The Loupe Index · September 21, 2026

What independent fashion is actually doing

Nobody measures the independent tier. Enterprise forecasting starts above it, wholesale data stays private, and a brand's own dashboard shows only its own store. So we built the instrument out of the thing we already had: a full snapshot of 200 labels' catalogues, taken every day since 2026-06-17.

200 labels 8,839 pieces tracked 92 daily snapshots 2026-06-17 → 2026-09-20

Four numbers nobody else has

Every rate on this page carries its sample size and a 95% confidence interval. Where the sample is too thin, the row is not printed rather than printed small.

Full price

97.6%

of 9,537 tracked pieces held their price across 13 days, 2026-09-08–2026-09-20. 127 of 159 labels did not mark down a single piece. That is a short window: our own price pipeline changed on 2026-09-05 and prices either side of it are not comparable, so this figure is not comparable with an earlier Index either. Over six weeks before that change, the same measure read 96.5%.

Sold out now

13.6%

of tracked pieces have no size left in any variant (8,839 pieces, 95% CI 12.9–14.3%).

42-day turnover

44.3%

of the pieces we tracked on 2026-08-09 are no longer in their brand's recent listings. Matched on title instead of id: 42.6%.

Newness

42.8%

of today's shelf was published in the last 42 days. The median label refreshed 32% of its shelf.

What is selling out

A piece counts as sold out when every size of it is gone and the brand has not restocked. Out-of-stock pieces stay in our feed, so this is measured directly rather than inferred from something vanishing.

New in the last 42 days

10.7%

sold out, off 3,783 arrivals.

On the shelf since 2026-08-09

15.7%

sold out, off 5,056 pieces. These two are not comparable on a 42-day window: an arrival has had at most 42 days to sell, standing stock has had months. Read each on its own.

Category	Tracked	Sold out	95% CI
tops	3,148	15.9%	14.6–17.2
accessories	2,019	13.0%	11.6–14.5
bottoms	1,571	10.5%	9.1–12.1
dresses	1,223	12.1%	10.4–14.0
outerwear	515	13.8%	11.1–17.0
shoes	363	15.4%	12.1–19.5

Price band	Tracked	Sold out	95% CI
under $60	500	22.4%	19.0–26.3
$60–119	1,916	14.5%	13.0–16.2
$120–199	2,278	12.2%	10.9–13.6
$200–349	1,925	14.2%	12.7–15.8
$350–599	1,327	12.1%	10.4–13.9
$600+	893	11.2%	9.3–13.4

Sell-out is a barbell, not a slope: the cheapest and the most expensive bands clear hardest, and the middle of the market sits.

Almost nobody discounts

Measured across 2026-09-08–2026-09-20, the longest stretch with no change in how we read prices.

Marked down

2.4%

of 9,537 tracked pieces ended the window below where they started. 58 ended above it.

When they do cut

16%

median markdown. Middle half of cuts: 6%–21%.

Full-price houses

127 / 159

labels with 20+ tracked pieces that did not cut a single price in the window.

Every one of these held full price on its whole tracked shelf:

SIEDRÉSFashion Brand CompanyTrashy ClothingVESTIGEShop PecheSt. AgniMartha CalvoDana FoleyPriscaveraTae ParkMiaouTyler McGillivaryThe Frankie ShopCafe ForgotDamson MadderBec + BridgeAudetteHand OverEB DenimGil Rodriguez6397Paris GeorgiaSandy LiangChristopher Esber

Some of the biggest "discounts" in this data were ours, and they are excluded. A real sale marks different pieces down by different amounts. A currency correction multiplies every piece by exactly the same number. We look for the second: a label whose whole line steps by one identical ratio on one day is treated as our pipeline changing, not the brand changing its mind, and prices are never compared across that day for that label. 74 such brand-days exist across the whole window and 58 of them moved by exactly a currency-pair ratio from our own conversion table — Stine Goya's line moved ×0.134, which is the Danish krone peg, not an 87% sale. 0 fall inside the window above:

This cuts against us and we would rather it did. A genuine sitewide "everything 20% off" is indistinguishable from a currency fix in catalogue data, so it would be voided too. The markdown figures on this page are therefore a floor: the tier discounts at least this little, possibly slightly more.

Where the tier prices

The question every small brand has and cannot answer: am I priced right? This is the whole tracked tier on 2026-09-20, by category.

Category	Pieces	10th	25th	Median	75th	90th
tops	3,148	$64	$89	$135	$225	$400
accessories	2,019	$60	$109	$200	$359	$560
bottoms	1,571	$85	$120	$175	$295	$465
dresses	1,223	$123	$175	$285	$500	$825
outerwear	515	$154	$233	$425	$755	$1,900
shoes	363	$140	$198	$293	$630	$845

Sitting between the 25th and 75th is not a virtue in itself — but a brand two whole quartiles off the tier in one category and on it in another is usually doing that by accident.

What the tier is making now

Share of the tracked shelf on 2026-08-09 against 2026-09-20, across the 220 labels present on both dates.

Colour	08-09	09-20	Shift	Pieces
neutral	44.4%	44.4%	+0.0	4,383
black	13.5%	14.1%	+0.5	1,388
blue	8.7%	8.4%	-0.3	825
white	9.0%	8.1%	-1.0	796
multicolor	7.6%	6.9%	-0.7	684
brown	5.2%	6.3%	+1.1	622
green	4.3%	4.5%	+0.1	440
pink	4.3%	4.1%	-0.2	400
red	3.0%	3.4%	+0.4	336

Price band	08-09	09-20	Shift	Pieces
under $60	6.0%	5.6%	-0.3	496
$60–119	23.5%	21.7%	-1.7	1,912
$120–199	26.8%	25.8%	-1.1	2,265
$200–349	22.0%	21.7%	-0.4	1,906
$350–599	13.1%	15.1%	+2.0	1,324
$600+	8.6%	10.2%	+1.5	893

The shelf's median price rose from $178 to $189 — and not because brands raised prices. New pieces are arriving below the ones that have stayed (median $210 arriving against $172 standing). The shelf gets more expensive because the cheap things leave it fastest: pieces under $60 turned over at 45.6% against 42.6% at $350–599.

Arriving vs standing

Share of pieces already on the shelf on 2026-08-09 against share of everything published since.

Price band	Standing	Arriving	Shift
under $60	6.1%	5.1%	-1.0
$60–119	25.2%	16.9%	-8.3
$120–199	26.6%	24.7%	-1.9
$200–349	20.8%	23.0%	+2.2
$350–599	13.7%	16.8%	+3.1
$600+	7.6%	13.5%	+6.0

Category	Standing	Arriving	Shift
tops	37.0%	33.7%	-3.3
accessories	24.9%	20.1%	-4.7
bottoms	17.5%	18.2%	+0.7
dresses	14.8%	12.6%	-2.2
outerwear	3.5%	9.0%	+5.5
shoes	2.4%	6.4%	+4.0

How fast the shelf empties

Of the pieces we tracked on 2026-08-09, the share no longer in their brand's recent listings on 2026-09-20. That is delisting and being pushed off by newer drops together — we can see both happen and cannot always tell which, so we say so rather than call it sell-through.

Price band	Tracked	Gone	95% CI
under $60	526	45.6%	41.4–49.9
$60–119	2,068	38.8%	36.8–40.9
$120–199	2,363	43.8%	41.8–45.8
$200–349	1,942	48.2%	46.0–50.4
$350–599	1,154	42.6%	39.8–45.5
$600+	759	52.4%	48.9–56.0

Category	Tracked	Gone	95% CI
tops	3,104	42.1%	40.4–43.9
accessories	2,019	39.2%	37.1–41.3
bottoms	1,688	48.6%	46.2–51.0
dresses	1,282	42.6%	39.9–45.3
outerwear	369	55.8%	50.7–60.8
shoes	350	66.6%	61.5–71.3

Sixteen looks, and which ones are moving

These groups come from the images, not the category labels — every piece is embedded with a fashion vision model and clustered by what it looks like, so a boxy cropped knit lands next to another boxy cropped knit whatever its brand called it. The names are the words that are unusually common in each group's own titles; nobody wrote them. Index 100 means the group is refreshing at exactly the tier's rate.

unisex, coat, jacket

776 pieces · 8.8% of the shelf · median $281

New 74% idx 173 Sold out 9.8%

collar, coat, jacket

403 pieces · 4.6% of the shelf · median $325

New 65% idx 151 Sold out 13.9%

sneakers, ankle, boots

585 pieces · 6.6% of the shelf · median $366

New 56% idx 131 Sold out 14.5%

sleeve, top, tee

788 pieces · 8.9% of the shelf · median $172

New 53% idx 123 Sold out 5.7%

wallet, socks, scarf

477 pieces · 5.4% of the shelf · median $110

New 47% idx 110 Sold out 20.1%

necklace, pendant, bracelet

569 pieces · 6.4% of the shelf · median $205

New 42% idx 97 Sold out 13.5%

gown, dress, skirt

627 pieces · 7.1% of the shelf · median $350

New 42% idx 97 Sold out 10.0%

skirt, pants, pant

887 pieces · 10.0% of the shelf · median $155

New 37% idx 86 Sold out 12.1%

dress, skirt, top

676 pieces · 7.6% of the shelf · median $226

New 36% idx 83 Sold out 10.8%

tank, cotton, tee

597 pieces · 6.8% of the shelf · median $125

New 35% idx 82 Sold out 15.2%

hoops, earrings, ring

616 pieces · 7.0% of the shelf · median $200

New 34% idx 79 Sold out 7.5%

pillow, lavender, shirt

277 pieces · 3.1% of the shelf · median $117

New 33% idx 78 Sold out 36.8%

top, tank, skirt

826 pieces · 9.3% of the shelf · median $120

New 26% idx 61 Sold out 16.3%

bottom, bikini, triangle

354 pieces · 4.0% of the shelf · median $83

New 26% idx 60 Sold out 9.6%

acetate, tort, sunglasses

323 pieces · 3.7% of the shelf · median $200

New 13% idx 31 Sold out 17.6%

The tier is two markets

Of 155 labels with 20+ tracked pieces, 15 (10%) published nothing at all in 42 days, while the fastest replaced their visible shelf several times over. Averages across this tier describe almost nobody.

Refreshing fastest

New pieces published since 2026-08-09, against the size of the shelf we track.

Label	Shelf	New	Refresh
The Frankie Shop	60	422	703%
Fashion Brand Company	60	323	538%
Chan Luu	60	234	390%
Paloma Wool	46	166	361%
VESTIGE	60	215	358%
INSIS	60	168	280%
Emily Dawn Long	60	163	272%
6397	60	140	233%
Gimaguas	52	120	231%
Damson Madder	60	135	225%
Bec + Bridge	60	132	220%
Baserange	52	109	210%

Selling through fastest

Share of the tracked shelf with every size gone, 20+ pieces. Only labels that have published something in the window, so this is live stock clearing rather than a shelf nobody has refreshed.

Label	Shelf	Sold out
Martine Rose	60	100%
Emily Dawn Long	60	100%
Mimi Wade	32	72%
Cowboys of Habit	56	62%
Dana Foley	60	58%
Little Yarn	60	58%
Gumi	52	58%
Addy Knits	58	53%
Mali Marea	60	47%
Woodrose Deli	40	45%

Free, for any label we track

Your own numbers against the tier

We build a private card for each of the 155 labels with enough tracked pieces to measure: your sell-out rate, your turnover, your refresh rate and where your prices sit in your own categories against everyone else's — each one next to the tier figure over the same window. It is free, it is yours, and nothing on it is published here — the card is stored encrypted and the link we send you is the only key to it.

Email tryloupeapp@gmail.com with your label's name and we will send the link.

How this is measured, and what it cannot tell you

What we see. Every label's public product feed, read once a day and committed. We keep up to 750 pieces per label, drawn from its 250 most recently published listings. For a small store that is the entire store; for a large one it is the front of the catalogue. Every number on this page is of what we track and we mean it literally.

Why 200 and not 232. The Loupe app carries 232 labels, but 32 of those reach it only through a partner boutique's shelf. What a boutique chooses to buy, and what it charges, is that boutique's decision — not the label's. Mixing the two would corrupt every number here, so the Index reads brand-direct storefronts only: labels publishing their own prices, their own assortment, their own drops.

The window. 92 daily snapshots, 2026-06-17 to 2026-09-20. Comparisons over time start at 2026-08-09, when our roster stopped growing — before that, new pieces are us adding labels, not brands publishing. The daily job did not run between 2026-07-24 and 2026-07-29, so 4 days are missing from the middle of the window.

Turnover is measured once, between two dates. Day-to-day disappearance is 45% noise in this data: of 10,059 overnight vanishings, 4,552 came back. Anything computed as a daily rate would be mostly measurement.

Our own decisions are removed. 8 labels left our roster in the window (Agus Chueizer, CHINI, Gitika, HOUGA, Linen & More, MAS1, Po-em, Wangda), some by our own choice. Every longitudinal number here uses only the 220 labels present on both dates.

Prices are never compared across a change in how we read them. On 2026-07-15 we pinned our scrape to the US market and 49 geo-priced labels flipped to USD; the next day 36.5% of the catalogue moved price. That whole boundary is void here. Separately, any single label whose line moves by one identical multiplier on one day is voided for that label — 74 such brand-days were found, 58 of them landing on an exact currency-pair ratio.

Stock data starts 2026-07-16. Before that we did not record whether a piece was in stock, so nothing earlier is asked. Partner-boutique listings are excluded from sell-out entirely, because we only ingest those in stock.

What we cannot tell you. Units, revenue, margin, returns, or whether a sold-out piece sold out at full price. We see a catalogue, not a till. We also cannot see a brand's stock depth — "sold out" here means no size is purchasable, which for a made-to-order label may never happen however well it sells.

Would our own users' swipes predict any of this? No — and we checked. Across 133 labels, in-app approval rate against sell-out rate gives r = +0.09, which at this sample is indistinguishable from nothing. The Index does not use app behaviour anywhere.

The record. Every figure above is in data.json, regenerated on the same schedule and kept in version control, so any week's numbers can be diffed against the last.