Singles' Day (Double 11)

Singles' Day, commonly called Double 11 after the date 11 November, is the main autumn shopping festival in Chinese e-commerce. It began as a one-day promotion and has become a multi-week cycle. Most errors in reading Chinese category data trace back to still treating it as a day.

The cycle, not the date#

A modern Double 11 typically runs through several distinct phases:

1. Pre-sale. Consumers place deposits on products days or weeks ahead, locking in a promotional price.
2. Settlement windows. Balances are paid, sometimes in more than one wave. This is when a large share of transactions is actually recorded.
3. The main date. 11 November itself, with its own peak.
4. The tail. Continuing promotion, followed by a trough as pulled-forward demand is worked through.

Two things follow immediately. First, when a transaction counts is a reporting choice, at the deposit or at the balance, and platforms have not always made it the same way. Second, the peak is not the event; a substantial share of the volume lands outside the headline date.

Why year-on-year comparison breaks#

The headline date is fixed. The cycle around it is not. Platforms adjust when pre-sale opens, how many settlement windows run, and how long promotion continues either side.

So comparing the same calendar window across two years can capture a different portion of each year's cycle, with one year including the pre-sale build and the other missing it. The comparison then reports a change in demand that is really a change in the promotional calendar.

The correction is to compare equivalent cycle positions rather than equivalent dates, and to state which windows are being compared. Where the cycle structure genuinely changed between years, a like-for-like comparison may not exist, and saying so is better than producing a number that cannot be defended.

A related trap: comparing a promotional window against a non-promotional one. This produces enormous apparent growth and means nothing. See GMV, where promotional inflation is treated in general.

Double 11 in 2025#

The 2025 festival opened earlier again. In 2024 the main marketplaces had opened on 14 October, already about a week earlier than in previous years, and in 2025 JD opened on 9 October while Tmall took deposits from 15 October and opened sales on 20 October. In both years, then, the whole cycle fell inside October and November, which makes those two calendar months the fairest window in which to compare one festival with the other.

Read that way, as one combined total in Moojing's e-commerce data, the festival did not lift every category in the same way:

In beauty, large appliances and snacks the festival months grew more slowly than the rest of the year, and in beauty they fell slightly while March to September grew, so a brand in those categories that judged its year on the festival alone would have understated how the year went. In large appliances units rose by a fifth while sales value stayed flat, which means the average price paid per unit fell by about 17%, and a volume figure on its own would have read as a strong festival when the money spent had not moved. In women's apparel the opposite happened, with value growing far faster than units, so the gain came from what each item cost rather than from more items being bought. In pet the festival months grew at about the pace of the rest of the year, which we believe says more about a growing category than about the festival.

The earlier start also shows in the timing. In four of the five categories a slightly larger share of the two months' sales fell in October in 2025 than in 2024, which is what one would expect when the cycle opens sooner, and in those four a comparison of November alone would have understated 2025 against 2024, which is why the two months have to be read together.

These figures are Moojing's estimates of combined online sales on the Chinese marketplaces in Moojing's e-commerce data, not of the whole Chinese market, and units are modelled. See GMV for how promotional pricing affects any sales total read from the festival.

Created demand versus moved demand#

Much of Double 11's volume is deferred, not incremental. Consumers hold planned purchases for the event and stock up on replenishable goods. The signature is a peak followed by a trough.

The balance varies by category:

• Long replacement cycles (appliances, electronics): deferral dominates. The event mostly redistributes purchases that would have happened anyway.
• Replenishables (household, personal care): stocking up dominates, and the trough is proportionally long as consumers work through inventory.
• Impulse and discovery-led categories: more of the volume is genuinely incremental, because the purchase was not planned at all. This is part of why content-led channels behave differently through the cycle; see Douyin E-Commerce.

The practical rule: never read the event window alone. The weeks before and after are part of the same measurement. A brand judging the event on its peak will overstate what it achieved, sometimes by a wide margin.

Price effects distort the category picture#

Realised prices during the cycle can sit well below their normal level, and unevenly across the category. Two consequences:

Price band work must exclude or isolate the period. A Price Band Analysis run across a window containing the event will show demand clustered in bands that reflect promotional pricing rather than the category's normal structure. Either exclude the period or band it separately and label it.

Volume bought at a deeper discount is not equivalent volume. Two brands with identical event sales can have made very different amounts of money. Any performance read that ignores realised discount depth is measuring activity rather than outcome.

Measuring a brand's performance properly#

1. Full cycle, including the trough. Pre-event, event, post-event, against the equivalent window a year earlier.
2. Realised discount depth. What the volume actually cost to win.
3. Return rate. It runs higher after promotional periods, and returns are settled well after the event closes, so an early read is always flattering.
4. Repeat behaviour of newly acquired customers. Whether event buyers come back is what separates a customer-acquisition investment from an expensive one-off. See Category Penetration and Repurchase.
5. Channel mix. Tmall, JD and Douyin run different cycle shapes, so a blended total hides how each channel contributed. See Tmall, JD and Douyin: How the Channels Differ.

Where to look next#

For where Singles' Day sits among the other festivals and holidays of the year, see China's E-Commerce Calendar. For the December festival that follows it, see Double 12 (12.12). For promotional inflation in sales figures generally, see GMV. For price distribution work around the event, see Price Band Analysis. For channel differences through the cycle, see Tmall, JD and Douyin: How the Channels Differ and Douyin E-Commerce. For whether acquired customers return, see Category Penetration and Repurchase.

Common questions#

What is Double 11?#

Double 11, also called Singles' Day, is China's main autumn shopping festival, named for the date 11 November. It began as a single-day promotional event and has become a cycle running several weeks, with a pre-sale phase where consumers place deposits, one or more settlement windows where balances are paid, and secondary peaks around the main date. It is now less a day than a season, and treating it as a day is the source of most misreadings of Chinese e-commerce data.

Why does Double 11 make year-on-year comparison unreliable?#

Because the shape of the cycle changes between years while the headline date does not. Platforms adjust when pre-sale opens, how many settlement windows there are, and how long the surrounding promotion runs. A comparison of the same calendar window across two years can therefore capture a different portion of each year's cycle, with one year including the pre-sale build and the other not. The result reads as a change in demand when it is a change in the promotional calendar. Comparing equivalent cycle positions rather than equivalent dates is the correction.

Does Double 11 create demand or move it?#

Substantially it moves it, and the proportion varies by category. Consumers defer planned purchases into the event and stock up on replenishable goods, which produces a peak followed by a trough as the pulled-forward demand is worked through. In categories with a long replacement cycle the deferral effect dominates; in impulse and discovery-led categories more of the volume is genuinely incremental. The practical implication is that the event window should never be read alone, because the weeks before and after are part of the same measurement, and a brand judging the event on its peak alone will overstate what it achieved.

How should a brand measure its Double 11 performance?#

Across the full cycle including the trough that follows, not the peak alone. Useful measures: sales across the whole pre-event, event and post-event window against the equivalent window in the prior year; the realised discount depth, since volume bought with a deeper discount is not equivalent volume; the return rate, which runs higher after promotional periods; and whether new customers acquired during the event buy again afterwards, which is what separates an expensive customer-acquisition exercise from a profitable one.

How did Double 11 go in 2025?#

It opened earlier again, from 9 October on one major marketplace and 15 October on another, against 14 October in 2024, so the whole cycle sat inside October and November in both years. Comparing those two months as one combined total in Moojing's e-commerce data, the festival lifted categories unevenly: women's apparel and pet grew strongly, snacks grew more slowly than the rest of the year, large appliances sold about a fifth more units for the same money, and beauty sales fell slightly while the rest of beauty's year grew. The lesson is the one the event always teaches, which is that the festival has to be read against the rest of the year and in units as well as value.