How we measure AI shopping visibility.
Every number Caeliai publishes comes from real shopping conversations with ChatGPT and Gemini, captured as evidence and classified with the same rules every time. This page is the full process, including what it cannot tell you.
The measurement, step by step.
Real conversations, not APIs alone.
We ask ChatGPT and Gemini the questions real shoppers ask, in the consumer apps shoppers actually use, phrased the way a buyer would phrase them. Where a study also reads catalog data (for example Shopify’s catalog endpoints), the paper says so explicitly.
Repeated prompts.
AI answers change between runs. Where the question is about a pattern rather than a single product, we run the same prompt multiple times and report rates across runs, not a single answer.
Record what appears.
For each answer we record which brands are named, which products appear, whether a visual product card renders, and where each product sits in the answer.
Follow the buy path.
The core question: where does the link send the buyer? We classify every observation as owns the path (the brand’s own product page), leaks the path (a retailer, marketplace, or reseller), or no path (named with no usable link).
Check the details.
Is it the right product? Is the price current? Is the first seller the official one? Wrong sellers and stale listings count against the brand even when the brand name appears.
Capture the evidence.
Observations are saved as screenshots, response records, and structured data with dates. Published figures state their sample size and measurement date. An undated chart is a decoration; we do not publish them.
Variability: the honest part.
No single AI answer is definitive. The same prompt can return different brands, different products, and different links an hour apart. That is not a flaw in the measurement; it is a property of the systems being measured, and any vendor who shows you one screenshot as proof is selling you the weather from one photo of the sky.
We separate signal from normal variation by repetition. A brand that appears in one run out of ten is noise. A brand that leaks its buy path to the same reseller across repeated runs and phrasings has a structural problem worth fixing. Rates across repeated observations, with the N stated, are the unit of evidence. Single observations are labeled as single observations.
Limitations we state up front.
These systems change without notice, so every finding is timestamped and can go stale. Most findings are correlational: when catalog presence predicts product cards, that makes a useful diagnostic, not proof of cause. Sample sizes are stated on every figure, and small ones are called early signal, not conclusions. And most of the evidence on this site comes from Caeliai’s own measurements: the methods are documented here so anyone, including a skeptical brand, can rerun them.
Disclosure.
Caeliai research is independent. No brand, platform, or vendor pays for placement in the research, and advisory clients do not buy better results in published studies. Where a study involves a client’s store, that is disclosed. Public research stays open and free to read.
Questions about the method, or think we got something wrong? Email contact@caeliai.com and the founder will answer directly. See also about Caeliai and a sample diagnosis built with this method.