Source data
Pulse collects publicly available search trend signals for supported regions. A snapshot records surfaced queries, approximate search volume, growth, related searches, source coverage, and collection time. Values are estimates and should be read as directional signals, not audited totals of every search performed.
Organization and classification
Queries are normalized and grouped when the evidence indicates the same real-world subject. Each topic is assigned a subject category and an apparent intent such as following, explaining, planning, buying, watching, worrying, remembering, debating, or playing. These labels are interpretations and can be imperfect, particularly for ambiguous names, multilingual terms, and fast-moving events.
Automated summaries
Explanatory text is generated with AI from the trend record, related searches, dated observations, and linked source titles supplied to the system. The system is instructed not to invent unsupported facts or treat a search-observation date as the date an event occurred. Automated output may still contain errors, so source links and correction channels remain part of the publication.
Images and source reporting
Images shown in trend interfaces are previews associated with search terms or linked source reporting and remain attributed to their source publishers. Pulse asks search engines not to index those publisher images from its pages. Image appearance does not imply ownership, endorsement, or that an image is direct evidence for every part of a summary.
Known limitations
- Search data reflects attention, not a representative opinion poll.
- Approximate volumes can be bucketed, delayed, or revised.
- Language and entity ambiguity can produce incorrect grouping.
- Source availability differs by topic, language, and region.
- Fast-moving stories can change after a snapshot is published.