2ndOpinion.FYI中文Log in
genius.wiki

#269 2015 · Pinduoduo · E-commerce

Pinduoduo dangled a free product one cent short, forever, to turn shoppers into recruiters

the problem

Customer acquisition cost was eating e-commerce margins

background

By 2015, Chinese e-commerce's biggest cost was no longer building a store online — it was buying a customer's first click. Alibaba and JD.com were locked into rising paid-acquisition costs in the wealthier coastal cities they already dominated, and lower-tier cities represented enormous untapped demand nobody could reach profitably through the same ad channels.

Colin Huang's Pinduoduo skipped the ad auction entirely. A shopper could get a product for free — but only by asking friends to each help shave a slice off the price. The trick was in the slices: instead of even, predictable cuts, the amounts were randomized and shrank unpredictably as the price neared zero, so the shopper always felt one more friend away from free. Researchers later showed this randomized structure recruits more sharers per yuan of discount given away than any smooth, predictable formula would — the platform had, in effect, engineered a slot machine into a checkout flow.

what everyone would do

The standard e-commerce growth playbook was to buy customers through paid ads, or at best structure referral discounts as a fixed, predictable schedule — invite five friends, get 20% off. A predictable formula is exactly what lets a rational shopper calculate how many invitations are actually needed and stop the moment further effort stops being worth it, capping how many new users each existing user brings in.

what they saw

Pinduoduo's designers saw that predictability itself was the growth constraint: a shopper who can calculate the remaining gap to zero can also calculate when to quit. Making each friend's discount cut random and unpredictable near the end removed that calculation entirely — the shopper could never be sure how many more invites stood between them and a free item, only that the next one might be the one that finishes it.

the move

Pinduoduo's "砍一刀" (bargain-help) feature let a shopper drive a product's price toward zero for free — but only if friends each tapped a link and sliced off a piece of it. Rather than shrinking those slices by a fixed, predictable formula, the platform's cuts were randomized and time-limited, an all-or-nothing structure that a 2024 peer-reviewed analysis showed systematically out-recruits and out-profits any deterministic discount schedule, at the cost of shoppers who chase the last few cents forever without quite reaching them.

why it works

Each invited friend's tap shaves off a random slice of the remaining price rather than a fixed fraction, so a shopper can never compute exactly how many more invites are needed — this is the same variable-ratio reinforcement structure that makes slot machines and loot boxes so persistent, applied to a checkout flow instead of a game. Because the uncertainty itself is what sustains recruiting effort, shoppers send more invitations per discount yuan given away than any smooth, predictable formula would produce, which is why the acquisition cost per new user ran an order of magnitude below what Alibaba and JD.com were paying for the same customers through advertising.

the payoff

Pinduoduo grew to over 300 million users by its July 2018 Nasdaq IPO (raising roughly $1.63 billion), concentrated heavily in the lower-tier Chinese cities Alibaba and JD.com's ad budgets barely reached. Contemporary estimates put its customer acquisition cost around ¥11 (about $1.64) per new user, versus roughly $46 for Alibaba's Taobao/Tmall and $35 for JD.com over the same period — built on friends recruiting friends rather than paid advertising.

where it breaks

The mechanism needs a product and audience where recruiting friends carries low social cost — it thrives inside China's app-embedded social graphs and fails wherever repeatedly asking contacts for favors reads as spam or damages a relationship, and it needs enough margin on the underlying goods to absorb giving some away free to the subset of shoppers who complete the chain. It also has a trust ceiling: users who recognize the pattern (the case's own backlash — the price that never quite reaches zero) grow resentful rather than engaged, so the same unpredictability that recruits new users can corrode goodwill among repeat ones once its mechanics become common knowledge.

what came after

The "砍一刀" loop became Pinduoduo's primary growth engine on the way to its 2018 Nasdaq listing and is now studied in academic mechanism-design literature as a rigorously optimal way to convert a discount budget into free customer recruitment — even as it drew a running backlash from Chinese users over the price that never quite reaches zero.

references

  1. [1]Help-and-Haggle: Social Commerce Through Randomized, All-or-Nothing DiscountsManagement Science, 2024pubsonline.informs.org
  2. [2]拼多多"砍一刀",永远差一刀?21世纪经济报道, 2022m.21jingji.com
  3. [3]Chinese e-commerce platform Pinduoduo debuts on NasdaqXinhua, 2018xinhuanet.com

keep it

same kind of clever

Back to the archive