On September 9, 2026, Cyrus Shepard and Dawn Shepard published the 2026 Google Ranking Factors Expert Survey at Zyppy. One hundred and thirty-one practitioners scored 103 candidate factors on a seven-point scale, then answered one open question: name your three most important ranking factors. That produced 13,665 data points and a top ten that looks almost exactly like it did a decade ago — relevance first (57.1%), backlinks second (54.8%), content quality third (47.6%).
Most coverage led with that stability. It is only half the story. The survey did not measure AI citations. It says so itself, and it has announced a companion study on visibility inside Google's AI answers that had not shipped when the data was released. So if you have seen this survey framed as proof that "AI citations are now a ranking factor," you have seen a claim the source does not make.
Three things the survey does support, in its own words: user and click signals took a large leap this year, content quality is where 2026 diverges most from past editions, and links from places with real audiences still carry weight. Each of those is worth acting on separately, because they fail in different ways.
What the survey actually measured
Item | Detail |
|---|---|
Published | September 9, 2026, by Cyrus Shepard and Dawn Shepard at Zyppy |
Respondents | 131 SEO practitioners, invited from a list of 200+ |
Scale | 103 candidate factors, rated −3 (strongly negative) to +3 (strongly positive) |
Volume | 13,665 ratings; contributors logged 2,620 minutes in total |
Second question | Open-ended: the three most important ranking factors |
The top ten, ranked by how many experts put each factor in their personal top three:
Tier | Factor | Named in top three |
|---|---|---|
Top | Relevance | 57.1% |
Top | Backlinks | 54.8% |
Top | Content quality | 47.6% |
Middle | Authority and trust | 36.5% |
Middle | Behavior and click signals | 29.4% |
Middle | Brand signals | 27.0% |
Base | User satisfaction (task completion) | 19.8% |
Base | Technical SEO health | 17.5% |
Base | Topical authority | 14.3% |
Base | Internal links | 11.1% |
One factor topped the whole rating exercise: search intent match, defined as how well a page satisfies the type of result or task the searcher appears to want. One factor sank to the bottom: the meta description, which most respondents judge as having no ranking effect at all. Both results survived a room of people who argue for a living.
Three numbers worth reading more carefully
The percentages are nominations, not weights. A factor named by 57.1% of experts is not 57% of Google. It is the share of a room that agrees on a priority. Treat the table as a ranked list of where to spend attention, not as a budget.
Backlinks at number two is not a contradiction of Google. Gary Illyes has said that links are not among Google's top three signals, and plenty of people read the survey as settling that argument. It does not settle anything: one is a description of a system, the other is what practitioners observe. Both can be true. What the survey actually adds is the texture, that experts rate links from trusted, topically relevant domains and pages with genuine visitors highest, and spam links among the biggest negatives.
Meta description last does not mean your snippet is irrelevant. It means it is not a ranking input, which is what Google has said for years. It still decides whether anyone clicks, and clicks are the signal that did move this year.
The blind spot: AI citations were not on the list
The only AI-visibility datum anywhere in the survey is a single respondent's report: after a client was mentioned and linked in a Reuters article, that client's visibility in ChatGPT rose fourfold over the following week. One anecdote, from one contributor, about one surface. Interesting. Not a factor.
That matters because of who this survey is for. Ranking factor surveys measure how pages earn positions in a list of blue links. AI answers do something different: they read passages, synthesize, and cite. A page can be excellent at the first job and invisible in the second, because the second job depends on whether the system can identify who is speaking and whether anyone else vouches for them.
One long-running entity SEO practitioner, Jason Barnard, made this point publicly after taking part in the survey. His argument, in short: ranking factors attach to a page, while the factors that decide AI visibility attach to a brand. He noted that the three factors he cared most about were not among the hundred-odd boxes he had just scored. That is a criticism of scope, not method, and it is the right lens for reading the results.
Put together, here is the split the survey data implies:
Layer | What lives here | Drawn from the survey | What it is not |
|---|---|---|---|
Page | Intent match, information gain, internal links, mobile usability | Relevance 57.1%, content quality 47.6%, internal links 11.1% | Not something you can fix by editing another page's copy |
Brand | Branded search demand, reputation, entity consistency, third-party corroboration | Brand signals 27.0%, authority and trust 36.5% | Not something you can fix by editing this page at all |
Internal links are worth pausing on. They scored lowest of the top ten, at 11.1%, and drew the warmest commentary in the report: several respondents called them the one lever a site controls end to end, and one compared a good internal link to a very good external backlink. Low nomination share, high control. That combination usually means underused.
What to do with this: two layers, one week
Page layer, where editing still works
Take your ten highest-impression pages. For each one, write down in a single sentence the task the searcher wanted. Then read only the first 150 words of the page and ask whether it delivers that task. If it does not, rewrite the opening to answer the task; do not bolt on another section.
Next, take your five commercial pages and list what a competitor cannot copy: your own numbers, pricing, test results, screenshots, customer language. If the list comes back empty, that emptiness is your brief. Respondents kept reaching for one phrase here, information gain, and it is the hardest of the top factors to fake with more word count.
Then fix internal links on the one page you most want to rank. Find five pages that already cover adjacent ground, and add one descriptive in-body link from each. A good check: the link text should make sense if read aloud alone.
Leave Core Web Vitals for last. It was the most contested block in the whole survey, with respondents split between calling it heavily overrated and calling it critical for large stores. Mobile usability drew far less disagreement and deserves your time first.
Brand layer, where publishing is not the lever
Brand signals drew 27.0% and branded search volume rated near the top of its group, with one respondent summarizing it as the new link. You cannot edit your way into that. What you can do is measure and feed it: count how often people search your name, make sure the same plain description of what you do appears on your site, in your profiles, and in third-party mentions, and earn pages that describe your work for real audiences rather than pages nobody reads.
And run the obvious test that this survey could not: ask the AI surfaces whether they know you. A quick pass through the AI Search Visibility Checker shows whether your brand appears in answers for the queries you care about, which is the closest thing to a baseline until Zyppy's AI study lands.
Where the evidence stops
Three gaps worth holding in mind before you rebuild a strategy on this.
The survey cannot tell you how much any factor weighs. It is expert perception, collected carefully, and the authors are explicit that the factor list was whittled from a larger pool, that some invitees never responded, and that no contributor panel is ever complete.
Nobody knows how satisfaction gets measured. One respondent raised the question of what happens to click-based signals when a growing share of browsing is performed by agents rather than people. The survey has no answer, and neither does anyone else yet.
The AI study is still pending. Until it is published, every statement about what makes a brand citable in AI answers, including ours, rests on inference from adjacent evidence rather than measurement. That is worth saying plainly in a year when a lot of confident AI-visibility advice is being written from the same position.
Our own read: the survey's most useful message is that the top of the list barely moved while the middle did. Click signals and brand signals are where the argument is now, and they are also the two factors least responsive to a content calendar.
FAQ
What were the top ranking factors in the 2026 survey? Relevance (57.1%), backlinks (54.8%) and content quality (47.6%) were named in the top three by the largest share of the 131 experts. Authority and trust (36.5%), behavior and click signals (29.4%) and brand signals (27.0%) formed the middle tier. Search intent match was the highest-rated individual factor of all 103 tested.
Do backlinks still matter in 2026? According to these experts, yes: links ranked second overall, and links from trusted, topically relevant domains with real visitors scored among the highest factors in the survey, while spam links scored among the most negative. The nuance is that quality and relevance dominate volume, and one respondent noted that anchor profile composition matters more than raw counts.
Did the survey measure AI citations or AI Overviews? No. It measured factors for traditional organic ranking. The only AI-related data point is one respondent's report of a client's ChatGPT visibility rising after a Reuters mention, and the survey team has announced a separate study on AI answers that had not been published at the time of writing.
Author: Lydia Hart, Brand Entity Strategist for 200+ Entity Audits at Auspia. Lydia writes about entity clarity, brand facts, and the signals that decide whether AI systems can identify and vouch for a company.
