Document 02 · Fabric Composition Score · v0.1 (MVP)

A composition score,
and only that.

Threadwise reads the fiber content on the label and nothing else, then returns a 0–100 score and a calm recommendation a shopper can read in the time it takes to check a price tag. Naming it precisely — a Fabric Composition Score — is the whole game.

Visual conversion of the MVP Fabric Composition Scoring logic · Threadwise v0.1 · July 2026

◆ Based on fabric label only This is a Fabric Composition Score, not a sustainability score and not a quality score. It uses only what is printed on the fiber-content label. Every screen, label, and line of copy calls it exactly what it is — that discipline is what makes it defensible.

01Input → output model

One clean pipeline: read the label, score each fiber, average by weight, apply at most two small adjustments, and return a score with a band, a recommendation, and a three-line explanation.

Input

Fiber-content label
Which fibers
Percentage of each
Natural / regenerated / synthetic
How many types blended

Engine

Base score per fiber (0–100)
Weighted average by %
− 5 mixed materials (if triggered)
− 5 complex blend (if triggered)
Max total adjustment: −10

Output

Score, 0–100
Band (one of five)
Recommendation label
3-line explanation
“What this can’t see” line

Nothing else touches the number. The score is a starting signal, not a final verdict on the garment.

02What the score can & can’t see

Stated plainly on the result screen, so the shopper is never misled. The exclusion is a visible feature, not a hidden gap.

✓ Can evaluate — from the label

Which fibers the garment is made from
The percentage of each fiber
Natural, regenerated, or synthetic type
How many fiber types are blended
Whether small synthetic percentages are present
The broad sustainability & longevity signal the fiber type suggests

✕ Cannot evaluate

Construction & stitching quality
Actual garment lifespan (fiber type is a weak predictor)
Price or value
Brand ethics or labor practices
Certifications (organic, recycled-content, etc.)
Dyeing & finishing process; the specific factory
Weight / construction (knit vs woven); how it’s worn & washed

◆ Based on fabric label only The result screen carries a short “what this score can’t see” line at all times. It usually points at construction, fit, and how much you’ll actually wear the garment.

03Fiber category map

Each fiber gets one base score, 0–100, higher is better — ordinal judgments about production footprint and end of life, set in steps of 5 to avoid pretending they are precise.

Strong naturallow-input, biodegrade cleanly

Linen90

Hemp90

Organic cotton85

Acceptable naturalbiodegradable, heavier footprint

Cotton (conventional)75

Wool75

Silk75

Regeneratedcellulosic; process varies

Lyocell / Tencel80

Modal65

Viscose / rayon55

Syntheticsfossil-based, persistent

Recycled polyester45

Polyester35

Nylon30

Acrylic25

Elastane / spandex20

Nothing reaches 100 — the empty 90–100 space is deliberate headroom, a signal that the score means “strong,” not “flawless.”

◆ The category name is not the ranking Lyocell (regenerated, 80) scores above conventional cotton (natural, 75) on purpose. The process behind the fiber is the ranking, not its family — the model resists the lazy “natural good, synthetic bad” shortcut.

04Composition rules

The garment score is the weighted average of fiber base scores, plus at most two small adjustments. To apply them, sort every fiber into one of two families.

Step 1 — Weighted average. Multiply each fiber’s base score by its percentage, add them up.
Step 2 — Up to two adjustments. Maximum total −10. Round to a whole number.

Biodegradable family

All natural fibers (cotton, linen, hemp, wool, silk) and all regenerated cellulosics (lyocell, modal, viscose).

Synthetic family

Polyester, recycled polyester, nylon, acrylic, elastane.

Rule	Trigger	Adjustment
Mixed materials	Both families present, each at 10% or more	−5
Complex blend	Three or more fiber types, each at 5% or more	−5

◆ Trace synthetic (1–5%) No penalty — the amount is too small to change which recycling stream the garment could enter — but a specific line in the explanation says so: “the small stretch/synthetic content means this can’t be recycled as a natural fiber.” Honesty lives in the copy, not in a fiddly penalty.

05Score bands

Five bands map the 0–100 score to a plain label. Each band is legible by its range and word alone — never by color only.

85–100

Great fabric choiceBand 1 · the strongest signal

70–84

Good fabric choiceBand 2 · solid, one or two things to check

50–69

Okay, with tradeoffsBand 3 · works, with a tradeoff worth knowing

30–49

Think twiceBand 4 · real tradeoffs; fine if you’ll wear it a lot

0–29

Weak fabric choiceBand 5 · working against you on fabric alone

The scale runs high (best) to low. Range and label carry the meaning; color reinforces but never stands alone.

06Recommendation labels

Calm, plain, and free of guilt. The label is the verdict; the shopper decides what to do with it. No language blames the shopper.

Band	Recommendation	One-line tone
Great	Go ahead	“A strong fabric. Little to watch here.”
Good	Solid pick	“A good choice. One or two things to check.”
Okay	Fine, with tradeoffs	“Works, but there’s a tradeoff worth knowing.”
Think twice	Worth a second look	“Some real tradeoffs here. Fine if you’ll wear it a lot.”
Weak	Probably skip	“This one’s working against you on fabric alone.”

Owned or thrifted garments are always best worn long, whatever the fabric scores — and the copy never implies otherwise.

07Explanation pattern

Every result shows the same three short lines, in this order. It keeps the tone consistent and makes “check yourself” a permanent, honest reminder.

Line 1

What’s good about this fabric — the strength that earned its score.

Line 2

The tradeoff — the honest cost, in one sentence.

Line 3

What to check yourself — usually construction, fit, and how much you’ll wear it.

The same pattern on every screen. Line 3 is where everything the score can’t see is named.

08Worked examples

All eight, fully worked, so any can be recomputed by hand. Distribution: 1 Great, 2 Good, 1 Okay, 4 Think twice — an honest skew that reflects what mass-market labels actually say.

#	Composition	Calculation	Score	Band → recommendation
1	100% cotton	1.00 × 75	75	Good → Solid pick
2	95% cotton, 5% elastane	71.25 + 1.0 = 72.25	72	Good → Solid pick
3	80% cotton, 20% polyester	60 + 7 = 67, −5	62	Okay → Fine, with tradeoffs
4	60% polyester, 40% cotton	21 + 30 = 51, −5	46	Think twice → Worth a second look
5	100% polyester	1.00 × 35	35	Think twice → Worth a second look
6	70% viscose, 30% nylon	38.5 + 9 = 47.5, −5	43	Think twice → Worth a second look
7	50% wool, 50% acrylic	37.5 + 12.5 = 50, −5	45	Think twice → Worth a second look
8	100% linen	1.00 × 90	90	Great → Go ahead

❄ Frozen · deliberate consequence Raising cotton to 75 pulls the 95/5 cotton-elastane tee from “Okay” (68) up to “Good” (72). That is a real consequence, not a rounding artifact: in a composition-only model, a 95% cotton garment genuinely is a good fabric, and the elastane’s cost is carried by the tradeoff line, not a heavy penalty. The only clean lever to reverse it is holding cotton at 70 — cotton off the boundary, or a firmer trace-elastane signal. The former was chosen, and it is reversible.

092–5 garment comparison ranking

How a set of already-scored garments would be ranked when a shopper compares several similar items. This is a Phase 2 surface — shown here so the ranking logic is understood, not because it is in the MVP.

◷ Parked for v2 · Phase 2 Comparison is designed and ready, but it sits outside the composition-only MVP. The single-item score is the v1 deliverable.

1 · Collect

2–5 scored items
Session-only tray
Opt-in “Add to compare”

2 · Line up & mark

One column per garment
Compare row by row
Mark the best value in each row

3 · Name the set

Highest score = “best of the set”
A gentle line, no shaming
All wearable; one is lowest impact

The comparison decides, it does not only display — but it names a winner without shaming the others.

Example set

Garment	Composition	Score	Band
Linen tee Best of set	100% linen	90	Great
Cotton tee	100% cotton	75	Good
Cotton-poly tee	80% cotton, 20% polyester	62	Okay

Reads as: “All three are wearable; the linen tee is the lowest impact of the set.”

10Frozen MVP decision

The scope call that keeps this model simple to build and honest about its limits.

❄ Frozen · MVP decision For v0.1, Threadwise scores garments on fabric composition only: one base score per fiber, a weighted average, and two small adjustments (mixed materials, complex blend). Nothing else.

✓ Shipped in v0.1

One base score per fiber
Weighted average by percentage
Mixed-materials adjustment (−5)
Complex-blend adjustment (−5)
Honest naming + “what to check yourself” line

✕ Parked for v2 (in Scoring Model V2)

Durability signals & care burden
Construction quality
Certifications
Lifecycle data
Brand & production information
The full four-pillar weighting

◷ Parked for v2 · revisit when Hi-fi user testing shows shoppers want more than a fabric signal, or garment-level inputs (construction, weight, verified certifications) become reliably available. The tradeoff accepted: the score can’t see durability — made safe by honest naming and the permanent “what to check yourself” line.

↑ Back to top