{
  "title": "Carpet and upholstery cleaning before-and-after data for AI",
  "source_page": "https://fiund.com/industries/carpet-cleaning-swatches-data",
  "authorship": "Original fictional Fiund editorial exercise. No actual operating records or third-party source assets are included.",
  "example": {
    "title": "A changed exposure can imitate an appearance improvement in a cleaning comparison",
    "context": "Two owned textile swatches use original image-review annotations. This text fixture contains no actual photos, customer furnishings, cleaning instructions or performance certification.",
    "records": [
      [
        "CW1",
        "Swatch A pair",
        "Same swatch ID, angle, crop, light and exposure before/after. Reviewer observes a smaller visible dark region after the staged task."
      ],
      [
        "CW2",
        "Swatch B pair",
        "Same swatch ID but after-view exposure increased. Dark region appears lighter; no exposure-matched recapture supplied."
      ],
      [
        "CW3",
        "Reference map",
        "A has a preexisting small dot outside the task region. Map identifies that dot in both views; it is not a new mark."
      ]
    ],
    "question": "Which pair supports a controlled appearance comparison, and what claims remain unsupported for B or cleaning effectiveness?",
    "answer": "CW1 supplies a matched A comparison with a reviewer-observed reduction in visible dark area under the stated capture controls. CW2 changes exposure, so its lighter appearance cannot be separated from the imaging change without recapture or another justified control. CW3 identifies A's dot as preexisting, not new damage. These observations evaluate image evidence and do not by themselves establish cleaning effectiveness, a cause or a certification outcome.",
    "failureModes": [
      "Attributing brightness changes to cleaning when exposure differs.",
      "Calling a preexisting reference mark new or treating appearance evidence as certified performance."
    ],
    "fields": [
      [
        "Swatch and task-region identity",
        "Defines the comparison target."
      ],
      [
        "Angle, crop, lighting and exposure",
        "Records capture controls."
      ],
      [
        "Reviewer change and preexisting marks",
        "Separates observations from causal claims."
      ]
    ],
    "holdout": "Provide B's exposure-matched recapture from a different angle. Require the remaining geometry difference to be considered before a controlled comparison is claimed.",
    "boundary": "Use business-owned swatches and authorized equipment. Exclude customer property, interiors and trademarks; retain photo authorship and reuse rights."
  }
}