Lumen AIScraping the web experience
We query chatgpt.com and gemini.google.com the way an anonymous user would, through DataForSEO, and analyze that answer.
Real answers
What your customer sees, not what the API returns
An LLM API and the app people actually use do not always recommend the same brands. Lumen measures the logged-out web answer of ChatGPT and Gemini, with its cited sources, so your reports reflect what is really shown.
Real answers
What you get
Closer to reality
In a test of ours with 48 runs, the top-5 brand overlap with the real web was 0.55 with this method, against 0.37 with the API.
Sources and searches included
Each answer carries the cited URLs with the surrounding text and, for ChatGPT, the internal searches the engine ran to build it.
Transparent about the transport
Every result records whether the web or the API answered. If the web read fails, or the location is a city, the API takes over and it is noted.
Real answers
How it works
You define your prompts
Your own or Lumen suggestions, per country and in Spanish, Portuguese or English.
Lumen queries the real web
It runs each prompt on the ChatGPT and Gemini web experience for the client’s country.
It analyzes the answer
It extracts mentions, position, competitors, sentiment and cited sources from what was shown.
You see it in your dashboard
With a history per prompt and engine, to compare runs over time.
Measure what people really see
Available for ChatGPT and Gemini on the plans that include them.