Pranjal Aggarwal

Pranjal Aggarwal led the paper "GEO: Generative Engine Optimization," posted in November 2023 and later accepted to KDD 2024. The work did more than introduce a label. It treated visibility inside generative engines as a black-box optimization problem that researchers could measure.

That framing gave practitioners and academics a shared vocabulary. Instead of arguing only about anecdotes from ChatGPT screenshots, the field could talk about interventions, benchmarks and measured lifts. Agencies adopted the name quickly because it named a commercial problem they were already being asked to solve.

Aggarwal's role as lead author puts him at the origin story every GEO deck now compresses into a single slide. The paper is still the citation people reach for when they need to show the discipline has a research backbone. Read it on arXiv.

The black-box framing also kept the paper honest about limits. The authors did not pretend generative engines expose a clean ranking API. They showed that even without that API, researchers can still measure how content changes move visibility inside the answers users see.

The paper introduced GEO-bench, a benchmark for studying how changes to source content affect whether generative engines surface that source. Among the reported results: targeted edits such as added statistics, quotations and citations could raise a source's visibility by up to 40 percent in the study's setting. Karthik Narasimhan was the senior academic on the same paper.

Those numbers are widely cited, sometimes more casually than the paper warrants. The careful reading is that content structure and evidence density matter to generative retrieval, and that the authors gave the industry a way to test that claim rather than only to mythologize it.

For working teams, the durable lesson is methodological. If you change a page, you should be able to describe the intervention and measure whether generative visibility moved. GEO-bench made that research habit feel normal outside a lab.

GEO-bench also invited follow-on work. Once a benchmark exists, other researchers can argue with it, extend it or break it, which is healthier than a field that only trades screenshots.

Start with the paper itself on arXiv 2311.09735. It remains the founding reference for people who want the academic origin story rather than a vendor summary, and it is short enough to read before a planning meeting.