Browse the evidence base

Showing 1–50 of 1751 claims

• Google’s current ranking-systems guide says passage ranking identifies individual sections or passages of a page to better understand the page’s relevance to a search.Supported
  mechanism
• Google’s current documentation says interaction data is transformed into signals that help machine-learned systems estimate relevance.Supported
  mechanismDo clicks move rankings? Documented vs. leaked
• Microsoft states that a Grounding with Bing Search model response includes a link to the Bing query used for the search.Supported
  mechanism
• Perplexity currently says Pro Search conducts multiple searches across the web and can draw from different source types depending on the selected focus.Supported
  mechanism
• Microsoft states that a model response using Grounding with Bing Search includes citations linking to websites used to generate the response.Supported
  mechanism
• Google’s current How Search Works documentation says aggregated and anonymized interaction data is used to assess whether search results are relevant to queries.Supported
  mechanismDo clicks move rankings? Documented vs. leaked
• Microsoft says Foundry grounding agents identify information gaps and construct search queries during query formulation.Supported
  mechanism
• Microsoft states that developers and end users do not receive the raw content returned by Grounding with Bing Search.Supported
  mechanism
• When Googlebot is allowed to fetch a page and sees a noindex directive, the directive is intended to keep that page out of Google's index.Supported
  causal
• Google has claimed that consistently fast server responses can allow Googlebot to increase its crawl rate or fetch more content with available connections.Supported
  causal
• Google says AI Overviews may use query fan-out that issues multiple related searches across subtopics and data sources.Supported
  mechanismAI search visibility: what is documented, what is early research
• Google described consolidating properties or signals such as link popularity from duplicate URLs to the representative canonical URL.Supported
  mechanismCanonical URLs: how duplicates get consolidated
• Google has claimed that exposing many low-value URLs can negatively affect a site's crawling and indexing.Supported
  causalRedirects and error codes: what Googlebot does with them
• Google described selecting one representative or canonical URL from a duplicate cluster for search results.Supported
  mechanismCanonical URLs: how duplicates get consolidated
• If robots.txt blocks Googlebot from fetching a page, Google cannot read robots meta directives located in that page's HTML.Supported
  mechanism
• Faceted navigation and other combinatorial URL patterns can create very large URL spaces that waste crawler resources and slow discovery of useful URLs.Supported
  causal
• Google has described a declared canonical preference as a hint rather than an absolute command or rule.Supported
  mechanismCanonical URLs: how duplicates get consolidated
• Server errors, timeouts, or overload signals can cause Googlebot to reduce crawling.Supported
  causal
• Google’s current ranking-systems guide says neural matching understands conceptual representations in queries and pages and matches them to each other.Supported
  mechanism
• Author bylines do not directly give a page a ranking boost in Google Search.Supported
  negative
• Google stated in 2024 that it has systems that can treat a distinct site section more like a standalone site so it does not receive a ranking boost merely from the host site’s reputation.Supported
  mechanism
• Ahrefs claimed in 2026 that stronger topical authority improves visibility in AI search results as well as conventional rankings.Supported
  causal
• A 2025 entity-SEO article claimed entity-based internal linking can clarify relationships among entities and may broaden the queries for which pages rank.Supported
  causal
• A 2026 SEO article claimed topically relevant internal links strengthen a destination page’s perceived authority on a subject, while unrelated links can dilute it.Supported
  causal
• Ahrefs claimed in 2026 that comprehensive topical coverage can increase AI-search visibility because query fan-out creates multiple related subqueries.Supported
  mechanism
• Ahrefs stated in 2026 that topical authority is built by comprehensive topic coverage connected through related content and internal links.Supported
  causal
• Bing's current Webmaster Guidelines state that Bing and Copilot search experiences rely on the same core crawling, indexing, and ranking foundation as traditional search.Supported
  mechanism
• Google Cloud says Grounding with Google Search results can be customized to an end user's geographic location using latitude and longitude.Supported
  mechanism
• Current Google documentation states that permanent redirects are used by the indexing pipeline as a signal that the redirect target should become canonical.Supported
  mechanismRedirects and error codes: what Googlebot does with them
• Microsoft says a Foundry model using Grounding with Bing Search may decide which pieces of tool output are used to generate the response.Supported
  mechanism
• Microsoft Bing stated in 2025 that canonical tags can improve the likelihood that the preferred original page is used in search results.Supported
  causalCanonical URLs: how duplicates get consolidated
• Google says its generative Search features can identify additional supporting webpages, enabling a wider and more diverse set of helpful links than classic web search.Supported
  mechanism
• Submitting or exposing a Sitemap can tell Google about URLs it may not otherwise know about.Supported
  causal
• Perplexity currently says Pro Search reads, analyzes, and compiles insights from dozens of sources before summarizing them into an answer.Supported
  mechanism
• A 2025 Search Engine Land guide claims that clear canonicalization helps generative search systems identify the preferred source/version to ingest, retrieve, summarize, or attribute.Supported
  causal
• Perplexity currently says its source labels rate the source website as a whole rather than each individual page.Supported
  mechanism
• OpenAI currently says ChatGPT Search may use relevant information from enabled memories when rewriting a user prompt into a search query.Supported
  mechanism
• Microsoft states that Bing's user-engagement ranking parameter considers whether users adjust or reformulate their query.Supported
  mechanism
• Microsoft Bing stated in 2025 that canonical tags can improve the likelihood that the preferred original page is used in AI answers.Supported
  causalCanonical URLs: how duplicates get consolidated
• FeatGEO represents webpages using interpretable structural, content, and linguistic properties for generative citation-visibility optimization.Supported
  mechanism
• Google states that membership in a consistent hreflang cluster can influence which regional URL is preferred as canonical.Supported
  mechanism
• Google described grouping detected duplicate URLs into a cluster.Supported
  mechanismCanonical URLs: how duplicates get consolidated
• OpenAI currently says ChatGPT Search may use general location inferred from a user IP address to improve the relevance of search results.Supported
  mechanism
• OpenAI currently says that when a site allows both OAI-SearchBot and GPTBot, it may reuse results from a single crawl for both search and foundation-model-training use cases.Supported
  mechanism
• Microsoft Bing stated in 2025 that duplicate content can dilute ranking or selection signals.Supported
  causalCanonical URLs: how duplicates get consolidated
• OpenAI currently says OAI-SearchBot and GPTBot robots controls are independent, so a site can allow search crawling while disallowing potential foundation-model training use.Supported
  mechanism
• Perplexity currently says the Sonar API can use a classifier to determine whether web search is needed for a query.Supported
  mechanism
• Current Google documentation states that a temporary redirect is followed but is not itself used by the indexing pipeline as a signal that the target should become canonical.Supported
  mechanismRedirects and error codes: what Googlebot does with them
• Backlinks with high Semrush toxicity ratings were claimed to be potentially capable of negatively affecting search rankings and therefore worth investigation.Supported
  causal
• Perplexity currently says its source review process uses objective site-level questions, including correction practices, author identification, and separation of news from advertising and opinion, when deciding whether a site receives a label.Supported
  mechanism