Methodology

Built for the sceptical reader

How we run the analysis, what the numbers mean, and where we draw the line on what we will and won't claim. The honesty is the credibility.

How we measure

We don't run prompts. We simulate the conversations your customers actually have, from inside their market.

Personas, not prompts

We simulate full conversations as your actual customer personas, not one-off prompts. Volume is driven by how many brands, how many markets, and how many buyer personas.

Multi-turn conversations

Each scenario runs as a multi-turn dialogue in the persona's own language, because a conversation is a chain of decisions, not a single answer.

In-country IP addresses

We run from in-country IP addresses, which is what makes it a true local-market analysis rather than a translated one.

Competitor set per brand

Even inside the same market, each brand faces a different competitor set, which means genuinely different conversations.

Math, not vibes

A consulting firm runs 1,000 calls and reports noise. We run 100,000 and give a recommendation rate with a mathematically correct confidence interval.

73.2%± 4.1%

How volume is sized

Volume is driven by three things: how many brands, how many markets, and how many buyer personas. Roughly 30 scenarios per brand produce around 3,000 conversations a month.

~30 scenarios / brand~3,000 conversations / month
Engines covered
ChatGPTGeminiPerplexityGoogle AI OverviewsCopilotClaudeGrokDeepSeekMeta AI
What we won't claim

Where we draw the line

Honesty about limits has consistently built more trust than confidence about everything.

Are these real customer questions?

No. The scenarios are realistically simulated, built as close as we can get to how your customers actually ask. That modelling choice is exactly why the sample can be large enough to be statistically meaningful.

Can you prove the revenue came from this?

Citation monitoring is correlational, not deterministic. We can show you which sources the models read, how often, and how that moved. We will not claim a causal chain we cannot prove.

How fresh are the numbers?

Targets are labelled as targets, not guarantees. Baselines under two weeks old are labelled directional until enough data accumulates.

Ready to win
AI recommendations?

Join leading brands using Genezio to understand, influence, and win AI driven purchase decisions.
Move beyond visibility, optimize for recommendations that drive conversions.