Ultimate GSO Guide for Skeptics: How to accurately estimate brand mentions in generative engines

Introduction: The doubt everyone has

What’s the point of ChatGPT, Gemini or Perplexity mentioning your brand if you don’t know how many people ask those questions? It’s a common objection. Marketing teams want to know:

• Is GSO worth investing in?
• Can we measure how often we’re mentioned?
• If there’s no public volume data, how do I calculate ROI?

The short answer is: yes, you can measure mentions in a statistically valid way. The longer answer is that you need a robust methodology and a working understanding of how generative engines use RAG (Retrieval-Augmented Generation). In this guide, we debunk the biggest myths and show how to turn uncertainty into actionable data.

1. What is GSO and why it matters more than ever

GSO (Generative Search Optimization) is the practice of maximizing your brand’s visibility in generative engines like ChatGPT, Google Gemini and Perplexity AI.

Traditional SEO focuses on ranking your content at the top of Google to get clicks, while GSO aims for your content to be used directly in AI-generated answers. This shift is powered by the adoption of RAG, which lets language models use external data to produce more accurate responses.

2. The role of RAG and Google’s “grounding”

RAG is a two-step process:

1. Retrieval: The AI looks up relevant information from a database or the web.
2. Generation: It uses that information to create the answer.

Google’s grounding is how Gemini 1.5 Pro uses RAG, connecting to Google Search to fetch real-time data and cite sources—reducing hallucinations and increasing reliability.

How does ChatGPT’s RAG work?

ChatGPT’s RAG operates via its web-browsing capability, often called “Browse with Bing.” When a query needs up-to-date or source-specific information, this tool is triggered. ChatGPT searches the web with Bing, retrieves relevant page content, then uses that text as context to generate its answer, citing links as footnotes. This enables answers about recent events or data from specific sites.

3. Web browsing for (almost) everyone

Today, most ChatGPT users have access to web browsing thanks to models like GPT-4o, which brings RAG features to the masses. Google Gemini has also opened web browsing for free users. This isn’t a future change—it’s the current reality.

4. The big shift in web traffic

The expansion of RAG by giants like OpenAI and Google is expected to restructure web traffic. While exact volumes are hard to predict, projections point to a significant rise in the importance of AI-generated answers, potentially diverting some traffic that historically came from organic clicks.

Being present in those answers provides direct visibility and builds brand authority by being cited as a trusted source.

5. How to measure your brand visibility in GSO

Despite the variability of AI-generated answers, it’s possible to estimate your brand’s visibility in a statistically valid way. The key is a robust methodology that accounts for the non-deterministic nature of generative engines. Unlike traditional search results that are almost identical for the same query, an AI’s answer can vary each time. This estimation is valid because, instead of measuring the true query volume (which isn’t publicly available), you measure how consistently your brand is mentioned across a predefined universe of queries. The data reveals mention frequency and confidence intervals for your positioning.

Factors that influence a generative engine’s answer

• a. Non-deterministic nature: Generative models are intrinsically stochastic and won’t always give the same answer.
• b. API parameters: Temperature, Top-P and Top-K control creativity, diversity and coherence.
• c. Training data and process: Training sources and knowledge cutoff affect what gets mentioned.
• d. RAG and real-time search: Quality and availability of retrieved sources.
• e. User and environment variables: Location, chat history, prompt formulation and language.

6. Why GSO isn’t optional

Visibility in generative engines is the new battleground of digital marketing. Unlike traditional rankings, once a brand is established as a trusted source in knowledge graphs, AI is far more likely to cite it repeatedly. Staying out now can mean years of delay.