SmartSize · Internal

Restricted area

Enter the access code to continue.

Incorrect code. Try again.
SmartSize · Algorithm explainer

How SmartSize picks the right size for your shoppers

This page walks through what happens behind the scenes every time a shopper uses your size quiz. It's written in plain English — no formulas — so you can tune your size charts, fit settings, and priorities with confidence.

1 · The big picture

Three things in, one recommendation out

Every recommendation is the result of mixing what you configured for the garment with the body measurements the shopper typed in.

You provide
Garment setup
Size chart, how roomy the cut is, how stretchy the fabric is, and which measurements matter most.
Shopper provides
Body measurements
Bust, waist, hip — whatever you ask for in your quiz, in cm or inches.
SmartSize returns
A recommended size
With a confidence level, a strength rating, and a short shopper-friendly explanation.
No dead ends
SmartSize always picks a size for a person. If nothing fits perfectly, it picks the closest match and clearly labels the recommendation as cautious or speculative so the shopper sees the uncertainty. (Pets are different — see §7.)
2 · The 4 things you control

Your four levers

These are the only inputs you set as a merchant. Once they're configured, the same logic runs for every shopper.

① Size chart

A min and max for each measurement, per size.
Example — bust column
S: 82–86 cm  ·  M: 86–90 cm  ·  L: 90–94 cm

② Fit ease

How roomy the garment is designed to be, in cm.
Example
A bodycon dress sits at −8 cm (compression). A loose hoodie sits at +25 cm (oversized).

③ Fabric stretch

How many extra cm the fabric can give before it pinches.
Example
A cotton shirt is 0 cm. A 4-way stretch legging is 14 cm.

④ Measurement priorities

Which body measurements matter most for this garment.
Example — skinny jeans
Waist HIGH, Hips HIGH, Inseam HIGH, Bust NONE.

Fit ease at a glance

Fit ease is the most important dial. It tells SmartSize whether the cut is tight, regular, or roomy by design — so it knows how to interpret the chart's min/max numbers.

Compression−8 cm · shapewear, bandage
Slim+5 cm · skinny jeans, bodycon
Fitted+9 cm · tailored shirts
Regular+11.5 cm · standard tees
Relaxed / Oversized+16 to +25 cm

Priority weights — how it works

You assign each measurement a priority: NONE, LOW, NORMAL, or HIGH. Those become weights of 0 / 1 / 2 / 3 and are scaled to add up to 100%.

Example — a tailored shirt: bust HIGH, waist NORMAL, hips LOW, inseam NONE.

Bust HIGH
50%
Waist NORMAL
33%
Hips LOW
17%
Inseam NONE
0%
Rule of thumb
A bust score on a tailored shirt counts three times more than a hip score. Set priorities the way you think about fitting the garment.
3 · Scoring a single measurement

Where does the shopper's number land?

For each size, SmartSize takes one measurement at a time and asks: "where on the spectrum does the shopper's number fall, relative to this size's min and max?"

Imagine the size's range as a green zone in the middle, with progressively worse zones on either side:

min
max
Too looseway under min
Loosenoticeably big
Relaxeda bit big
Perfect fitinside the range
Snuga bit tight
Tightnoticeably tight
Too tightway over max

Two important asymmetries

  • Loose forgiveness depends on fit ease. A roomy garment is allowed to sit looser on the body before it starts losing points. An oversized hoodie won't penalise a slim shopper much; a bodycon dress will.
  • Tight forgiveness depends on stretch. A stretchy fabric is allowed to sit a bit over the max before it starts losing points. A non-stretch poplin won't get the same allowance.
Length measurements skip the stretch bonus
Fabric usually stretches sideways, not lengthwise. So inseam, arm length, and total height get the fit-ease allowance but not the stretch allowance. A stretchy legging won't be forgiven for being too short.
4 · Combining measurements

One weighted score per size

Each measurement gets a score from the previous step. SmartSize then averages them using the priority weights you set, producing one final score per size.

Example — same tailored shirt setup as before:

MeasurementWeightScore for size MContribution
Bust (HIGH)50%Perfect · 1.00+0.50
Waist (NORMAL)33%Perfect · 1.00+0.33
Hips (LOW)17%Snug · 0.40+0.07
Total for size M+0.90

The same calculation runs for every size in the chart. The highest total wins — unless the size has been ruled out (see §5) or there's a tie (see §6).

Why weighted averages?
A perfect bust on a shirt should matter more than a perfect hip. Weights let the algorithm respect your judgement about what's important for each garment.
5 · When a size is ruled out

Disqualification (the "won't fit" gate)

A great score on bust doesn't save a size if the waist is hopelessly tight. SmartSize ejects sizes from consideration when any important measurement is too far outside the range.

The rules in plain English

  • HIGH priority measurements are strictly judged. If the shopper's value is more than the tolerance over the size's max (or under the min), the size is out.
  • NORMAL priority measurements get a small extra buffer — 3 cm for adults, 2 cm for kids — before they trigger disqualification.
  • LOW and NONE priority measurements never disqualify a size on their own. They only nudge the score.

Garment-type extras

Picking an archetype (like skinny jeans or bodycon dress) turns on a few extra guardrails:

Garment typeWhat also disqualifies a size
Bottoms (jeans, trousers)Hip is way too tight, or waist is bigger than hip's best match.
One-piece (dresses, swimwear)Any critical girth (bust, waist, or hip) is too tight, even by a smaller margin.
Tops, outerwear, generalNo extra rules — just the standard ones above.
Below the whole chart? We back off.
If the shopper's measurement is smaller than even the smallest size's min, we don't disqualify every size for being "too loose." Instead, every size remains a candidate (with reduced score) and the shopper is shown an "off-chart, below" warning. This avoids the absurd outcome of recommending nothing to a petite shopper.
6 · Picking the winner

The highest score wins — usually

Among the sizes that survived disqualification, the one with the highest total wins. When two are basically tied (within 0.15 points), we have to pick a direction.

Customer typeTie-break directionWhy
Adults (man, woman, unisex)Larger sizeEasier to alter down than up; more comfortable to wear.
Children (boy, girl)Always largerKids grow; sleeves can be rolled.
Adults in compression/slim + stretchy garmentsSmaller sizeThat's the design intent — bandage dresses, leggings, swimwear hug the body.

What "tied" really means

"Tied" here means the two top sizes scored within 0.15 points of each other on the −1.0-to-+1.0 scale. Anything bigger than that and we just pick the higher score outright.

7 · Special cases

When the rules bend

Three situations get bespoke treatment. They're worth understanding because they explain why two shoppers with the same numbers may get different sizes.

Kids: growth bias

For boys and girls, the next-larger size gets a small bonus (about 20% of the score scale, dialled down for narrow charts). If a slightly larger size is otherwise close in score, it wins. Sleeves get rolled; jeans get hemmed.

Slim + stretchy

When the garment is cut slim or compression and the fabric is stretchy, tied scores go to the smaller size. This is the design intent for bodycon dresses, sports bras, swim one-pieces, and leggings. You'll see a special shopper banner explaining this.

Pets: pass/fail

For dogs and cats, no scoring or weighted averages. Each measurement either falls inside a size's range or it doesn't. If every measurement agrees on a size, we recommend it; if they disagree wildly, we say so instead of guessing.

8 · Confidence & strength

Two labels you'll see in the popup

Every recommendation comes with a confidence rating (how clear the data was) and a strength rating (how well the size actually fits). They are independent — high confidence in a so-so size is possible.

Confidence

How well the shopper's measurements agreed on one size.

highEvery measurement points at the same size.
mediumMeasurements split across two adjacent sizes.
lowMeasurements span 3+ sizes or some are missing.

Strength

How well the chosen size actually fits the shopper.

strongEverything falls inside the chart.
moderateSlightly off-chart, on the loose side.
cautiousOff-chart with at least one tight area.
speculativeWay off-chart, or no size really fits.
What this means for shopper copy
The popup wording adapts to these ratings. A strong recommendation reads confident ("We recommend M"); a speculative one is honest ("Closest match — measurements are outside our chart").
9 · A full worked example

Maria orders a tailored shirt

Putting it all together with a single, end-to-end story.

Setup

Garment: tailored shirt  ·  Fit: +9 cm (Fitted)  ·  Stretch: 0 cm (Non-stretch)  ·  Priorities: Bust HIGH, Waist NORMAL, Hips LOW, Inseam NONE.

SizeBust (cm)Waist (cm)Hips (cm)
S82 – 8664 – 6890 – 94
M86 – 9068 – 7294 – 98
L90 – 9472 – 7698 – 102

Maria's measurements: bust 88, waist 70, hips 96.

1

Compute priority weights. Bust HIGH(3) + Waist NORMAL(2) + Hips LOW(1) + Inseam NONE(0) = 6. So bust = 50%, waist = 33%, hips = 17%, inseam ignored.

2

Score each size. Maria's numbers slot neatly into M's ranges, so every measurement is a perfect fit for M. For S, every measurement is over the max — way too tight.

SizeBust 88Waist 70Hips 96Total
SToo tightToo tightToo tight−1.43
MPerfectPerfectPerfect+1.00
LRelaxedRelaxedRelaxed−0.18
3

Check disqualifications. Size S fails because Maria's bust (HIGH priority) is 2 cm over the max, with zero stretch and zero fit tolerance to absorb it. S disqualified

4

Pick the winner. M's score of +1.00 beats L's −0.18 by far more than 0.15 — no tie-break needed. M wins

5

Set confidence & strength. All three measurements landed in M's ranges, so: confidence: high strength: strong

The shopper sees: "We recommend M — fits comfortably."

A trickier case — Maria's stretchy bodycon dress

Same Maria (bust 88, waist 70, hips 96), but now shopping for a bodycon dress cut at +5 cm with moderate stretch (7 cm). The dress is rated bust HIGH, waist HIGH, hips HIGH.

  • The 7 cm of stretch and slim cut activates the slim+stretchy rule: tied sizes go to the smaller one.
  • If Maria scores roughly the same on size M and L, the smaller (M) wins — that's the design intent of a body-hugging dress.
  • The shopper sees a banner: "Bodycon dresses are designed to hug — we picked the smaller of two close matches."