Observe the consumer search experience
To measure recommendations and citations your customers encounter, observe the consumer surface. querying.ai collects the answer, sources, timing and surface-specific extras. Treat these observations separately from a model API response.
One submission contract, surface-specific responses
Submit taskType and payload to POST /v1/async/task to receive an ID. Collect the result by webhook or polling. Search queries, Google search envelopes and Naver product cards have different fields; use the engine reference rather than assuming every response is identical.
Repeatable market comparisons
Set payload.country to a code such as KR or US and write the prompt in the desired language. Retain the supported market and collection time. Repeat a fixed prompt set instead of treating one answer as representative of the whole market.
Requests, costs and failure handling
Public answer engines cost 1–2 credits per completed task. Plans include monthly credits and concurrency limits; failed tasks are not charged. Distinguish a collection failure from a Google or Naver search that did not produce an AI summary.
Compare citations across AI search engines
Submit the same earbud recommendation prompt to ChatGPT, Gemini and Perplexity, then include Naver AI Briefing and AI Tab for Korean market research. Store prose, citations and shopping cards separately to compare which products each engine recommends.
- Store the answer with its taskType, prompt, country and collection time.
- Group source URLs by domain while counting AI citations separately from organic results.
- Use Monitors for recurring brand tracking and Source Influence to check whether cited pages support the answer.
Your first request
Start with GEMINI to verify your integration, then apply the input fields and response schema for each engine.
{
"taskType": "GEMINI",
"payload": {
"prompt": "best wireless earbuds for commuting",
"country": "US"
}
}