AI-visibility research · Practical tools
AI decides who gets recommended. We study why and give you tools to improve AI visibility.
Research-based insights into how AI systems retrieve, cite, and recommend information.
How BLURSOR works
-
We monitor new research papers on AI visibility, LLM ranking factors, and how AI finds and cites sources.
-
We filter for findings with practical implications and distill the useful parts into readable articles.
-
We publish the findings, supporting evidence, and what they could mean for your work.
-
We turn useful findings into tools that help you investigate problems and improve AI visibility.
From the papers we've distilled
What the research actually says
85.7%
Brand-reputation citations pointed to third-party pages. Find the domains AI cites in your category.
Source · arXiv:2606.25787
1 in 10
citation failures came from technical problems. Check access, JavaScript rendering and extractability before rewriting.
Source · arXiv:2603.09296
AI Crawlability Checker
Check whether AI crawlers can access and read a page. See the applicable rules, server responses, raw content, and JavaScript-rendered version.
Check a page →
+29.6%
In one RAG setup, entity-focused pages improved answer accuracy; appended JSON-LD had little effect.
Source · arXiv:2603.10700
21–32 pts
Rewording the same question cut overlap between brand recommendations by 21–32 points. Test several phrasings for each intent.
Source · arXiv:2605.27440
The research, distilled
Latest digests
01
Body-copy rewrites can cut AI answer visibility
In a realistic AI-search benchmark, the common instinct to rewrite page text for LLMs often hurt the odds of being retrieved, reranked, and cited.
02
Citation manipulation lost ground in one AI-search lab test
Before buying a citation-boosting rewrite, ask what the claimed improvement was tested against.
03
Source-attributed cues can push multiple-choice models off the right answer
How “source said so” cues destabilize correct answers in multiple-choice QA