Do Backlinks Still Matter for AI Search?

Short answer: Yes — and they matter for a second reason most technical marketers haven't priced in yet.Zyppy's 2026 Google Ranking Factors Expert Survey collected 13,665 ratings from 131 SEO professionals across 103 potential ranking factors. Relevance (57.1%), backlinks (54.8%), and content quality (47.6%) came out as the three factors experts pick most often as Google's top signals. 

The finding that matters most for scientific marketers: original research and first-party data are the highest-scoring content signals in the survey — well above AI-generated content produced at scale with no added value.

Key Takeaways

• Relevance wins. Matching what the searcher actually needs — not just the keyword — was the single highest-rated factor of the 103 evaluated.
• Backlinks are alive. 54.8% of experts named them a top-3 ranking factor. The shift is quality over volume: links from trusted, topically relevant sites with real visitors score high; spam links are one of the strongest negative signals.
• Content quality means original work. Original research and first-party data rated highest. AI-generated content only scored negatively when it was scaled without adding anything new — quality AI content with real human review and unique value held up fine.
• Technical SEO is table stakes, not a differentiator. Experts described it as foundational: bad technical SEO can sink a site, but good technical SEO alone won't lift mediocre content.
• Backlinks now double as AI training signal. One surveyed practitioner reported a client's ChatGPT visibility went up 4x within weeks of being mentioned and linked by Reuters — the same mechanic behind Covalent Bond's Credibility Bridge trade media strategy and HORIBA's AI search visibility win.

What the survey actually measured

Zyppy (led by Cyrus Shepard) ran the survey to reflect a year of real disruption to how Google ranks pages: a 2024 leak of Google's internal API documentation, U.S. v. Google antitrust trial testimony in which Google's own engineers confirmed under oath that ranking systems like Navboost — which re-ranks search results using real click and engagement data, rather than links or on-page content alone — actually exist, and the shift toward AI Overviews and answer engines. (For the full technical breakdown of what the leak and testimony revealed, see iPullRank's analysis.)

131 SEO practitioners — from agencies and platforms including Ahrefs, HubSpot, Moz, Wix, iPullRank, and SearchPilot — rated each of 103 potential ranking factors on a seven-point scale, then separately named their top three. That produced 13,665 individual data points, representing roughly 44 hours of expert time.

It's worth being precise about what this is: an expert opinion survey, not a controlled ranking test. The value is in aggregating what 131 people who watch Google's algorithm for a living believe is actually happening. Take it as a useful directional signal, not causal proof.

The result that matters most for scientific marketing: links are doing double duty

Ten years ago, "are backlinks dead" was a live debate. The 2026 survey settles it in the opposite direction experts expected: 54.8% still name backlinks a top-3 factor, second only to relevance. What's changed is why they matter. Links from real, topically relevant sites with actual audiences score highest; the rest barely move the needle or actively hurt.

That distinction is exactly where AEO — Answer Engine Optimization — and traditional SEO now overlap. Google's ranking algorithm and AI models like ChatGPT and Perplexity both draw on the same evidence of trust: who links to you, and who those linkers are. The Reuters example in the survey — a single mention and link that quadrupled a client's ChatGPT visibility in weeks — is a direct data point on that overlap. A backlink from a trusted trade publication isn't just a ranking signal anymore. It's training data for the systems buyers now ask instead of Google.

For instrument, reagent, CRO, CDMO, and pharma/biotech supplier companies, that reframes where content and PR budget should go. A placement in a trade publication or a peer-reviewed citation isn't a vanity metric — it's a direct input into whether an AI engine cites your company as an authoritative source when a scientist asks it a technical question. We've written before about using backlink and audience data to pick the trade publications actually worth the placement fee — this survey is independent confirmation of why that selection work pays off in AI citations, not just impressions.

What the content-quality finding confirms

The survey's content-quality result lines up with something most scientific companies already have and underuse: original research and first-party data. Experts rated these the highest-scoring content signals — well above AI-generated content produced at scale with no added value. Scientific companies sit on proprietary data, application notes, and technical findings that generic content marketers can't fabricate. That's a structural advantage in an environment where Google and AI engines are both filtering out commoditized content.

The practical implication: publish the technical findings you already have, structured so both Google and AI models can extract and cite them — clear claims, defined terms, and a direct answer near the top of the page, not buried three paragraphs into scene-setting.

What this means for your AEO strategy

1. Prioritize earned links over volume. One citation from a trusted, relevant publication outweighs a dozen low-quality links — for rankings and for AI citation odds alike.
2. Publish what only you know. Original data, application notes, and technical findings outperform generic or lightly-edited AI content in both channels.
3. Treat technical SEO as the floor, not the strategy. Fix crawlability, schema, and indexing — then put the real effort into relevance and evidence, not more technical tuning.
4. Structure content to answer the question first. Relevance — matching what the reader or the AI model's user actually needs — was the single highest-rated factor in the entire survey.
5. Go deep on a topic, not wide across many. Topical authority scored as its own distinct factor in the survey (14.3% named it top-3). We've covered why owning a whole topic cluster beats ranking for a single prompt — this survey adds Google's ranking side of that same argument.

For a fuller audit of where your site stands on these factors, our GEO/AEO Checklist walks through them one by one.

Frequently Asked Questions

Do backlinks still matter for SEO in 2026?

Yes. 54.8% of the 131 experts surveyed named backlinks a top-3 Google ranking factor — second only to relevance. Quality from trusted, topically relevant sites matters far more than volume.

Does AI-generated content hurt search rankings?

Only when it's produced at scale without adding anything new. AI content paired with real human review and unique value rated favorably in the survey; low-effort scaled AI content rated negatively.

What's the single most important Google ranking factor, according to the survey?

Relevance — how well content matches what the searcher actually needs, beyond keyword matching. It was the highest-rated factor among all 103 evaluated.

Do meta descriptions still matter if they don't affect rankings?

For rankings, barely — the survey rated them near the bottom of all 103 factors. They still matter for click-through rate: as we've covered, a meta description that matches actual search intent is what gets a searcher to click your result over the one above it, even though it won't move you up the page.

Do backlinks affect AI search visibility, not just Google rankings?

The survey included a reported case of a client's ChatGPT visibility increasing 4x within weeks of a link and mention from Reuters, suggesting backlinks now influence both traditional rankings and AI citation likelihood.