Unusual queries.
Useful clues.

Queries can contain instructions, long persona prompts, short replies and repeated questions. Some shapes are consistent with AI search or automated monitoring. The scanner reads a requested 16-month window and flags patterns for inspection. Query text alone does not identify an AI service or prove how a search started.

The scanner looks for recurring prompt instructions, persona wrappers, tracker probes, conversation pivots and reply fragments. Ordinary searches can share these shapes. A flagged row is a candidate to inspect, not a confirmed AI query.

It can group recurring templates and surface brands named in them. The analyst may suggest a pattern with a confidence label, but Search Console queries do not verify the source or operator. The member queries and their metrics remain available so you can assess the inference.

Open query evidence to read the returned query text, impressions, clicks, dates, days seen, country and associated page when supplied. Expand a program to inspect its member examples and grouping rationale. Export the displayed queries as CSV or the complete returned report as JSON. Totals describe returned Search Console rows, which may be limited or aggregated.

The scan is free. Google asks for openid, email and Search Console read-only access. Your Google address identifies the report owner and receives the saved report. Choose a property available to that account. If only one property is returned, scanning starts automatically.

The Search Console permission is read-only. Selected query evidence and metrics are sent to DeepSeek for analysis. Google tokens, your email and the private report have a 30-day retention window. The report link requires its requesting Google account; copying it does not grant access. Your address is used for report ownership and delivery, not automatic marketing enrollment.

Low click counts can support a hypothesis of automated search, but they do not prove it. Ordinary searches also produce impressions without clicks. Read the phrasing, repetition, dates and other evidence together before drawing a conclusion.

The scanner inspects past query rows from Search Console. Trakkr asks selected questions across AI models over time and tracks brand mentions, citations and rank in their answers. Signed-in users can review selected questions in their prompt form. Trial setup accepts one editable question suggestion.