83 Questions, Most Built From Google's Own Advice — Fifty-four Land on One Dealer Page

One consultant just republished a free set of 83 questions, most of them drawn from Google's own public advice. Aleyda Solis announced the revised checklist on November 2, 2026. Google didn't write it and nothing suggests Google reviewed it. But for a dealer with a website, the questions are the useful part.

83 Questions, 10 Areas, and One Big Asterisk

On November 2, 2026, Aleyda Solis put out a new edition of a checklist that tests web pages against Google's public quality advice. Solis is a consultant in SEO and AI search and founded the firm Orainti. She also shipped a Claude Skill meant to help run the review.

The sheet carries an October 29, 2026 date. It has 83 questions across 10 areas; 76 of them point to 11 Google documents.

One thing to settle first: Google did not make this sheet, and there is no sign in her post or the sheet that Google reviewed it.

Every row is a question, not a rule. The main tab gives you 10 columns. They cover the topic area, the question to ask, when it applies, a way to test it, the wording from Google, where that wording came from, a box for your assessment, room to paste evidence or links or explain why a row does not apply, a mark against the top-ranking page, and what the check is reviewing. The assessment menu starts at Not assessed and ends at Not applicable. In between are Meets, Partly meets and Does not meet. All 83 rows start as Not assessed. The sheet carries no numeric score, and the areas are not weighted against each other.

Trust and authorship has the biggest group with 13 rows. Editorial validation has the smallest with two.

Where each check applies matters more than the total. Fifty-four apply to one page. Twelve apply across the site. Nine apply to a page template, and eight apply to the production workflow. Page experience owns every one of the nine template checks. Only 22 rows cover all content with no ifs. The rest are conditional. Some get very narrow: bought or reused expired domains, partner or white-label sections, or AI-made product data and images sent to Merchant Center.

The source column puts seventy-one under Google Search guidance, five under the rater guidelines, and seven under Solis's own practical advice. Over a third of the sheet traces back to one document: the helpful content page.

Then there is the new piece. In her post on X, she says the update includes a Claude Skill meant to help run the review. The header shows version 1.2.1 and the date October 29, 2026. The download contains 27 files. Its rules require a note near the top of each report: not a rating from Google, not a score for E-E-A-T, not a certification, and not a prediction about rankings.

That raises a fair question for anyone who owns a store with a website: if it doesn't produce a score, why would a dealer bother? The answer sits in what Google changed in recent weeks.

Google Kept Moving the Ground This Year

The update tracks changes in Google's own documents. Google told sites on October 1, 2026 to check AI-made content by hand before it goes live. That review also covers page titles, meta descriptions, structured data and alt text on images.

A day later, Google added new sections to its helpful-content page covering main content, effort, skill and accuracy. It also added a passage about fake author profiles. Each of those additions shows up as a row in her sheet.

It goes beyond page advice. In a May 15, 2026 clarification, Google said its spam rules cover AI answers in Search, AI Overviews and AI Mode included.

Enforcement has been steady. Four spam updates began in 2026 ahead of the sheet's date: March 24, then June 24, then August 18 and September 24. The last one ran 13 days and 16 hours and ended on October 8.

People who watch search already rank these signals high. In a survey of 131 SEOs from September 2026, 47.6% named content quality as a top-three ranking factor, and 36.5% said the same of authority and trust. That is a survey of SEOs, not of Google's algorithm. Still, it shows where the people who track this put their money.

None of which explains what a page-experience row has to do with a guy selling travel trailers. That's next.

What It Means If You Sell Powersports or RVs

My read: the sheet is not a ranking formula, and the skill has to say so. But it's the closest thing to a free second opinion on the pages your customers read before they call.

Picture a powersports store with a page on one used sport side-by-side — the one with the cage, the tires and the tune sheet. The trust rows want proof, not adjectives. They ask for original photos, records or notes behind any first-hand claim, author links that show a relevant background, cited sources for the exact claim, and a log of corrected errors.

My read: that is where most unit pages fall apart. Factory stock photography and a spec table lifted from the manufacturer don't show anybody's first-hand experience. Photos taken in your service bay do. Ten minutes with a phone in the shop produces more of this than a week of writing.

Now picture an RV dealer with a wall of comparison pages. The sheet pulls 11 rows from Google's guide on writing high-quality reviews. They read less like a technical SEO audit and more like a style guide for editors. For a performance claim, the how-to column wants a table showing what was measured, under what conditions, and on how many samples. The example given is repeating a battery test under identical settings. For a ranked best list, every item needs a reason it is there, plus clear strengths and limits, so the page works without the linked reviews.

Prices are the other quiet one. Two maintenance rows put one person in charge of prices, availability, specs and guidance, plus a trigger for review. They also give readers a way to report errors, with a record of corrections. If your lot re-prices units weekly and your site re-prices them quarterly, that row is a full afternoon.

Then speed, where the sheet stops being polite. Page experience is the one section with hard numbers. The cutoffs track Google's Core Web Vitals guidance, measured on the 75th percentile of real-user data. Real-user data is required. If there is too little of it, the check gets marked Not assessed. The rows cover HTTPS, whether the page works on a real phone, ads that get in the way after load and scroll, pop-ups on entry and return visits, and how easily a visitor can tell main content from ads and recommendations.

My read: if you only work two groups, work the 13 trust rows and the page-experience rows.

My read: for a dealership, those rows are the ones that cost cash. Not because Google is watching, but because the shopper on your lot at 7 p.m. is.

How to Use It This Week

Seven steps, in order. None of them require an agency retainer.

1. Pick your money page. Not the homepage — the one page per store that produces the most calls. Print it, then open it on an actual phone on cell data, not the office Wi-Fi.
2. Go get the sheet. Solis announced the update on X; the checklist lives in a Google Sheets document, and the skill downloads from her website. I'm not handing you a link I can't vouch for, so search her name plus the words content quality checklist.
3. Start with the 13 trust rows. Answer each one with a link, a record, or a plain reason it doesn't apply. The evidence field is the entire point — a debate about quality becomes something you can document and hand to a writer.
4. Do not chase a score. The sheet gives you no score and does not weight one area against another. You're building a fix list, not a grade.
5. Put a name on the price rows. The maintenance rows ask you to name an owner and set a review trigger. Type a person's name into the cell, not a department.
6. Test the phone before you touch the copy. Watch whether an overlay blocks the page on the way in, and whether an ad shoves the main content down after load. Take screenshots with a timestamp.
7. Know what the AI piece is before you run it. A Claude Skill is a packaged set of instructions that Anthropic's assistant loads to handle one task. Inside the download are the rules file, a second copy of the sheet, a data file, presets and a folder of sample cases. The write-up leaves open how it handles evidence it can't see — your field speed data, your Search Console reports, your workflow records — and seventeen checks sit at the template or workflow level, where that evidence usually lives.

If you want to move now, here are three scripts. Hand them to whoever writes your pages, or paste them into whatever assistant you already use.

• For a used unit page: Here are 10 photos from a used side-by-side and my inspection notes. Write 200 words on what the photos and notes actually show. Flag every claim I can't back with a photo or a record.
• For a new model page: I'm pasting a factory spec sheet. Draft the page, then tell me which line of the sheet supports each spec claim. Do not add a spec that isn't in the sheet.
• For what's already live: Read this page and list every claim a customer could check — price, availability, specs, dates. Give me a table with the claim, where it appears, and who should own it.

Mistakes to Avoid

• Don't buy a score. No numeric grade appears anywhere in the spreadsheet, and no area carries more weight than another, while the skill must state that what it produces is not a rating from Google or a forecast of rankings. Anyone offering you a letter grade is offering a different product.