Limitations

The boundaries of what FabricIQ measures and claims. Honesty here strengthens credibility — every transparent product on the market wins trust by being explicit about its limits.

This document is the public-facing companion to §10 of the methodology whitepaper. It is deliberately defensive in tone.

What we measure

Composition at the fiber-percentage level (parsed and normalized from product labels and brand websites).
Quality as a composite of 7 material-axis scores (durability, comfort, breathability, warmth, moisture wicking, sustainability, care ease) computed from a curated 206-material DB.
Health Impact as a composite of three sub-signals: microplastic shedding (per-fiber published rates), skin-contact synthetic load (continuous, weighted by category), and chemical treatments — both heuristic flags (PFAS, formaldehyde, antimony — composition × category thresholds) AND, since v0.38.0, observed finishes captured directly from the product page (PFAS, formaldehyde, biocide, flame retardant), which are stronger than heuristics and register as likely.
Sustainability as one of the 7 Quality axes, exposed independently for UI use.

All scores are deterministic — same inputs always produce the same outputs. Reproducible from any clean checkout.

What we explicitly do NOT measure

This is the important part.

1. We do not test individual SKUs in a lab

No specific garment has had its antimony, PFAS, formaldehyde, or any other chemical content lab-measured by FabricIQ.
Chemical-treatment flags (PFAS likely / formaldehyde possible / antimony trace) are heuristic based on industry-standard manufacturing practice for that composition × category combination.
A heuristic flag means "in our analysis, items matching this profile commonly carry this treatment in published industry practice." It does not mean any specific item has been tested or confirmed.
Some flags are now observed — the chemical finish (e.g. "wrinkle-free", "DWR", "antimicrobial") is stated on the brand's own product page. This is a stronger signal than the heuristic, but it is still "stated on the page," not lab-tested by us. A brand's claim that a finish is absent is also taken at face value (e.g. a "PFC-free" note suppresses the PFAS flag).
This is consistent with how all consumer-facing tools at this layer operate (Yuka, Oasis, Good On You) and is documented in §10 of the methodology whitepaper.

2. We do not have factory- or mill-level data

Same product from the same brand with the same composition gets the same score regardless of which factory made it.
This is the boundary that separates our POD-founder audience use case from the "real fashion startup" use case where mill provenance matters.
Closing this gap requires partnerships with mills, supplier databases, or licensed feeds — a growth investment, not a current capability.

3. We do not measure dye chemistry

Bright neon synthetics may use azo dyes that can degrade to aromatic amines (a regulatory concern), but we do not currently flag this.
Future addition: a heuristic flag based on (color × material × dye-class) inference would be valuable for the chemical-treatments layer.

4. We do not score garment-construction quality

Stitch count, seam type, button quality, hardware durability, and fit consistency are out of scope.
Our Quality grade reflects the fabric's intrinsic properties — not how well the garment was constructed.
A poorly-constructed item from a great fabric can still earn a high Quality grade.

5. The Health Impact score is not medical advice

It is a heuristic-based exposure-likelihood signal, not a diagnostic tool.
Every Health Impact UI surface includes the disclaimer "Not medical advice — based on material composition."
A consumer with sensitivities should consult an allergist.
A brand with regulatory questions should consult its compliance counsel and a chemical-testing lab.

6. Microplastic shedding rates are estimates, not measurements

Real-world shedding varies by detergent, water temperature, machine type, agitation, garment age, and washing frequency.
We use mid-range published estimates from peer-reviewed laundry literature (see §1 of the research bibliography).
A specific 6 kg load may release more or fewer fibers than our estimate; we are calibrated to be informative, not predictive at the wash-level granularity.

7. Brand-level scores require sample size

Brand averages with fewer than 30 analyzed products show a "Limited data" warning on every public scorecard and dashboard.
Below that threshold, the scores may not be representative of the brand's full catalog.
This applies particularly to small brands, blank manufacturers with niche specialty SKUs, and any brand whose product catalog isn't well-represented on a publicly-scrapeable site.

8. Recycled and organic claims are sourced from labels

We trust the label. If a label says "Recycled Polyester," we score it as recycled polyester.
We do not currently verify recycled or organic claims against certification documents (GRS, GOTS).
This is a known gap — verification via certification database integration is on the roadmap.
Mitigation: the Label Confidence engine flags recycled/organic claims without explicit certification mention as moderate confidence. Surfaces on every product page as a "Label Confidence" card. See also the dedicated topic page at /health/label-trust.

8a. Composition labels can lie

Independent fiber-content testing routinely finds more synthetic content than labels claim. The 2020 Indian organic-cotton fraud investigation removed GOTS certificates from suppliers covering roughly half of the global certified organic cotton supply — meaning products labeled "100% Organic Cotton" during that period often were not. "Natural" outerwear — jackets, suits, dress shirts — frequently has undisclosed synthetic linings, PFAS-treated DWR coatings, or formaldehyde resin finishes that the composition label does not mention.

This is the most fundamental limitation in the entire system: a score is only as honest as the label it was computed from. We cannot lab-test garments. We cannot audit factories. We trust labels by default.

What we can do, and now do, is surface the trust signal:

Vague composition language ("eco-friendly fabric", "recycled materials") triggers a low Label Confidence flag
"100% Cotton" on a jacket triggers a low flag (lining is almost certainly undisclosed)
Organic cotton claims without certification trigger moderate
Multi-section compositions where percentages don't sum to ~100% trigger moderate (lining undisclosed)
Marketing prose where structured composition data should be triggers moderate

The Label Confidence card surfaces alongside Health Impact and is documented at /health/label-trust. It does not change the underlying scores; it tells the reader how skeptically to read them.

9. We are not a certification body

We do not award OEKO-TEX, GOTS, GRS, Bluesign, Cradle to Cradle, or any other third-party seal.
Our scores are independent assessments and should not be confused with certification.
We respect existing certifications and may extend coverage to mark certified items in our catalog at some point.

10. Coverage is not comprehensive

We cover a curated subset of brands and products — not the entire global apparel market.
Brands in our database are primarily mid-to-large US, EU, and Japanese apparel companies plus 42 blank manufacturers with public catalogs.
Many smaller / regional / emerging brands are not covered.
Adding a brand requires a custom scraper and ~1 week of curation work.

What this means in practice

For shoppers

Treat Health Impact as a directional signal — "all else equal, lean toward the lower-impact alternative" — not a diagnosis.
A Health A or B grade is not a guarantee of safety; an F is not a guarantee of harm. Ask a doctor for actual medical assessment.
Quality grade A or B reflects the fabric, not whether the seams will hold up. Read reviews.

For POD founders

/blanks data is fabric-level. The same SKU could be sewn at different factories with different production-quality outcomes.
Use our data to make the fabric choice; verify the manufacturer separately (Printful's vendor reviews, MOQ tests, sample orders).
Plastic-Free filter is composition-based and reliable.

For brand sustainability leads

Risk Briefing flags are heuristic. They are a starting point for internal compliance audits, not a substitute for them.
For EU REACH PFAS-inventory work, our SKU counts indicate scope — formal compliance requires lab testing on production samples plus legal counsel.
The Pro dashboard accelerates the audit; it does not replace the audit.

For partners and integrators

API consumers see the same Health Impact heuristics with the same caveats. Anyone embedding FabricIQ scores into a product surface should preserve the disclaimers.
For commercial redistribution, see API terms.

Why we publish this document

Most rating tools obscure their limits. We document them prominently because:

Trust compounds. A user who can find our limitations document trusts the scores more, not less.
It defines the upgrade path. Every limitation listed above is a future investment opportunity (mill-level data, dye chemistry, construction quality, certification verification) that an investor or partner can underwrite.
Regulatory environments reward transparency. EU REACH requirements lean toward "show your work." A published limits document is consistent with that direction.
It positions us against opaque competitors. Higg/Cascale's credibility damage came from opacity. We are explicit by design.

How this document is maintained

When the engine adds a new measurement (e.g., dye chemistry flagging), the limit gets removed here and the methodology is added to the methodology whitepaper.

When a new boundary is discovered (e.g., a category formaldehyde flags miss), it gets added here.

Updated alongside the methodology in every release that touches the scoring engine.