Aleyda Solis turns Google's content quality guidance into 83 checks

On November 2, 2026, Aleyda Solis, the SEO and AI search consultant who founded Orainti, published a revised version of her checklist for judging web pages against Google's public quality guidance and paired it with a Claude Skill built to run the assessment. The spreadsheet, dated October 29, 2026, holds 83 checks in 10 areas and links 76 of them to 11 Google documents.

According to Solis's post on X, she "Updated the Google's Content Quality [EEAT], Helpfulness & Page Experience Checklist with the latest criteria, made the guidance more comprehensive and added a Claude Skill to help with the assessment." The checklist is a practitioner's tool. Google did not write it, and nothing in the post or the spreadsheet suggests Google reviewed it. The skill's own instructions make the same distinction: every report it produces must say, near the top, that it is not a Google rating, a certification or a prediction of rankings.

In Short

An SEO consultant named Aleyda Solis has updated a free spreadsheet that turns Google's public advice about useful, trustworthy web pages into 83 questions. It matters to anyone who publishes content and wants to compare their pages with what Google says it values, from honest author bios to pages that load quickly. The new version comes with an add-on for Anthropic's Claude that walks through the questions and writes up the evidence, and it is told to say plainly that its report is not a Google score.

What the update says, and what it leaves out

The post makes three claims: the criteria are current, the guidance is more comprehensive, and a Claude Skill now accompanies the sheet. It does not list which checks are new, which were rewritten or which were dropped. The spreadsheet has no change log either. Its only version marker is a banner, repeated at the top and bottom of the main tab, reading "Checklist Updated on October 29th 2026 / Created by Aleyda Solis / SEO & AI Search Optimization Consultant." A line-by-line comparison with the previous edition is therefore not possible from the published material.

What the sheet does show is how recent its sources are. Several checks rest on changes Google made to its documentation in the five weeks before the update date. On October 1, 2026, Google told sites to manually fact-check all AI-generated content before publishing and extended that review to titles, meta descriptions, structured data and image alt text. One day later, Google added sections on main content, effort, talent or skill and accuracy to its "Creating helpful, reliable, people-first content" page, along with a passage on fake author profiles that PPC Land reported on October 7. Each of those additions has a matching row in Solis's checklist.

Ten areas, 83 questions

Every check is phrased as a question. The main tab has 10 columns: the area, the actionable check, when it applies, how to check it, the source wording or guidance, the primary source, an assessment field, a field for evidence, URLs or a reason the check does not apply, a benchmark against the top-ranking result, and the review scope.

Trust and authorship is the largest group, at 13 rows. Editorial validation is the smallest, with two.

The assessment column carries a drop-down with five values: Not assessed, Meets, Partly meets, Does not meet and Not applicable. All 83 rows ship set to "Not assessed." There is no numeric score anywhere in the sheet, and no weighting between areas.

The review-scope column says what unit each check examines. Fifty-four checks apply to an individual page, 12 to the whole site, nine to a page template and eight to the production workflow. All nine template checks sit in page experience, which the sheet treats as a property of shared templates rather than of individual articles.

Only 22 rows apply to "All content" without qualification. The rest are conditional. Some apply to all sites or all factual content, while others narrow sharply: "Acquired or repurposed expired domains," "Partner, white-label or other third-party sections," or "AI-generated product data or images submitted to Merchant Center." Two rows are reserved for health, finance, safety or societal welfare topics, which Google calls YMYL.

Where each check comes from

The primary-source column splits the 83 checks three ways. Seventy-one are labelled Google Search guidance, five cite Google's rater guidelines and seven are marked "Practical recommendation," meaning they come from Solis rather than from Google.

The 76 sourced checks point to 11 distinct Google URLs:

• "Creating helpful, reliable, people-first content": 31 checks
• "Write high quality reviews": 11
• Google's spam policies: 10
• Search Quality Evaluator Guidelines: 5
• The page experience page: 5
• Google's AI optimization guide for Search: 4
• The guide on using generative AI content: 3
• The Core Web Vitals page: 3
• The May 2025 Search Central blog post on succeeding in AI search: 2
• The February 2023 blog post on Google Search and AI content: 1
• The documentation page on AI features: 1

More than a third of the checklist, then, traces to one document: the helpful content page.

A second column records how closely each row follows Google's text, and here the sheet is unusually explicit. Twenty-six rows are labelled "Checklist question adapted from Google guidance." Twenty-nine are "Guidance summary," 10 are "Policy summary" and five are "Rater guidance summary." Only six quote Google directly, three as "Google question" and three as "Google excerpt." The seven practical recommendations say so in the source column too, sometimes with a disclaimer attached. The row on disclosing commercial relationships, for instance, ends: "This is an editorial transparency check, not a quoted Google requirement."

The helpful content questions

The 31 rows drawn from the helpful content page cover purpose, originality and trust. Google's original self-assessment questions appear in adapted form: whether the content gives "original information, reporting, research, or analysis," whether it is "mainly summarizing what others have to say without adding much value," and whether it was mass-produced or outsourced so widely that individual pages "don't get as much attention or care."

The checklist rewrites many of these as instructions for a reviewer. Google's question about writing to a preferred word count becomes "Is the depth determined by the user's task rather than a supposed Google word count?", with the note "Choose length according to the task. Remove padding."

Three rows quote Google's questions verbatim: "Does the content have any easily-verified factual errors?", "Is this content written or reviewed by an expert or enthusiast who demonstrably knows the topic well?" and "Does the content provide a substantial, complete, or comprehensive description of the topic?" For the second, the how-to column adds that "A reviewer's name alone is insufficient."

The October 2 additions appear as separate rows. One asks whether the main content shows "meaningful human effort or well-built supporting systems," with the reviewer told to "Assess the useful result, not an arbitrary number of hours." Another asks whether the content shows the skill needed for its purpose, noting that "Specialist credentials are only relevant when the topic requires them." A third applies Google's accuracy point to high-stakes topics. The fourth quotes Google's new wording on authorship directly: "Fabricating creator profiles" is "a form of deception."

This is the territory Google labels E-E-A-T, for experience, expertise, authoritativeness and trust. The checklist puts the acronym in its title, but none of the 83 rows treats it as a score. The trust rows ask for evidence instead: original photos, records or observations behind claims of first-hand experience, author links that show relevant background, cited sources that support "the exact claim," and corrected errors.

Spam policies

Ten rows map to Google's spam policies, one per policy area: scaled content abuse, scraping, site reputation abuse, doorway abuse, misleading functionality, link spam, manipulation of generative responses, expired domain abuse, thin affiliation and user-generated spam.

Some rows add precision that Google's policy names leave out. The link spam row notes that "a disclosure alone does not qualify a link," which means paid links need rel="sponsored" or rel="nofollow" in the rendered page. The site reputation row says "Third-party authorship alone is insufficient; regional treatment differs," pointing the reviewer to the policy's regional provisions. That regional split is recent. Since August 30, 2026, Google no longer applies site reputation abuse manual actions to searches made inside the European Economic Area, a change it said followed discussion with the European Commission.

The generative-response row is careful about what Google has and has not said. On May 15, 2026, Google clarified that its spam policies apply to generative AI responses in Search, including AI Overviews and AI Mode. The checklist turns that into a check for "deliberate attempts to manipulate Google's generative AI responses," then adds that "it is not a separately defined prompt-injection policy." The rule comes from the introduction to Google's spam policies, not from a policy of its own.

Enforcement under those policies has been regular. Google started four spam updates in 2026 before the checklist's update date: on March 24, June 24, August 18 and September 24. The last of them ran for 13 days and 16 hours before finishing on October 8.

Rater guidelines

Five rows cite the search quality rater guidelines, each pinned to the September 2025 edition and to specific sections. Sections 12.1 to 12.4 cover the searcher's goal, language, region and location. Section 5.2.2 covers filler that obstructs main content. Sections 3.3 to 3.3.5 cover independent reputation research, and the row adds a caution: "No reputation information alone does not establish low quality." Section 3.4.1 separates safe personal experience from high-stakes advice that needs qualified expertise. Sections 2.5.2 to 2.5.3 cover contact and customer service information, with "stronger expectations for transactions."

Google revises the guidelines regularly. The January 2025 edition, for example, added 11 pages and took the document from 170 to 181 pages, and introduced criteria for evaluating AI-generated content. The checklist cites none of the earlier editions.

Reviews and comparisons

The 11 review rows come from Google's guide to writing high-quality reviews, and they read more like an editorial standard than an SEO audit. Two quote Google directly: "Evaluate from a user's perspective" and "Explain what sets something apart from its competitors." The how-to column gives concrete examples. For performance claims it asks for "a results table with units, conditions and sample size," for instance by repeating battery tests under the same settings. For ranked "best" lists, every entry should explain why it was included, with distinguishing strengths and important limitations, so the list stands on its own without the linked reviews.

The checklist also marks where Google's review guidance stops short of a rule. Linking to several sellers, it says, is "a recommendation, not a universal requirement."

AI production and AI search

The five AI and production rows avoid treating AI use as a fault in itself. "AI use alone is neither a failure nor a quality guarantee," the first row says. The disclosure row adds that "A blanket label is not always required," and asks for an explanation of what was automated, how and why when readers would reasonably expect one.

The next two rows follow Google's October 1 edit to its generative AI content guide. One requires that a person check AI-generated material before publication and keep "reviewer evidence." The other extends that review to generated titles, meta descriptions, alt text and structured data. The last row covers a narrower case. For Merchant Center submissions, AI-generated images need IPTC DigitalSourceType metadata set to TrainedAlgorithmicMedia, and AI-generated titles and descriptions need separate AI-labelled attributes. The checklist describes this as "metadata, not a universal visible-image label," and calls the check "conditional, not a rule for every website image."

The eight Search presentation rows deal with Google's AI features directly, and several rule things out. One states that Google-Extended "governs certain other AI uses; it does not control inclusion or ranking in Google Search," and tells reviewers to inspect Googlebot access, indexability and snippet controls instead. Snippet controls have covered Google's AI features since at least March 2025, when Google added AI Mode to its robots meta tag documentation and said nosnippet keeps content out of AI Overviews and AI Mode.

Another row, citing the AI optimization guide Google published on May 15, 2026, says: "Google Search requires neither artificial chunk sizes nor special AI markup or llms.txt." Google's Chrome team treats the file differently. On May 5, 2026 it added llms.txt to Lighthouse in a new section of agentic browsing audits, though Lighthouse marks a missing file as not applicable. The checklist addresses Google Search only.

A further row covers what it calls manufactured brand mentions. It tells reviewers to check outreach and agency briefs for "fake endorsements or forum conversations" and summarises Google's mythbusting section: such mentions "are not a dependable optimization tactic." Two of Solis's own practical recommendations sit beside it. One keeps "product, market, date, units and qualifications near the claim" so excerpts stay accurate, while warning against implying "guaranteed AI citation benefits." The other reconciles prices, restrictions and availability across article, product page, offer terms and checkout.

The checklist's final row brings in measurement. It points reviewers to Search Console's Generative AI performance report, which Google added to Search Console in June 2026, and asks them to combine it with "relevant conversions and task outcomes." The source column marks that combination as the checklist's own recommendation, "not a quoted reporting requirement."

Page experience, in numbers

The page experience section is the only one with numeric thresholds. They follow Google's Core Web Vitals guidance, all measured at the 75th percentile of field data:

• Largest Contentful Paint: 2.5 seconds or less
• Interaction to Next Paint: 200 milliseconds or less
• Cumulative Layout Shift: 0.1 or less

The checklist insists on real-user data. Insufficient field data means a check is marked "Not assessed," and the INP row warns that "Lighthouse TBT is not a direct INP measurement." The layout-stability row asks for mobile and desktop to be judged separately. The other six rows cover HTTPS, usability on "a real phone," ad interference after loading and scrolling, intrusive overlays on entry and return visits, how easily the main content can be told apart from ads and recommendations, and whether the key journey works by keyboard and assistive technology. That last check is labelled "a practical usability check, not a complete accessibility audit."

Maintenance and validation

Four maintenance rows deal with freshness. Displayed update dates should "reflect substantive changes," and the how-to column says plainly: "do not refresh the date alone." Mass changes made only to make a site look fresh are discouraged, citing Google's helpful content page. Two practical recommendations assign an owner and a review trigger for prices, availability, specifications and guidance, and require a working route for readers to report errors, along with a correction record.

The two editorial validation rows close the sheet. One adapts Google's suggestion to seek "an honest assessment from trusted people unaffiliated with the site," and notes that doing this before publication "is a recommended workflow, not a timing requirement set by Google." The other is the Search Console measurement row described above.

The Claude Skill

A second tab, titled "EEAT Claude Skill," holds a single line: a download link to the skill, hosted on Solis's website. Claude Skills are packaged instruction sets that Anthropic's assistant can load for a specific task, and software vendors have begun building skills for their own products. In August, for example, Salesforce put 37 prebuilt sales skills inside Claude for pilot customers.

In the Claude interface Solis shared alongside the checklist, the skill is named eeat-helpfulness-assessment. Its header gives the version as 1.2.1, dated October 29, 2026, and describes it as "Based on Google's Content Quality, E-E-A-T, Helpfulness & Page Experience Checklist by Aleyda Solis." The package holds 27 files. Besides the main instruction file, they include a copy of the checklist, a checklist data file, a presets file and a folder of worked examples. Two examples are labelled fictional, an evidence packet and a review audit, and two more are "focused zero-findings" cases. The package also contains a configuration file for OpenAI agents, which neither the post nor the checklist explains.

The instructions set out the job in a few sentences: "Turn the checklist into an evidence-backed assessment and a prioritized improvement plan. Lead with evidenced findings and practical improvements. Zero problems is a valid outcome." The skill is told to keep "material uncertainty beside the affected conclusion" and to consolidate any remaining evidence requests. It must also say once, near the top of the report, that the output is "an expert assessment of the inspected evidence, not Google's rating, an E-E-A-T score, a certification or a prediction of rankings or AI citations."

The explicit allowance for zero findings, and two worked examples built around it, suggest the package is designed not to manufacture problems. Neither the post nor the checklist describes how the skill handles evidence it cannot see, such as field Core Web Vitals data, Search Console reports or editorial workflow records. Seventeen checks are scoped to templates or workflows, where that kind of evidence usually sits outside the page itself.

What the checklist does not settle

The sheet also has some loose ends. One how-to note on content depth tells reviewers to "Assess completeness separately in CQ12," but the spreadsheet has no identifier column. CQ12 matches the twelfth content quality row, the completeness check, which suggests a numbering scheme kept elsewhere, possibly in the skill's data file.

The bigger limit is the one the skill states itself. Google's documentation sets out what the company says it values, and the rater guidelines describe how human evaluators judge results. Neither is a ranking formula, and the checklist does not claim otherwise. The seven practical recommendations, from accessibility testing to correction logs, go beyond what Google's documents ask for, and the sheet labels each one.

Why this matters for marketers

For marketing teams, the checklist's value lies less in any single question than in the evidence column. Each check asks for a URL, a record or a reason it does not apply, which turns a debate about quality into something that can be documented and handed between SEO, editorial and engineering teams.

Practitioners already rank these signals highly. A September 2026 survey of 131 SEOs found content quality in the top three ranking factors for 47.6% of respondents and authority and trust for 36.5%. Google's own messaging in 2026 has pushed in the same direction. In June, Brendon Kraham, Google's vice president of Search and Commerce for Global Ads Solutions, argued that good SEO is good GEO and listed llms.txt and content chunking among the tactics Google Search does not need. John Mueller made a sharper version of the point in August 2025, when he said that the urgency with which new AI SEO acronyms are pushed signals spam.

The checklist follows the same line. Its AI rows mostly explain what Google says is not required, rather than adding new tactics. For affiliate publishers, retailers running Merchant Center feeds and sites that carry advertising, the commercial rows are the most concrete. They cover paid-link attributes, thin affiliation, ad interference with main content, AI image metadata and price consistency between article and checkout. All but the last rest on written Google documentation; price consistency is one of Solis's own recommendations.

What the update cannot show is how much any of it moves rankings or AI citations. The skill's required disclaimer says as much.

Timeline

• February 2023: Google publishes guidance on AI-generated content, saying automation used mainly to manipulate rankings violates its spam policies
• May 7, 2024: Google says it is enforcing its site reputation abuse policy through manual reviews
• January 23, 2025: Google revises its rater guidelines from 170 to 181 pages
• March 5, 2025: Google adds AI Mode to its robots meta tag documentation
• August 14, 2025: John Mueller warns that pushing AI SEO acronyms signals spam
• September 2025: Google publishes the edition of its rater guidelines that the checklist cites
• March 24, 2026: Google starts its first spam update of 2026
• May 5, 2026: Google adds llms.txt to Chrome Lighthouse audits
• May 15, 2026: Google publishes its guide to optimising for generative AI features in Search
• May 15, 2026: Google clarifies that its spam policies apply to generative AI responses
• June 2026: Search Console gains Generative AI performance reports
• June 2026: Google's Brendon Kraham argues that good SEO is good GEO
• June 24, 2026: Google starts its second spam update of 2026
• August 18, 2026: Google starts its third spam update of 2026, across all languages
• August 26, 2026: Salesforce and Anthropic announce 37 prebuilt sales skills inside Claude
• August 30, 2026: Site reputation abuse manual actions stop applying to searches inside the European Economic Area
• September 9, 2026: Survey of 131 SEOs puts content quality among the top ranking factors
• September 24, 2026: Google starts its fourth spam update of 2026
• October 1, 2026: Google tells sites to manually fact-check AI-generated content before publishing
• October 2, 2026: Google adds main content, effort, skill and accuracy sections and a fake author warning to its helpful content guidance
• October 8, 2026: The September 2026 spam update finishes after 13 days and 16 hours
• October 29, 2026: Solis's checklist is updated, and the Claude Skill reaches version 1.2.1
• November 2, 2026: Solis announces the updated checklist and Claude Skill on X

Related PPC Land coverage

• Google says fake author profiles signal a low-quality page - Google's October 2, 2026 additions to its helpful content guidance on main content and fabricated creator profiles.
• Google AI Overviews cut self-ranking listicle citations 38%, Lily Ray finds - Lily Ray's research on AI Overview citations, published alongside Google's helpful content additions.
• Google tells sites to manually factcheck all AI content before publishing - Google's October 1, 2026 edit to its generative AI content guide, extending review to titles, meta descriptions, structured data and alt text.
• Google's new guide for AI search: what SEO really needs now - The May 2026 Search Central guide to AI Overviews and AI Mode and its mythbusting section.
• Google spam policies now officially cover AI Overviews and AI Mode in Search - The May 15, 2026 clarification that spam policies apply to generative AI responses.
• Google finally gives Search Console its own generative AI visibility reports - The launch of Search Console reports for AI Overviews, AI Mode and Discover.
• Google adds llms.txt to Lighthouse as agentic web standards heat up - Chrome's new agentic browsing audits, which treat llms.txt as optional.
• Google's VP of Search tells CMOs: good SEO is still all you need for AI - Brendon Kraham's list of tactics Google Search does not need, including llms.txt and chunking.
• Google's third spam update of 2026 rolls out to every language - The August 18, 2026 spam update and the two that preceded it.
• Google's September spam update ends after 13 days as manual actions rise - The September 24 to October 8, 2026 spam update and anecdotal reports of more manual actions.
• Google drops EEA search penalties under site reputation policy August 30 - How site reputation abuse enforcement split by searcher location after talks with the European Commission.
• Google's updated search quality guidelines add 11 pages about spam policies - The January 2025 rater guidelines revision and its E-E-A-T changes.
• Google updates quality rater guidelines with AI content evaluation criteria - How the January 2025 rater guidelines handle generative AI content and filler.
• Google adds AI Mode to robots meta tag documentation - The March 2025 change that extended nosnippet to AI Overviews and AI Mode.
• Site Reputation Abuse: manual reviews currently underway - Google's May 2024 statement on how it enforces the site reputation abuse policy.
• Google releases guidance over AI-generated content - Google's February 2023 position on automation, E-E-A-T and spam.
• Google's John Mueller warns AI SEO acronyms signal spam tactics - Mueller's August 2025 comments on GEO, AIO and AEO.
• 131 SEOs rank relevance top Google factor, meta description dead last - The Zyppy Signal 2026 survey of practitioner views on ranking factors.
• Salesforce puts 37 prebuilt sales skills inside Claude for pilot customers - Salesforce's August 2026 skills package inside Anthropic's Claude.

Summary

Who: Aleyda Solis, an SEO and AI search consultant and the founder of Orainti, who also runs the SEOFOMO newsletter. The checklist is aimed at publishers, SEO teams, content editors and commercial sites that want to compare their pages with Google's published guidance.

What: An updated Google Sheets checklist of 83 questions in 10 areas. Seventy-six are linked to 11 Google documents, including the helpful content page, the spam policies, the reviews guide, the page experience and Core Web Vitals pages, the AI optimization guide and the September 2025 rater guidelines. The other seven are Solis's own practical recommendations. A downloadable Claude Skill, version 1.2.1, turns the checklist into an evidence-based assessment and improvement plan, and must state that its report is not a Google rating, an E-E-A-T score, a certification or a ranking prediction.

When: Solis announced the update on November 2, 2026. The checklist and the skill are both dated October 29, 2026.

Where: The announcement was made on X. The checklist is published as a Google Sheets document, and the skill is a download from Solis's website.

Why: Google changed several documents in 2026 that the checklist draws on. It published an AI optimization guide and extended its spam policies to AI responses in May, called for manual fact-checking of AI content on October 1 and added main content and fake author sections to its helpful content guidance on October 2. The checklist turns that guidance into checks that require evidence, and it labels which checks come from Google and which do not.