{
  "measured_by": "GetIntel",
  "measured_on": "2026-08-11",
  "question": "What does share of voice mean in generative engines, and how much does the definition change the number?",
  "method": "One brand, 60 tracked prompts, four engines, 10 July to 10 August 2026, one set of probe runs. Share of voice was then computed four defensible ways, changing only the denominator.",
  "results": [
    {
      "definition": "Share of all citations returned",
      "numerator": "160",
      "denominator": "16,982",
      "share_of_voice_pct": 0.94,
      "note": "Every domain any engine cited, including news sites, forums and documentation."
    },
    {
      "definition": "Share of citations to category vendors only",
      "numerator": "160",
      "denominator": "2,855",
      "share_of_voice_pct": 5.6,
      "note": "Restricted to the 22 domains selling tools in this category. Excludes Reddit, YouTube, arXiv and every other non-vendor source."
    },
    {
      "definition": "Share of answers that name us",
      "numerator": "162",
      "denominator": "2,320",
      "share_of_voice_pct": 6.98,
      "note": "Answer-level rather than citation-level: what fraction of responses mentioned us at all."
    },
    {
      "definition": "Share of prompts that name us at least once",
      "numerator": "24",
      "denominator": "60",
      "share_of_voice_pct": 40.0,
      "note": "Prompt-level, any-run. The most generous defensible reading."
    }
  ],
  "range": {
    "low_pct": 0.94,
    "high_pct": 40.0,
    "ratio": 42.45,
    "ratio_note": "Computed from unrounded values (40.000/0.9422 = 42.46), not from the rounded 0.94."
  },
  "finding": "The same brand's share of voice is 0.94% or 40% depending only on what you divide by, a 42-fold spread. There is no standard definition of share of voice in generative engines, so a number quoted without its denominator is not comparable to anything.",
  "data_quality_note": "The answer-level count is 162 while the citation-level count is 160. Two runs are flagged as mentioning us but did not yield a parseable getintel.ai citation URL. We report both rather than reconciling them silently.",
  "which_to_use": "For tracking your own progress, prompt-level any-run is the most stable and the least flattering to churn. For competitive comparison, share of citations to category vendors is the only one where the denominator is a comparable field. Never compare your number to a competitor's published figure without knowing which of these they used.",
  "limits": "One brand, one category, one month. The 22-domain vendor list behind the 2,855 denominator is our own classification and is published in full in this file under vendor_denominator.",
  "vendor_denominator": {
    "count": 22,
    "domains": [
      "ahrefs.com",
      "aiclicks.io",
      "conbersa.ai",
      "dageno.ai",
      "foglift.io",
      "frase.io",
      "getintel.ai",
      "growthproai.com",
      "kime.ai",
      "llmpulse.ai",
      "llmrefs.com",
      "otterly.ai",
      "rankability.com",
      "sanbi.ai",
      "searchscore.ai",
      "semrush.com",
      "seranking.com",
      "siftly.ai",
      "therankmasters.com",
      "trysight.ai",
      "tryprofound.com",
      "useomnia.com"
    ],
    "note": "This is the list behind the 2,855 denominator, published here because it was not published anywhere else. An earlier version of this file pointed at llms-txt-citations-2026-08-11.json as the source, which was wrong: that file flags 18 vendor domains drawn from a different and larger corpus. The four additional domains here - conbersa.ai, growthproai.com, sanbi.ai and searchscore.ai - surfaced in the prompt-ownership data instead.",
    "classification": "Ours, and disputable. Semrush and Ahrefs are general SEO suites with AI-visibility features rather than dedicated tools; moving them out lowers the denominator and raises our 5.60%."
  }
}