Scoring rubric

What the score means

Every product is broken into named parts (body, lid, handle, gasket…). Each part is in exactly one of four states:

• non-plastic — captured evidence confirms a non-plastic material.
• plastic — captured evidence confirms plastic.
• silicone — captured evidence confirms silicone or another elastomer. Counted at half penalty: neither plastic nor fully plastic-free.
• unknown — no evidence either way. Unknown is a first-class value, not a gap.

Score = percent of parts confirmed non-plastic (0–100), with silicone parts earning half credit. Confidence = percent of parts that are not unknown, capped at 80% when every evidence row comes from a listing or brand page (no independent verification such as a review, teardown, or user photo). The score is never displayed without its confidence.

Rules we never break

• We score only from captured evidence. We never infer or guess a material from photos, price, or brand reputation.
• Unknown parts count against a product: they can never raise a score, and each product page says why.
• “BPA free” is treated as a strong flag that plastic is present — not as a plastic-free signal.
• Every score traces to at least one evidence row: an exact quote and a link to its source. Evidence is append-only.
• We state observed evidence, not verdicts. “Listing says PP lid” — never “this product is toxic.”
• We make no health claims about microplastics. For the research, see sources such as PubMed Central.

Where the data comes from

Retailer-first: the listing is the unit of record. Evidence can come from listings, reviews, brand pages, teardowns, user photos, or manual verification — each evidence row records its source type and URL. Manufacturer data comes at a later stage.