bridal dresses benchmarks, observed from public storefronts.

How these numbers were produced

• They are observed public storefront data. oddly reads publicly visible storefronts and aggregates what is published there: product prices, catalog size, market and currency. They are not first-party merchant data, they are not survey data, and no figure here is estimated.
• Every figure rests on at least 8 observations. A cell below that floor is withheld rather than shown thin, and each figure states the number of observations behind it. That floor counts OBSERVATIONS, which is the honest unit for public data; the tighter contributor floor applies to opted-in first-party cohorts, which no figure on this page uses.
• Prices are medians and quartiles of observed distributions, never averages of averages. Two markets are never pooled into one number, and every amount states its currency.
• Each figure carries the date it was observed. When a category is crawled again, the newer observation replaces the older one on the page rather than sitting beside it.
• A sample floor is not a sanity check, so two further rules apply. A cell reporting no spread at all is not a distribution and is withheld. A market or product figure that disagrees with its own category by more than a factor of 10 is withheld pending review, because at that distance it is a measurement fault rather than a real difference. Across the whole layer 23 figures are currently withheld on those two rules, and 0 for resting below the floor. Nothing is ever adjusted to fit.
• Stored metric keys behind this page: price.median, price.dresse.
• Free to cite under CC BY 4.0.

Corpus: 603 observed storefronts, 2,037,725 observed data points, latest observation 2026-08-05. Pages are published for 9 of the 11 category nodes the corpus covers; the rest carry only measurements this layer does not publish.