Tercile

Segment everything. See every opportunity.

If wool is 30% of your rail but 50% of your sales, you should be buying more wool. Tercile reads every product page in your store — descriptions, tags, metafields, titles, and soon the images themselves — and splits revenue, sell-through and returns by every attribute a buyer thinks in. Colour. Fabric. Fit. Construction. The model it was shot on. Your strongest segments, your weakest, and every gap between shelf and demand.

Decisions, not dashboards.

01

Tercile builds a dataset most brands never have — every attribute × every order × every session × every day — then an AI analyst reads it and hands you the conclusions as cards. Three to ten per page, only when they're earned. No padding, ever. And the big ones can come to you — an insight digest by email, weekly or monthly, your call.

RevenueUnitsAOVFull-price shareAvg discountReturn rateSell-through speedDemand vs shelfViews *Add-to-cart *Conversion *Profit †Margin rate †

* with the storefront pixel, coming in beta  ·  † when unit costs live in Shopify

Opportunity · colour
Black outsells its shelf 1.6 to 1
Black holds 36% of your assortment but takes 59% of revenue, nearly all at full price. The strongest signal in your catalogue — deepen the black buy before AW ordering closes.
+63% vs shelf · 94% full-price
Opportunity · shoot
Amélie sells the clothes
Styles shot on Amélie sell +52% above their shelf share; flat-lay shots run −42% on the same categories. Rebook her for the next campaign and reshoot the flat-lay best-sellers.
+52% vs −42% flat-lay
Inefficiency · fabric
Viscose only moves at a discount
Viscose sits on 16% of your shelf, returns 7% of revenue, and 37% of its units sold on markdown at an average 28% off. Cut the depth or reprice the category.
−55% vs shelf · avg discount 28%
Watch · fit
Oversized sells — and comes back
Oversized sells +30% above its shelf share, but returns 23% of units against 10% for regular fits. Strong demand, fragile margin. Watch the next drop's return window before reordering.
+30% vs shelf · returns 23% vs 10%

The AI reads the whole product page.

02

Not just the title. Tercile mines the full page — description copy, existing tags, metafields, product names — plus a vision pass over product imagery in beta — and turns it into clean, consistent attribute data across your whole catalogue. The tags you never had time to create.

Mara Coat — Washed Black
Description

An oversized, double-breasted coat cut from 80% virgin wool. Generous through the shoulder, dropped sleeve, falls below the knee. Shot on Amélie, who wears a size S. Garment-washed for a soft, lived-in black.

Tags
outerwearAWwoolnew-in
Metafields
custom.composition: "80% wool / 20% PA"
custom.fit_note: "runs large"
product imagery → vision pass
Extracted — with confidence
ColourBlack0.98
FabricWool0.97
CategoryCoat0.99
FitOversized0.93
ConstructionWoven0.91
SeasonAW0.88
ModelAmélie0.74 → review

Every segment. One wall.

03

Twelve segments out of the box — plus anything you define yourself. Each one shows revenue share and one honest number: how far its revenue runs above or below its share of shelf. +63% means customers buy it faster than you stocked it. −55% means you overbought.

Colourtagged
black
59%+63%
ecru
20%+4%
olive
11%−8%
rust
6%−39%
Fabrictagged
wool
44%+38%
cotton
27%+1%
denim
17%−4%
viscose
7%−55%
Fittagged
oversized
48%+30%
regular
36%−2%
slim
16%−30%
returns: 23% vs 10%
⚠
Shot on modelyours
amélie
41%+52%
jonas
24%+8%
flat-lay
16%−42%
ghost
19%−6%

colour · fabric · construction · fit · silhouette · print · gender · season · category + size, price band & newness computed automatically + anything you invent

See exactly where you over- and under-bought.

04

The gap chart puts assortment share against revenue share. Green right of the line: this attribute earns more than its shelf space. Red left of it: shelf space wasted. This is the chart your next buy should be built on.

black
+22.6 pp
wool
+12.1 pp
oversized
+11.0 pp
ecru
+0.8 pp
slim
−6.9 pp
rust
−7.8 pp
viscose
−18.4 pp
← overboughtassortment = revenueunderbought →

When it sells is half the insight.

05

Start-of-winter wool is not end-of-winter wool. Tercile tracks every segment over time: what your sale periods actually clear and at what average discount, which drops sell hard then tail off, which gain momentum week after week.

Wool — revenue by monthpeaks Nov–Jan
JFMAMJJASOND
Linen — revenue by monthpeaks May–Aug
JFMAMJJASOND
Drop 3 — revenue by monthhard launch, long tail
M1M4M8M12

Any catalogue with character.

06

The segments are yours to define. Whatever your product thinks in — metal, shape, capacity, designer — Tercile tags it and prices its shelf space.

Fashion colour · fabric · fit · construction · silhouette · season · shot on model oversized +30% vs shelf — but 23% returns
Jewelry metal · stone · setting · carat band · collection · designer gold: 30% of assortment, 50% of sales — buy more gold
Bags & gear shape · capacity · travel vs city · material · colourway city silhouettes outsell travel 2:1 per shelf metre

Plain verdicts, honest numbers.

07
BUY MORE

At least +15% vs shelf share, with enough sales behind it. Customers are pulling this attribute off the shelf faster than you stock it.

REDUCE

−30% vs shelf or worse — plus markdown dependency. If 37% of viscose units only sell on discount, the chart says it before the season ends.

HOLD

In balance. No action needed — and Tercile says so instead of inventing urgency.

— COLLECTING

Below 30 net units, we suppress the signal entirely. No verdict on noise. Your buying decisions deserve a sample size.

The AI suggests.
You decide.

Every tag carries a confidence score. High-confidence tags apply automatically. Anything uncertain lands in a review queue — confirm, reject, or fix in one click. And once you've touched a tag, the AI never overwrites it. Your merchandising judgment outranks the model, permanently.

Insight cards on every page are written by a senior-analyst-grade AI reading your actual numbers — 3 to 10 cards, only when they're earned. No padding.

Silk Slip Dress — Pistachiocolour → green? description says "pistachio"
0.71
ConfirmRejectFix
Boxy Overshirt — Indigofit → oversized? "generous, boxy cut"
0.78
ConfirmRejectFix
Mara Coat — Washed Blackmodel → amélie? "shot on Amélie"
0.74
ConfirmRejectFix
0
Segments out of the box
0 min
From install to first wall
0
Net units before we call a signal
∞
Custom segments — designer, drop, shoot…

Pricing

08

The beta deal: everything free for your first 3 months. After that you pick a plan — or stay on Free. No card up front, no auto-upgrade.

Free
$0
try it on your store
  • Up to 50 products
  • 3 segments
  • Segment wall + demand-vs-shelf
  • 30-day history
Starter
$19/mo
for small stores
  • Up to 100 products
  • All preset segments
  • AI insight cards on the overview
  • 12-month history
Beta: 3 months free
Pro
$49/mo
for growing brands
  • Up to 1,000 products
  • Unlimited custom segments
  • Gap charts + AI insights on every page
  • Insight digest by email — weekly or monthly
  • Full order history
Beta: 3 months free
Growth
$149/mo
for teams & multi-store
  • Up to 2,500 products
  • Everything in Pro
  • Seasonality + newness analysis
  • Priority tagging + support

More than 2,500 products? Enterprise — on demand →

Stop buying
on gut feel.

Tercile is in free beta for Shopify stores with catalogues that have character — fashion, jewelry, accessories, gear. Early access installs get every feature free for 3 months, white-glove segment setup, and a direct line to the founder.

Get early access