft in Or a saved draft.
  3. Click Grade.

Or a saved draft appears only when the project has saved Content Editor drafts. Save your Content Editor changes before you choose the draft, because a draft grade reads the saved content only. Grade the public URL after you publish to include the page-level evidence that a draft cannot carry.

Read the report

The report opens with an overall score out of 100, a letter grade, and one score for each of five sections. Start with Do these first, which lists the findings that move the grade most, then read the sections:

  • Page basics: the title, the description, the heading structure, and the text alternatives for images.
  • Experience: whether the page shows first-hand use of its subject.
  • Expertise: who wrote the page, what the byline says about that person, and whether a reviewer checked the work.
  • Authoritativeness: the sources the page cites and the identity information the site publishes about itself.
  • Trustworthiness: contact routes, policy pages, secure delivery, dates, disclosures, and the advertising experience.

Each finding carries a status, a severity where one applies, the detail behind it, and a recommendation where one applies. Use the grade to support your own content review. The grade does not predict rankings, traffic, or AI citations.

Understand the statuses

Unknown means the page supplies too little information to judge the check. E-E-A-T Analysis excludes unknown findings from the score, so they neither raise nor lower the grade. Read them anyway, because several name evidence that a reader expects to find.

Off-page reputation is reported as not measured. Backlinks, press coverage, and independent reviews sit outside the page, so the report says so instead of scoring their absence.

Word count and sentence length appear as information only and never change the score.

Hard gates

Two signals block a good grade on their own: hidden iframes and stacked scam-pattern urgency language. A page that carries either one shows a Blocked notice, drops into the F band whatever the five sections score, and lists the reason at the top of Do these first.

Fix the flagged problem and grade the page again. No other improvement lifts a blocked page.

YMYL topics

Your money or your life (YMYL) topics, such as health, finance, safety, and legal advice, are graded more strictly. The report shows a YMYL topic notice when it detects one, and findings that a general page can treat as opportunities count as issues instead.

A YMYL page cannot grade above C unless it carries all of the following:

  • a named author, not a generic byline such as "Admin" or "Staff."
  • stated credentials for that author, or a credited reviewer.
  • at least one citation to an authoritative source.

Run an AI review

The AI review is optional, and it never changes the score.

To use it, save an OpenRouter key and select a model in Account Settings. Read API keys for the setup steps. Grade a page, then click Run AI review at the end of the report. You can leave and reopen the audit while the review runs.

The review asks nine judgment questions that a signal check cannot answer, such as whether the page shows experience instead of claiming it, whether evidence sits next to the claims that need it, and whether the page acknowledges its limitations. Each question returns pass, partial, fail, or not evidenced, with a short quote from the page as its evidence.

Read the verdicts beside the graded findings. They record what a model found in the text you submitted, so check each quote before you act on it.

Rewrite a section

Some findings describe the article's own text, and those findings show Rewrite this section:

  • little or no first-hand experience described.
  • no duration of use stated.
  • no credentials or experience stated.
  • no subheadings.
  • unusual sentence length.
  • unedited AI boilerplate.
  • scam-pattern urgency language.
  • no outbound citations, or too few for the page's length.

Other findings, such as the author page, dates, structured data, contact and policy links, HTTPS, advertising, disclosures, and the title, description, H1, and image text, depend on the page template or the site rather than the article text. A rewrite cannot fix them, so they show no button.

The rewrite uses your OpenRouter key and model, the same as the AI review.

  1. Under a finding, choose the section to change in Choose a section.
  2. Optionally, add instructions of up to 500 characters, such as a fact the rewrite should use.
  3. Click Rewrite this section.

Each click applies one finding to one section. The Article card highlights the rewritten section; click Revert to restore the original text. Use Copy HTML or Download to take the edited article to your editor.

A rewrite never invents facts. When a finding needs something that only you know, such as how long you used a product or which study a figure comes from, the rewrite marks the spot with a bracketed prompt, for example [Add: how long you have used this product] or [Source needed: cite the study this figure comes from]. Replace every bracketed prompt with the real fact, or delete it, before you publish.

Re-evaluate

After you rewrite at least one section, click Re-evaluate in the Article card. Re-evaluate rescans the edited article instantly and at no cost, and the report shows the score before and after, for example was 32/100 (F) → now 41/100 (D). Findings about the page template or the site keep their original result until you publish the changes and grade the page again. Re-evaluate leaves an existing AI review unchanged.

An audit graded before section rewrites existed has no Article card. Grade the page again to rewrite its sections.

Fix and recheck

Make your changes in Content Editor or in your website editor, then grade the updated page. For author and reviewer findings, check the linked author page as well as the article.

Reuse results

Past audits lists the grades saved in the active project, each with its score and letter grade. Click an audit to reopen it and compare its findings with a later run, or click the delete control to remove it. A reopened audit still shows its AI review if you ran one.

An audit graded before the E-E-A-T update shows a notice to grade the page again for the current report.

Language support

Several text checks expect English, including bylines, credential wording, and disclosure wording. Non-English pages can return findings that need manual review.

Use the language-neutral findings, such as headings, dates, links, and structured data, alongside your own review. Read Languages for the full limits.