{
  "measured_by": "GetIntel",
  "measured_on": "2026-08-11",
  "question": "What does a normal AI visibility level actually look like across brands?",
  "privacy": "Aggregate only. No brand names, identifiers, industries or per-brand rows are published. Percentiles and counts across 63 brands, with no minimum or maximum reported, so no individual brand is identifiable from this file.",
  "method": "Every brand in our system with at least 20 probe runs on its live prompt set. For each we computed the share of its own tracked prompts where at least one run named the brand, then took the distribution across brands. Brands track different prompt-set sizes, so the split by set size is reported to show the result is not an artefact of that.",
  "scope": {
    "brands": 63,
    "total_probe_runs": 58016,
    "prompt_set_size_percentiles": {
      "p25": 10,
      "p50": 60,
      "p75": 60
    }
  },
  "distribution_pct_of_own_prompts_naming_the_brand": {
    "p10": 0.0,
    "p25": 3.3,
    "p50": 28.8,
    "p75": 90.0,
    "p90": 98.5,
    "mean": 41.4
  },
  "shape": {
    "named_on_zero_prompts": 14,
    "under_10_pct": 21,
    "between_10_and_50_pct": 16,
    "at_or_above_50_pct": 26,
    "note": "Bimodal. Only 16 of 63 brands sit in the 10-50% middle; the rest cluster at one end or the other."
  },
  "not_an_artefact_of_prompt_set_size": {
    "brands_with_50plus_prompts": {
      "n": 37,
      "zero": 6,
      "under_10": 12,
      "mid": 11,
      "at_or_above_50": 14,
      "median": 28.8
    },
    "brands_with_under_50_prompts": {
      "n": 26,
      "zero": 8,
      "under_10": 9,
      "mid": 5,
      "at_or_above_50": 12,
      "median": 30.0
    },
    "conclusion": "Medians are close, 28.8% against 30.0%, and the bimodal shape is present in both groups. But they are not identical: the smaller-prompt-set group is more polarised, with 30.8% of its brands at zero against 16.2%, and a thinner middle at 19.2% against 29.7%. Prompt-set size does not create the bimodality but it does sharpen it, so the rebuttal is that size is not the cause rather than that size is irrelevant."
  },
  "not_explained_by_recency": "Brands named on zero of their prompts have a median of 300 probe runs, identical to the median for every other brand. The invisible ones are not simply newer or less measured.",
  "limits": "One platform's customer base, which self-selects for companies that already suspect they have a visibility problem, so this is not a random sample of businesses. Prompt sets are chosen per brand and are not comparable instruments. Read the shape of the distribution rather than any single percentile as a target."
}