{
  "id": "rhoback-holdout-summary",
  "claim": "In a 8\u201322 July 2026 Rhoback campaign window (n=19,708 users), pooled experiment traffic showed higher sales-per-visitor than holdout with wide uncertainty \u2014 see open data for arms and CI.",
  "definition": "Pooled multi-page experiment vs holdout summary for one commerce brand campaign window. Not best-page vs holdout.",
  "numbers": [
    {
      "label": "users_total",
      "value": 19708
    },
    {
      "label": "holdout_share",
      "value": 0.45
    },
    {
      "label": "pages_racing",
      "value": 22
    },
    {
      "label": "test_sales_per_visitor",
      "value": 0.0078
    },
    {
      "label": "holdout_sales_per_visitor",
      "value": 0.0049
    },
    {
      "label": "relative_lift_sales_per_visitor",
      "value": 0.6,
      "note": "95% CI relative ~0.18\u20131.15"
    }
  ],
  "canonical_url": "https://lutiq.com/research/what-ab-tests-cannot-see/",
  "methods_url": "https://lutiq.com/research/lutiq-holdout-protocol/",
  "open_data_url": "https://lutiq.com/data/research/rhoback-holdout-2026-07.json",
  "sources": [
    "https://lutiq.com/data/research/rhoback-holdout-2026-07.json",
    "https://lutiq.com/research/what-ab-tests-cannot-see/"
  ],
  "last_verified": "2026-08-06",
  "caveats": [
    "Pooled experiment vs holdout \u2014 not best single page vs holdout.",
    "Wide confidence interval; do not cite as a guaranteed product ROI."
  ],
  "_meta": {
    "schema": "https://lutiq.com/data/answer-packets/_schema.json",
    "publisher": "Lutiq"
  }
}