{
 "id": "people",
 "title": "People search",
 "question": "Which source finds the learning and development people at a company?",
 "n": 100,
 "positives": 127,
 "primary_metric": {
  "name": "recall of the known people",
  "definition": "Share of the known learning and development people at the companies a route took that it returned. Wilson 95% interval."
 },
 "rows": [
  {
   "vendor": "Parallel Task",
   "n": 100,
   "metric": 0.5669,
   "ci_low": 0.48,
   "ci_high": 0.6499,
   "secondary": {
    "precision": 0.8256,
    "recall": 0.5669,
    "coverage": 0.65
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Parallel",
     "usd_per_call": 0.005,
     "usd_per_1k_correct": 0.62,
     "median_latency_s": null
    }
   ],
   "note": ""
  },
  {
   "vendor": "Exa people search",
   "n": 100,
   "metric": 0.5669,
   "ci_low": 0.48,
   "ci_high": 0.6499,
   "secondary": {
    "precision": 0.5365,
    "recall": 0.5669,
    "coverage": 0.64
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Exa",
     "usd_per_call": 0.022,
     "usd_per_1k_correct": 1.72,
     "median_latency_s": 2.1
    }
   ],
   "note": ""
  },
  {
   "vendor": "Exa agent",
   "n": 100,
   "metric": 0.5591,
   "ci_low": 0.4722,
   "ci_high": 0.6424,
   "secondary": {
    "precision": 0.8201,
    "recall": 0.5591,
    "coverage": 0.6
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Exa",
     "usd_per_call": 0.025,
     "usd_per_1k_correct": 3.01,
     "median_latency_s": null
    }
   ],
   "note": ""
  },
  {
   "vendor": "Clay Find People",
   "n": 100,
   "metric": 0.5197,
   "ci_low": 0.4335,
   "ci_high": 0.6047,
   "secondary": {
    "precision": 0.7584,
    "recall": 0.5197,
    "coverage": 0.58
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Clay",
     "usd_per_call": 0.0,
     "usd_per_1k_correct": 0.0,
     "median_latency_s": null
    }
   ],
   "note": ""
  },
  {
   "vendor": "Crustdata",
   "n": 30,
   "metric": 0.5122,
   "ci_low": 0.3648,
   "ci_high": 0.6575,
   "secondary": {
    "precision": 0.7756,
    "recall": 0.5122,
    "coverage": 0.5667
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.1311,
     "usd_per_1k_correct": 13.16,
     "median_latency_s": 1.7
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Lusha",
   "n": 30,
   "metric": 0.4878,
   "ci_low": 0.3425,
   "ci_high": 0.6352,
   "secondary": {
    "precision": 0.7327,
    "recall": 0.4878,
    "coverage": 0.5667
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.11648,
     "usd_per_1k_correct": 11.76,
     "median_latency_s": 1.8
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Prospeo",
   "n": 100,
   "metric": 0.4567,
   "ci_low": 0.3726,
   "ci_high": 0.5433,
   "secondary": {
    "precision": 0.7999,
    "recall": 0.4567,
    "coverage": 0.52
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.02254,
     "usd_per_1k_correct": 2.22,
     "median_latency_s": 1.7
    }
   ],
   "note": ""
  },
  {
   "vendor": "Apify B2Bsearch employees",
   "n": 30,
   "metric": 0.439,
   "ci_low": 0.2989,
   "ci_high": 0.5896,
   "secondary": {
    "precision": 0.7575,
    "recall": 0.439,
    "coverage": 0.4333
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Apify",
     "usd_per_call": 0.0232,
     "usd_per_1k_correct": 2.21,
     "median_latency_s": 0.7
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "AI Ark",
   "n": 30,
   "metric": 0.439,
   "ci_low": 0.2989,
   "ci_high": 0.5896,
   "secondary": {
    "precision": 0.7691,
    "recall": 0.439,
    "coverage": 0.5
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.0389832,
     "usd_per_1k_correct": 3.63,
     "median_latency_s": 33.5
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Apify Atomus employees",
   "n": 30,
   "metric": 0.439,
   "ci_low": 0.2989,
   "ci_high": 0.5896,
   "secondary": {
    "precision": 0.7691,
    "recall": 0.439,
    "coverage": 0.5
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Apify",
     "usd_per_call": 0.08885,
     "usd_per_1k_correct": 8.27,
     "median_latency_s": 0.7
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Clay Surfe",
   "n": 30,
   "metric": 0.4146,
   "ci_low": 0.2776,
   "ci_high": 0.5663,
   "secondary": {
    "precision": 0.7026,
    "recall": 0.4146,
    "coverage": 0.4333
   },
   "tie_group": 1,
   "stores": [
    {
     "store": "Clay",
     "usd_per_call": 0.078963,
     "usd_per_1k_correct": 8.59,
     "median_latency_s": null
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Icypeas",
   "n": 100,
   "metric": 0.3858,
   "ci_low": 0.3057,
   "ci_high": 0.4727,
   "secondary": {
    "precision": 0.7593,
    "recall": 0.3858,
    "coverage": 0.44
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.0046778,
     "usd_per_1k_correct": 0.58,
     "median_latency_s": 1.8
    }
   ],
   "note": ""
  },
  {
   "vendor": "Ocean.io",
   "n": 30,
   "metric": 0.3659,
   "ci_low": 0.2359,
   "ci_high": 0.5188,
   "secondary": {
    "precision": 0.8939,
    "recall": 0.3659,
    "coverage": 0.4
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.10908,
     "usd_per_1k_correct": 21.92,
     "median_latency_s": 2.1
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "QuickEnrich",
   "n": 100,
   "metric": 0.3622,
   "ci_low": 0.2838,
   "ci_high": 0.4487,
   "secondary": {
    "precision": 0.7731,
    "recall": 0.3622,
    "coverage": 0.43
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.0,
     "usd_per_1k_correct": 0.0,
     "median_latency_s": 0.9
    }
   ],
   "note": ""
  },
  {
   "vendor": "Parallel Entity",
   "n": 100,
   "metric": 0.3622,
   "ci_low": 0.2838,
   "ci_high": 0.4487,
   "secondary": {
    "precision": 0.3148,
    "recall": 0.3622,
    "coverage": 0.43
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Parallel",
     "usd_per_call": 0.005,
     "usd_per_1k_correct": 0.7,
     "median_latency_s": 2.1
    }
   ],
   "note": ""
  },
  {
   "vendor": "Crawl4AI",
   "n": 100,
   "metric": 0.3228,
   "ci_low": 0.2478,
   "ci_high": 0.4083,
   "secondary": {
    "precision": 0.7931,
    "recall": 0.3228,
    "coverage": 0.37
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.00041,
     "usd_per_1k_correct": 0.05,
     "median_latency_s": 1.4
    }
   ],
   "note": ""
  },
  {
   "vendor": "Leadsforge",
   "n": 100,
   "metric": 0.2992,
   "ci_low": 0.2264,
   "ci_high": 0.3838,
   "secondary": {
    "precision": 0.8043,
    "recall": 0.2992,
    "coverage": 0.34
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.00025,
     "usd_per_1k_correct": 0.03,
     "median_latency_s": 2.1
    }
   ],
   "note": ""
  },
  {
   "vendor": "LeadOcean",
   "n": 100,
   "metric": 0.2913,
   "ci_low": 0.2194,
   "ci_high": 0.3756,
   "secondary": {
    "precision": 0.6481,
    "recall": 0.2913,
    "coverage": 0.35
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.00721,
     "usd_per_1k_correct": 1.83,
     "median_latency_s": 1.3
    }
   ],
   "note": ""
  },
  {
   "vendor": "Datagma",
   "n": 30,
   "metric": 0.2195,
   "ci_low": 0.12,
   "ci_high": 0.3671,
   "secondary": {
    "precision": 0.2316,
    "recall": 0.2195,
    "coverage": 0.3
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "DeepLine",
     "usd_per_call": 0.014,
     "usd_per_1k_correct": 6.83,
     "median_latency_s": 2.5
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Dropleads",
   "n": 100,
   "metric": 0.1969,
   "ci_low": 0.137,
   "ci_high": 0.2745,
   "secondary": {
    "precision": 0.6652,
    "recall": 0.1969,
    "coverage": 0.22
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.0,
     "usd_per_1k_correct": 0.0,
     "median_latency_s": 3.2
    }
   ],
   "note": ""
  },
  {
   "vendor": "HarvestAPI",
   "n": 97,
   "metric": 0.1371,
   "ci_low": 0.0874,
   "ci_high": 0.2086,
   "secondary": {
    "precision": 0.8048,
    "recall": 0.1371,
    "coverage": 0.1649
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.006,
     "usd_per_1k_correct": 1.5,
     "median_latency_s": 6.6
    }
   ],
   "note": "ran on the 97 companies with a resolved LinkedIn URL"
  },
  {
   "vendor": "Apify memo23 employees",
   "n": 30,
   "metric": 0.122,
   "ci_low": 0.0532,
   "ci_high": 0.2554,
   "secondary": {
    "precision": 0.0683,
    "recall": 0.122,
    "coverage": 0.1667
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Apify",
     "usd_per_call": 0.08433333,
     "usd_per_1k_correct": 62.81,
     "median_latency_s": 0.7
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "LeadMagic employees",
   "n": 30,
   "metric": 0.0976,
   "ci_low": 0.0386,
   "ci_high": 0.2255,
   "secondary": {
    "precision": 0.0246,
    "recall": 0.0976,
    "coverage": 0.1333
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.0285,
     "usd_per_1k_correct": 50.89,
     "median_latency_s": 1.8
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "BlitzAPI",
   "n": 97,
   "metric": 0.0645,
   "ci_low": 0.0331,
   "ci_high": 0.1222,
   "secondary": {
    "precision": 0.0606,
    "recall": 0.0645,
    "coverage": 0.0825
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.0087467,
     "usd_per_1k_correct": 8.31,
     "median_latency_s": 1.5
    }
   ],
   "note": "ran on the 97 companies with a resolved LinkedIn URL"
  },
  {
   "vendor": "Findymail domain search",
   "n": 30,
   "metric": 0.0488,
   "ci_low": 0.0135,
   "ci_high": 0.1614,
   "secondary": {
    "precision": 0.6,
    "recall": 0.0488,
    "coverage": 0.0667
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.01518,
     "usd_per_1k_correct": 37.95,
     "median_latency_s": 9.1
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Fiber AI",
   "n": 30,
   "metric": 0.0488,
   "ci_low": 0.0135,
   "ci_high": 0.1614,
   "secondary": {
    "precision": 0.0493,
    "recall": 0.0488,
    "coverage": 0.0667
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.19,
     "usd_per_1k_correct": 407.14,
     "median_latency_s": 3.7
    }
   ],
   "note": "this endpoint filters by seniority and employer only, not by role; ran on the frozen 30-company subset"
  },
  {
   "vendor": "Apify Thirdwatch employees",
   "n": 30,
   "metric": 0.0244,
   "ci_low": 0.0043,
   "ci_high": 0.126,
   "secondary": {
    "precision": 0.225,
    "recall": 0.0244,
    "coverage": 0.0333
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Apify",
     "usd_per_call": 0.02791667,
     "usd_per_1k_correct": 46.53,
     "median_latency_s": 0.7
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Apify lead scraper",
   "n": 100,
   "metric": 0.0236,
   "ci_low": 0.0081,
   "ci_high": 0.0672,
   "secondary": {
    "precision": 0.2769,
    "recall": 0.0236,
    "coverage": 0.03
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Apify",
     "usd_per_call": 0.00911,
     "usd_per_1k_correct": 4.57,
     "median_latency_s": 0.8
    }
   ],
   "note": ""
  },
  {
   "vendor": "LeadMagic role finder",
   "n": 30,
   "metric": 0.0,
   "ci_low": 0.0,
   "ci_high": 0.0857,
   "secondary": {
    "precision": 0.9333,
    "recall": 0.0,
    "coverage": 0.0
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.03,
     "usd_per_1k_correct": 64.29,
     "median_latency_s": 2.5
    }
   ],
   "note": "ran on the frozen 30-company subset"
  },
  {
   "vendor": "Anymailfinder decision makers",
   "n": 30,
   "metric": 0.0,
   "ci_low": 0.0,
   "ci_high": 0.0857,
   "secondary": {
    "precision": 0.0,
    "recall": 0.0,
    "coverage": 0.0
   },
   "tie_group": 2,
   "stores": [
    {
     "store": "Treg",
     "usd_per_call": 0.02133333,
     "usd_per_1k_correct": null,
     "median_latency_s": 17.0
    }
   ],
   "note": "ran on the frozen 30-company subset"
  }
 ],
 "not_measured": [
  {
   "vendor": "Wiza",
   "reason": "refused 97 of 100 calls (\"size parameter not allowed\") while Treg's spec requires it."
  },
  {
   "vendor": "Apollo",
   "reason": "its free search hides last names, so answers can't be matched."
  },
  {
   "vendor": "Lemlist",
   "reason": "needs its own account."
  },
  {
   "vendor": "Muse Spark",
   "reason": "outside the data policy."
  },
  {
   "vendor": "ContactOut",
   "reason": "not bought on cost: 25\u00a2 to $3.50 a company."
  },
  {
   "vendor": "CompanyEnrich",
   "reason": "not bought on cost: 25\u00a2 to $3.50 a company."
  },
  {
   "vendor": "People Data Labs",
   "reason": "not bought on cost: 25\u00a2 to $3.50 a company."
  }
 ],
 "tests": "Holm-corrected exact paired McNemar against the leader, alpha 0.05, over the known people at the companies both took; a correct item is a known person found, the same exact test report.py pairs recall with.",
 "caveats": [
  "The key comes from one company's records, skewed to learning and development buyers: 100 companies holding 127 known L&D people.",
  "Precision is an estimate, not a count: a frozen sample of up to 100 people a route is judged by a model, and its unknown answers are left out of the rate.",
  "A company cannot be sampled: coverage counts the companies where a known person was found, over the companies asked.",
  "The mid tier and the role finders ran on the frozen 30 companies; the cheap tier ran on all 100.",
  "Wiza, Apollo, Lemlist, Muse Spark, ContactOut, CompanyEnrich and People Data Labs were not measured; see not_measured."
 ]
}