get_ai_visibility

How often AI assistants (ChatGPT, Perplexity, and any other engine the deployment probes) mention, recommend or cite the brand when asked the organization’s buyer-intent prompts: the generative-engine (GEO) measurement. Per engine and overall, over a rolling window compared with the prior one, with competitor mention rates on the same neutral questions, the brand’s pages the engines cite, and the latest answer per prompt.

AI answers are non-deterministic, so the rolling window (default four-week window; cadence is configurable) is the unit; a single week’s move is noise. Rates are over sampled answers, never over “the internet”.

Parameters#

Returns#

{
  "status": "ok",
  "window": { "from": "2026-08-25", "to": "2026-09-21", "weeks": 4 },
  "active_prompts": 24,
  "overall": {
    "samples": 192,
    "prompts": 24,
    "mentioned_rate": 37.5,
    "recommended_rate": 21.4,
    "citation_rate": 12.0,
    "avg_position": 2.6,
    "mentioned_rate_delta": -6.2,
    "recommended_rate_delta": -3.1,
    "citation_rate_delta": 1.0,
    "competitor_rates": { "hoka.com": 61.5, "salomon.com": 40.1 },
    "top_cited_pages": [{ "url": "https://example.com/blog/best-trail-shoes", "count": 9 }],
    "errors": 0
  },
  "engines": [
    {
      "engine": "perplexity",
      "label": "Perplexity",
      "samples": 96,
      "mentioned_rate": 45.8,
      "recommended_rate": 27.1,
      "citation_rate": 18.7,
      "mentioned_rate_delta": -4.2
    }
  ],
  "prompts": [
    {
      "id": 12,
      "text": "best trail running shoes for ultra runners",
      "category": "discovery",
      "origin": "auto",
      "engines": [
        {
          "engine": "perplexity",
          "run_date": "2026-09-21",
          "mentioned": true,
          "recommended": false,
          "position": 3,
          "cited": true,
          "our_cited_urls": ["https://example.com/blog/best-trail-shoes"],
          "competitors": ["hoka.com"],
          "brands_mentioned": ["Hoka", "Example"]
        }
      ]
    }
  ],
  "last_run_date": "2026-09-21",
  "source": "adlyse.ai_visibility@v2"
}

status is no_prompts before the first derivation and no_data before the first run.

Errors#

site_source_id scopes this operation to a site in the current organization; omitted selects the primary site. Results are sampled API answers. Comparative rates use matched prompt/engine/model/locale cohorts, excluding failed scoring; configuration changes require a new baseline. Only implemented and configured engine adapters can run.

Overall and per-engine rates now cover active neutral discovery questions. measurement.geo contains separate discovery, evaluation, and brand-check benchmarks, question topics and provenance, and evidence-backed investigation leads. Recommendations, text mentions, and own-domain citations are distinct. Each question/configuration receives equal weight; deltas use matched cohorts. Historical scoring remains inspectable but is excluded from the new baseline.

measurement.geo.source_gaps lists sources cited in active neutral discovery answers where competitors are mentioned and the brand is absent. Rows are grouped by URL, topic and engine, with gap_answers, cited_answers, affected questions, answer_ids, dates and competitors. The response returns up to 100 ranked rows and a total; this is a bounded investigation list, not an exhaustive market index. Inspect the linked answer evidence before proposing a change. Citation association does not establish that the source recommended a competitor or caused a result; counts may span different model configurations.

For entity configuration, approved facts, portfolio research, log imports and page outcomes, use get_geo_workspace and update_geo_workspace.