New research guide: The Confidence Economy, how AI is changing fashion commerce.

Read it →

Read it →

Home

/

Features

/

Style Graph

[ 05 ] · Style Graph

A stylist that remembers. 

A stylist that remembers. 

A stylist that remembers. 

Every piece mapped to every other — colour, silhouette, occasion, season. Every skip, save and try-on moves the shopper’s position on it.

Start 30-day free trial
Experience Elara

No credit card · Live in under an hour

Save · silk

[ 01 ]

in one sentence

What is the Style Graph?

The Style Graph is Elara’s model of your catalog and your shoppers’ taste. It maps every product to every other by colour, silhouette, fabric, occasion and season — built from your catalog on the first sync, not from traffic — and then adjusts weights per shopper from what they skip, save, try on and buy. It is why a 200-product store gets a stylist that knows every piece on day one, and why a returning shopper is not treated like a stranger.

Also called · AI taste model · personalisation graph · catalog knowledge graph

In the glossary →

[ 02 ]

the gap

Why does it matter?

01

Your best customer gets the same store as a stranger.

02

Recommendations are built from what sold, not from what she skipped.

03

Taste is a pattern. Nothing on your site is watching for it.

[ 03 ]

how it works

How does the Style Graph work?

  1. 01
    Built on first sync
    Attributes read from your catalog and imagery. No tagging.
  2. 02
    Tuned per shopper
    Skips, saves, try-ons and purchases adjust that person’s weights only.
  3. 03
    Carried back
    On the next visit the stylist already knows the budget, the silhouette, the colours she never picks.
  4. 04
    Surfaced to you
    Aggregate patterns appear in the merchant dashboard as occasion and catalog-gap insights.
STYLE GRAPH · FRAGMENT
Silk slip dress
  ├ occasion · dinner
  │   → Camel coat
  └ palette · neutral
      → Slingbacks
Printed midi
  └ SKIPPED ×3 · weight ↓
Per shopper, per store. Never shared.

[ 04 ]

the numbers

What does it do to the numbers?

0.0×
Items per session
Per shopper
Weights, per store
0
Personal data required
Industry standards for AI-assisted shopping sessions. Your holdout report shows your own.

Online fashion conversion rates typically sit in the 1–3% range, according to the industry benchmarks in our stats post. When most visitors are treated as strangers, that ceiling is hard to lift. A graph that remembers taste turns each returning visit into a continuation rather than a fresh start, which is where conversion gains come from — not from showing more products, but from showing fewer, better-judged ones.

Source · industry benchmarks, via our 2026 fashion ecommerce stats →

[ 05 ]

versus

How is it different from collaborative filtering and manual merchandising rules?

ElaraCollaborative filteringManual rules
Works on day one without traffic
Learns from skips, not only purchases
Understands occasion and silhouette
Per-shopper, per-store
Updates on every catalog sync

Compare ·

Elara vs Nosto →

·

Elara vs FindMine →

[ 06 ]

use cases

Which use cases does it power?

[ 07 ]

who it is for

Who is it for?

Brands with returning customers and a catalog that changes every season — and anyone whose “you might also like” row shows more of the same.

25K–600K SESSIONS / MO

SHOPIFY & HEADLESS

APPAREL, FOOTWEAR, ACCESSORIES

[ 08 ]

pairs with

[ ? ]

questions

Questions a Head of Ecommerce asks first. 

Questions a Head of Ecommerce asks first. 

Questions a Head of Ecommerce asks first. 

Anything else, book twenty minutes and ask the founder directly.

Ask a question

No. Each store has its own graph and each shopper’s weights stay with that store. Nothing is pooled across brands and nothing trains models for anyone else.

No. Attributes are read from your catalog data and product imagery on sync. You can override or add attributes in the dashboard if you want to.

Collaborative filtering needs traffic and learns from purchases, so it is weakest for new stores, new products and new shoppers. The Style Graph is built from the catalog itself and learns from skips as well as buys, so it works on day one.

Skips, saves, try-ons, swaps in Playground, briefs and purchases — all on-site, all anonymous, no personal data required.

Yes. Merchandising rules pin, boost or exclude products; Scale plans include custom graph tuning with our team.

Per-shopper weights update in the session. Catalog changes enter the graph on the next hourly sync.

[ 09 ]

go deeper

Go deeper

What is a style graph — and why it matters

A plain-language explainer of style graphs in fashion AI.

Why generic recommendations fail fashion stores

Why purchase-based recommenders struggle with taste.

Why “You might also like” is killing your AOV

The case against more-of-the-same product rows.

Give your store a stylist that actually sells. 

Give your store a stylist that actually sells. 

Give your store a stylist that actually sells. 

Thirty days free. Live in under an hour. Measured against a holdout from day one.

Slide to start your trial

No credit card · Cancel any time