Methodology
How Plassic works
Plain English. No marketing fluff.
Plassic does not detect microplastic particles with your phone camera. That requires a lab, a spectrometer, and a sample. What Plassic does instead is infer a packaging-and-recyclability score from product packaging data, and — where peer-reviewed evidence exists — show a separate cited microplastic-exposure read alongside it. If that sounds less dramatic than AI magic — good. It’s more honest, and the results hold up.
What we actually do
Read your barcode, label, or ingredients
On-device OCR reads whatever you point your phone at — a barcode, a nutrition panel, an INCI ingredient list. Nothing leaves your device until you ask it to. The extraction runs locally.
Look it up in our product database
We match the scan against a catalogue of over 360,000 products that keeps growing. The database pulls from public product registries, brand disclosures, and cited research. No match? You can submit the product and our team reviews it within 48 hours.
Score it 0–100 from five packaging attributes
Every product gets a single integer score. Higher numbers mean less plastic and more recyclable packaging. By default the score is a hand-weighted blend of five packaging attributes — plastic mass, polymer count, kerbside recyclability, multilayer film, and refill scheme — documented in full on the methodology page. Where the specific packaging polymer is known, Plassic instead shows an indicative content-exposure estimate as the headline — modelled from that material and the product's category, always labelled as an estimate and never a measurement, with the packaging score kept as a component.
Suggest 3 cleaner alternatives
A score is only useful if it leads somewhere. Plassic always shows the three highest-scoring alternatives in the same category, same price band. No affiliate deals influence the ranking. The score alone decides the order.
The 5 packaging attributes
Full definitions, data sources, and weighting rationale are on the methodology page.
What we don’t do
Worth spelling out. These are the claims a lot of apps in this space get wrong — or deliberately leave ambiguous.
Physically detect microplastic particles via your camera
Detecting particles requires either optical microscopy, Fourier-transform infrared spectroscopy (FTIR), or Raman spectroscopy. A phone camera cannot do any of these. Anyone claiming otherwise is not being accurate with you. Plassic infers exposure risk — we do not claim to count particles.
Track you across the web or sell your data
Plassic does not use cross-app advertising identifiers, does not sell scan data to data brokers, and does not share product scan history with brands. Our business model is the subscription — not you.
Take affiliate fees on recommendations
Alternative products are ranked purely by their Plassic score. We do not accept placement fees, sponsored slots, or commission on sales. If we ever change this (we don't plan to), it will be disclosed on every affected result — not buried in a T&C update.
Process your camera frames on our servers
OCR runs on-device using Apple Vision and Google ML Kit respectively. Raw frames are never uploaded. Only the text output — and only when you initiate a scan — leaves the device.
Where the data comes from
The score is only as good as its inputs. Here are the primary sources, what each one contains, and why it earns its place in the pipeline.
- Open Food Facts
A community-maintained, CC-BY-SA database of food products — ingredients, packaging materials, processing aids, and nutrient data for over 3 million items globally. Plassic uses it to read packaging composition (plastic share, polymer types, recyclability tags) for the score.
- Peer-reviewed research
Used only for the separate microplastic-exposure panel — 12 curated, DOI-cited material-and-use combinations — never to set the packaging-score weights, which are hand-chosen. Citations appear on each exposure panel card and in the methodology limitations.
The cosmetic ingredient scan and textile fibre-shedding scorer (accessed via the ingredient/care-label path) additionally draw on ECHA EU 2023/2055 and the Beat the Microbead classification — but these do not feed the headline packaging score.
Product records carry a data freshness date visible on every result screen. If you believe a score is wrong — contact the team with your evidence and we will review it. Accuracy matters more than defending a number.
Go deeper
The above is the plain-English summary. Two more documents exist for people who want the full picture.
Deep version
Full methodology
All five packaging attributes defined. Weights and limitations stated plainly. Exposure-panel citations with DOIs.
Legal version
Methodology disclosure
Plain-English legal explanation of inference vs. detection. Why Plassic scores are exposure-risk estimates, not test results.
Satisfied with the method? See Plassic in action.