Short answer
In the GEO paper's benchmark, adding quotations raised a source's visibility (position-adjusted word count) by about 41%, adding statistics by about 31%, citing sources by about 28%, and fluency improvements by about 28%. Keyword stuffing lowered it by about 8%. Gains were largest for lower-ranked sources.
Results (GPT-3.5 generative engine)
| Method | Visibility change | Impression change |
|---|---|---|
| Quotation Addition | +40.9% | +28.0% |
| Statistics Addition | +30.6% | +22.8% |
| Cite Sources | +27.5% | +13.5% |
| Fluency Optimization | +28.0% | +13.5% |
| Technical Terms | +17.6% | +10.9% |
| Easy-to-Understand | +14.0% | +6.2% |
| Authoritative tone | +10.4% | +18.7% |
| Unique Words | +6.2% | +5.7% |
| Keyword Stuffing | −8.3% | +4.7% |
Smaller sites benefit most
For sources ranked fifth, visibility rose by about 115% with cited sources, 100% with quotations and 98% with statistics. For top-ranked sources the same methods slightly reduced visibility.
Caveats
- Lab benchmark, 2023–2024 models
- Effects vary by domain
- On a Perplexity subset, gains were smaller but still positive for quotations and statistics
Questions
Should we add statistics everywhere?
Add real, sourced numbers where they help the reader. Invented or irrelevant statistics damage trust.
Facts checked against the sources below on 12 October 2026. Platforms change quickly: if something here no longer matches what you see, tell us and we will update it.
