Model limitations

Geoduck measures controlled responses from selected model APIs. It does not access, reproduce, or rank the logged-in consumer applications called ChatGPT, Claude, Gemini, or similar products.

Results are observations from a specific prompt, configuration, model, mode, and time. They are not guarantees of product placement, traffic, conversion, source causation, or future model behavior.

Web-Grounded and Knowledge-Only results are reported separately. Only models that pass Geoduck's capability checks are eligible for a given mode. Provider failures, incomplete samples, missing provenance, and insufficient evidence are shown as such rather than treated as zero visibility.

Prompt variants test response robustness. They are not a recommendation that wording improves merchant visibility. Evidence, citations, and mention classifications should be reviewed before making commercial decisions.

Read the methodology and Privacy Policy for more detail.