B
Evidence Grade B
Graded
under review
Safe to Cite
Recent
25%
25% -- preference flip rate for LLM-based rerankers (2025)
Attribution:
View source
Lighthouse Research Team
Last verified:
·
How we verify data →
Source verified
Evidence graded
Value
25%
Unit
Percentage
Data Vintage
2025
Entity
Evidence Quality
B
Strong, reliable evidence
Sourced from a well-regarded research firm or industry leader with transparent methodology and a reasonably large sample.
Single source - not yet corroborated
Score breakdown
Extraction
75
Source Trust
60
Citation Chain
20
Independence
30
Methodology
45
Freshness
95
Source
resollm.ai
Primary
Tier B
https://resollm.ai/blog/do-llms-favor-recent-content/ ↗
Source Article
Do Large Language Models Favor Recent Content?
Two search results that say the same thing. Same wording. Same meaning.
The only difference is that one looks “newer” because it has a recent-looking date at the top.
Researchers recently tested whether that single cue is enough to sway large language models. They took identical passages and changed…
Original Research
resollm.ai
· Original study ↗
Methodology
Side-by-side pairwise relevance tests with identical passages differing only in date
Methodology documented
Segmentation
Industrymarketing
Cite this stat
Plain Text
LLM-based rerankers: 25% preference flip rate. Source: resollm.ai (2025). Via Lighthouse Intelligence -- https://lighthousedata.io/data/llm-preference-flip-25-percent
HTML Embed
<blockquote cite="https://lighthousedata.io/data/llm-preference-flip-25-percent" style="border-left:3px solid #2563eb;padding:12px 16px;margin:16px 0;font-family:system-ui,sans-serif;background:#f0f4ff;"><strong>25%</strong> -- preference flip rate for LLM-based rerankers<br><small>Source: resollm.ai (2025) · <a href="https://lighthousedata.io/data/llm-preference-flip-25-percent" target="_blank" rel="noopener">Lighthouse Intelligence</a></small></blockquote>
Share
Use this data in your work
Copy the citation, embed the widget, or explore thousands more verified marketing statistics -- all free.