{
 "name": "How SG&A Benchmarking numbers are produced",
 "url": "https://sgabenchmarking.com/methodology.json",
 "taxonomy": "https://sgabenchmarking.com/taxonomy.json",
 "products": {
  "industryFunctionBenchmarks": {
   "what": "Share of core SG&A headcount by function across a named cohort, one page per industry.",
   "how": [
    {
     "step": "Collect real titles",
     "detail": "We pull each company’s public workforce signals: department tallies and 5,962 actual job titles across Finance, Operations and the C-suite, the same way for all seven companies."
    },
    {
     "step": "Classify every title",
     "detail": "Each title is classified into the ten-function SG&A map . This is where 34% of department labels move: procurement buyers filed under “Finance,” whole Sourcing and Facilities teams hiding inside “Operations.”"
    },
    {
     "step": "Count only what we can see",
     "detail": "Levels are the classified contacts we actually observed, with no modeled multipliers. That makes them a conservative floor, works identically for public and private companies, and keeps every number traceable to real titles. The census closes the gap to a verified total."
    }
   ],
   "basis": "Estimate from public signals: department tallies corrected by individually classified job titles, capped at each company's social-profile-visible workforce. Levels are observed floors; the function MIX is the like-for-like read.",
   "notPublished": [
    "named-profile verification of every count (these are sampled-title splits)",
    "off-platform gross-up and confidence per cell",
    "sub-function detail (which Finance: FP&A vs controllership vs tax)",
    "subsidiary / brand-entity coverage"
   ],
   "perIndustryNotes": {
    "building-products": {
     "provenance": "ESTIMATED — title-refined: ZI department tallies corrected by Haiku-classified job titles (sampled), anchored to public headcount and censused-industry ratios. Not a census.",
     "methodNote": "Estimate from observed public signals: real job titles, classified one by one, split the department buckets — recovering Sourcing and Facilities, removing procurement and plant roles mislabeled as Finance. Levels are the classified contacts we can actually observe (a floor — coverage varies by company); the function mix is the comparable read. A census verifies every person, adds subsidiaries and the off-platform population.",
     "asOf": "2026-07-15",
     "titlesClassified": 7169
    },
    "tools": {
     "provenance": "ESTIMATED — title-refined: ZI department tallies corrected by Haiku-classified job titles (sampled), anchored to public headcount and censused-industry ratios. Not a census.",
     "methodNote": "Estimate from observed public signals: real job titles, classified one by one, split the department buckets — recovering Sourcing and Facilities, removing procurement and plant roles mislabeled as Finance. Levels are the classified contacts we can actually observe (a floor — coverage varies by company); the function mix is the comparable read. A census verifies every person, adds subsidiaries and the off-platform population.",
     "asOf": "2026-07-16",
     "titlesClassified": 5962
    },
    "cpg": {
     "provenance": "ESTIMATED — title-refined: ZI department tallies corrected by Haiku-classified job titles (sampled), anchored to public headcount and censused-industry ratios. Not a census.",
     "methodNote": "Estimate from observed public signals: real job titles, classified one by one, split the department buckets — recovering Sourcing and Facilities, removing procurement and plant roles mislabeled as Finance. Levels are the classified contacts we can actually observe (a floor — coverage varies by company); the function mix is the comparable read. A census verifies every person, adds subsidiaries and the off-platform population.",
     "asOf": "2026-07-16",
     "titlesClassified": 16498
    },
    "toys": {
     "provenance": "ESTIMATED — title-refined: ZI department tallies corrected by Haiku-classified job titles (sampled), anchored to public headcount and censused-industry ratios. Not a census.",
     "methodNote": "Estimate from observed public signals: real job titles, classified one by one, split the department buckets — recovering Sourcing and Facilities, removing procurement and plant roles mislabeled as Finance. Levels are the classified contacts we can actually observe (a floor — coverage varies by company); the function mix is the comparable read. A census verifies every person, adds subsidiaries and the off-platform population.",
     "asOf": "2026-08-11",
     "titlesClassified": 15736
    },
    "apparel": {
     "provenance": "ESTIMATED — title-refined: ZI department tallies corrected by Haiku-classified job titles (sampled), anchored to public headcount and censused-industry ratios. Not a census.",
     "methodNote": "Estimate from observed public signals: real job titles, classified one by one, split the department buckets — recovering Sourcing and Facilities, removing procurement and plant roles mislabeled as Finance. Levels are the classified contacts we can actually observe (a floor — coverage varies by company); the function mix is the comparable read. A census verifies every person, adds subsidiaries and the off-platform population.",
     "asOf": "2026-08-18",
     "titlesClassified": 11845
    },
    "saas": {
     "provenance": "ESTIMATED — title-refined: ZI department tallies corrected by Haiku-classified job titles (sampled), anchored to public headcount and censused-industry ratios. Not a census.",
     "methodNote": "Estimate from observed public signals: real job titles, classified one by one, split the department buckets — recovering Sourcing and Facilities, removing procurement and plant roles mislabeled as Finance. Levels are the classified contacts we can actually observe (a floor — coverage varies by company); the function mix is the comparable read. A census verifies every person, adds subsidiaries and the off-platform population.",
     "asOf": "2026-08-27",
     "titlesClassified": 8383
    }
   }
  },
  "companyBriefings": {
   "what": "One public company's SG&A as a percent of revenue on a normalized basis against its US-GAAP peer set, the disclosed lines held outside it, the gap to the peer median in dollars, and a TSR driver bridge.",
   "basis": "10-K and 10-Q filings and XBRL company facts. Normalized SG&A = reported SG&A less disclosed advertising, R&D, pension, stock-based compensation, restructuring and one-time items, plus logistics reclassed where filers park it in SG&A. Every adjustment is traceable to the filing.",
   "index": "https://sgabenchmarking.com/reports.html",
   "perCompanyJson": "https://sgabenchmarking.com/r/{TICKER}.json"
  },
  "census": {
   "what": "The paid engagement: every person at the company and its chosen peers classified to a function and sub-function from named public profiles, subsidiaries and off-platform population included, with spans, layers, geography and tenure.",
   "accuracy": "Typically within 5 to 8 percent of true headcount per function.",
   "pricing": "https://sgabenchmarking.com/index.html#pricing",
   "request": "https://sgabenchmarking.com/request"
  }
 },
 "limits": [
  "Industry-page levels are observed floors from public data and understate true headcount by a company-specific amount; compare the mix, not the absolute counts, across companies.",
  "Public-profile coverage varies by company and skews North American for globally distributed companies.",
  "Per-$B figures are shown only where filed revenue exists; private companies show no per-$B value.",
  "Some companies carry structural artefacts named in the page footnotes (for example a franchise network that public records cannot separate from employees).",
  "Company briefings use the filer's own disclosures; where a line such as advertising is not disclosed it is absent, not zero.",
  "Nothing here is audited. Figures are directional outside-in estimates."
 ],
 "citation": "Cite as: SG&A Benchmarking (sgabenchmarking.com), page URL and as-of date.",
 "contact": "kyle@sgabenchmarking.com"
}