PAPER
GEO — the paper that named generative engine optimization
Can you make your content more visible inside a generative engine's answer?
GEO: Generative Engine Optimization
- Authors
- Pranjal Aggarwal · Vishvak Murahari · Tanmay Rajpurohit · Ashwin Kalyan · Karthik Narasimhan · Ameet Deshpande
- Affiliation
- Academic research (Princeton and others)
- Venue
- Published at KDD 2024
- Submitted
- 2023-11-16 · revised 2024-06-28 (v3)
- arXiv
- arXiv:2311.09735
- We verified
- 2026-08-22
WHAT THE PAPER SAYS
The authors concluded you can. This paper introduced the term GEO (Generative Engine Optimization) and defined it as a black-box optimization framework for raising content visibility in generative engine responses. On their own benchmark, GEO-bench, they reported visibility gains of up to 40%.
Read the original on arXiv ↗METHOD
How it was measured
Read the conditions before the numbers. The same figure means something different under a different sample or environment.
- Approach
- Treat the generative engine as a black box and optimize a visibility metric by changing wording and structure on the content side
- Evaluation environment
- GEO-bench, a benchmark the authors built
- Definition of visibility
- The paper defines its own metric for exposure within a response
FINDINGS
What came out
- Headline number
- Visibility in generative engine responses improved by up to 40%
- ⚠️ The condition on that number
- It is a value inside GEO-bench, a benchmark the authors built themselves. It is not a guaranteed result in a live service such as ChatGPT or Perplexity, and the paper does not claim otherwise
LIMITATIONS
Limitations the authors state themselves
Not our criticism — this is what the authors wrote in the paper.
- Evaluation ran in a benchmark environment the authors assembled — not the full pipeline of a commercial engine.
- The paper defines its own 'visibility' metric, so its numbers cannot be compared directly with studies using a different definition.
NAVIRANG'S READING — NOT THE PAPER'S CONCLUSION
This paper set the concept of GEO and deserves the credit for it. The problem is how it gets cited in the market — proposals routinely quote 'the paper says a 40% increase' with nothing attached. The 40% belongs to GEO-bench under specific conditions, and SAGEO Arena below reports that the same family of methods performed worse when reproduced in a realistic pipeline. This is the textbook case of a number that must not be separated from its conditions.
IN COMPARISON
Where it diverges from other papers
The claims-table entries this paper appears in.
RELATED
Related reading
If you want your own brand's numbers rather than a paper's
Every figure here came from someone else's sample. Send us a URL and we ask all 7 answer engines directly and measure yours.
We reply within one business day.