Summary

“Deal hunters” have evolved from individual consumers who knew how to work coupons into professional arbitrage chains powered by tutorials, crawlers, and account farms. Return-shipping insurance arbitrageurs use roughly 2 yuan in partner logistics costs to collect 8-12 yuan in reimbursement per order; at scale, cases can involve several million yuan. Fulfillment-compensation arbitrage targets low-activity stores or installation products that sellers cannot fulfill, generating 400-500 yuan per order. What began as individual behavior is now organized and professionalized, turning loopholes into steady income.
The spread of refund-only policies is fundamentally the result of platforms using merchants’ money to compete for users as e-commerce shifts into a mature, zero-sum market. Pinduoduo pioneered the policy in 2021, winning consumers over with the refrain, “The prices are great, and refund-only is even better.” By 2023, its growth, profits, and market-cap performance had peers questioning their own restraint; Taobao, JD.com, and later content-commerce platforms followed. Yuan Bu’s view is that once “users first” became synonymous with more aggressive, more extreme compensation, the ecosystem’s existing balance began to unravel.
Malicious consumers exploit merchants’ no-win choice—fail to ship and pay compensation, or ship and still risk losing money—not simply the chance to buy cheaply. One buyer ordered 100 large wall paintings at once: the merchant would owe a contractual penalty if it did not ship, but would absorb production, packaging, transport, and return damage if it did. Other buyers request an in-transit refund immediately after the merchant clicks “ship,” then pressure the seller to transfer money privately. RMB199 courses promising “free shopping” and “earn RMB1,000 a day” have brought these tactics to a mass audience.
Easy returns have released legitimate consumer demand while materially amplifying opportunistic and malicious behavior. After the pandemic, return rates at some stores rose from roughly 20% to 35-40%, especially in apparel, sporting goods, and mother-and-baby products. Content-driven impulse buying, ubiquitous return-shipping insurance, and the habit of buying more to try more have all pushed returns higher, a trend Wei Shijie considers essentially irreversible. The problem is that genuine demand is mixed with buying authentic goods and returning counterfeits, cross-platform shipments of cheaper substitutes, and customers returning more because unlimited shipping coverage makes it feel like an all-you-can-eat buffet.
Platforms have built mature merchant-credit systems but still lack a consumer-credit system that can operate openly across the market. Taobao currently relies mainly on internal abnormal-behavior tags, real-person verification, purchase restrictions, and downgraded benefits for extreme accounts. But incompatible data systems, privacy concerns, differential treatment, and false positives have kept a market-wide consumer score at the discussion stage. Yuan Bu explicitly agrees that “credit must be bilateral,” while acknowledging that black- and gray-market operators can rapidly replace accounts, leaving governance in a constant security arms race.
Taobao’s decision to loosen refund-only rules starting last July shows that the industry has shifted from user acquisition to rebuilding merchant confidence. The platform first relaxed the policy for merchants with experience scores above 4.8, then lowered the threshold to 4.6 before the Lunar New Year. As of the point discussed in the episode, non-essential refund-only scenarios had been cut by roughly 80-90%, with exceptions limited to “fake-crack” claims, goods unsuitable for return, perishables that would spoil in transit, or items worth less than shipping costs. Merchant appeals may have fallen by more than 60%, while the platform intercepted roughly 400,000 malicious refund orders a day around Singles’ Day.
The most concealed effect of refund-only is to shift governance costs that should sit with platforms onto merchants and the consumer ecosystem. Automated refunds reduce evidence review, customer-service mediation, product screening, and compliance spending, but force merchants to raise prices, cut quality, or exit—leaving compliant consumers to pay for a small minority’s outsized benefits. “The platform is a signal amplifier,” so the sustainable path is not to compete over who will compensate more aggressively, but to restore rewards and penalties, build bilateral credit, and let platforms compete on their own supply and user positioning.

Deep dive

1. Deal Hunters: From Smart Shopping to Professional Arbitrage

Wei Shijie looked back at the media portrayal of deal hunters in 2018. They were then seen as consumers who were good at calculating discounts, understood the rules, and used coupons rationally—“almost a middle-class lifestyle.” The term was still broadly neutral.
Yuan Bu declined to assign a precise year to the turning point, but said 10 years of observation since the 2014 Consumer Rights Protection Law revealed 2 changes: the behavior moved from individuals to organized groups, and the tactics evolved from basic discount calculations to precisely identifying exploitable gaps across platforms.
The defining question today is no longer how much money someone saved, but whether they are earning in bulk without any intention to consume. Tutorials, social-media communities, data software, and account networks now work together, making “earn-money” technical arbitrage the public’s dominant image of deal hunters.

2. Return-Shipping Insurance Turns a 2-Yuan Logistics Cost into Nearly 10 Yuan per Order

Yuan Bu broke down return-shipping insurance arbitrage: users place large numbers of orders and return everything, sending the parcels themselves through offline channels and uploading the tracking numbers. If they have a deal with a logistics outlet, the return cost can be roughly 2 yuan per package while the reimbursement runs 8-12 yuan—“nearly 10 yuan per order.”
Officially scheduled home pickup is settled directly by the platform and carries relatively little risk. High-risk accounts prefer self-shipping because the spread between the reimbursement and the actual logistics cost can be monetized. At scale, cases involving more than RMB1M are common.
The money is not created out of thin air. A small number of people extracting return-shipping insurance raises the cost of the entire system, eventually showing up as merchants no longer offering the insurance, higher premiums, or product prices that absorb the additional losses. “The welfare of the entire consumer base is declining.”

3. Mature-Market Competition Turns Consumer Protection into an Arms Race

From 2014 to the eve of the pandemic, Chinese e-commerce was still in a high-growth penetration phase. Platforms used subsidies and benefits to attract new users, while the consumption narrative was one of upgrading into an “ideal life on Tmall.” Platforms had begun offering 7-day no-questions-asked returns and instant refunds as early as 2008. Early shipping-compensation claims still required consumer complaints; later, especially after the pandemic, compensation increasingly became automatic.
Once the post-pandemic traffic dividend faded, e-commerce moved into a stock market. To retain users, platforms began competing on “more extreme service,” including refund-only policies and proactive compensation for late shipments. Compensation that once required a consumer complaint gradually became an automatic system payment.
Wei Shijie describes the period as a “four-kingdoms war.” Research firms once estimated Taobao and Tmall’s combined share at roughly 65-70%; low-price e-commerce, content commerce, and JD.com’s quality-and-logistics strategy then entered the same contest. Pricing power and user subsidies became the most visible tools for taking share.
Yuan Bu sees Pinduoduo’s launch of refund-only in 2021 as a possible turning point. By 2023, its strong growth, profit performance, and market capitalization—surpassing Alibaba’s—put pressure on peers to follow. They began questioning their restraint: “You start to think that holding the line could cost you the market.”

4. 100 Wall Paintings Show How Merchants Can Lose Either Way

Around 2018, Taobao noticed bulk accounts tying up inventory and forcing compensation, so it established a malicious-behavior complaints center, later renamed Merchant Protection. One typical case involved a buyer ordering 100 large wall paintings at once: “Should you ship or not? If you don’t ship, you owe compensation.”
If the merchant shipped, it had to absorb production, packaging, and oversized-item transport costs, while the buyer could still invoke the 7-day no-questions-asked return policy. If it did not ship, it faced a contractual penalty. A more advanced tactic was to request an in-transit refund immediately after the merchant clicked “ship,” then use the seller’s fear of round-trip damage to demand a private transfer.
The platform can waive shipping compensation when an address is abnormal or contact information is invalid, and it may contact the account holder to verify the purchase. Attackers have since shifted to large numbers of accounts, placing only 1 or 2 orders per account so each transaction looks normal. Detecting the organization requires backend data to connect the accounts.

5. RMB199 Courses Package Black-and-Gray Tactics as “Free Shopping”

Yuan Bu said the market now offers RMB199 courses advertising “free shopping” or “earn RMB1,000 a day.” They teach compensation and refund-only tactics, and provide software that monitors product information across multiple e-commerce platforms.
Upstream crawlers search for products that have not been edited for a long time and low-activity stores with limited sales, inferring that they will be unable to ship on time. Downstream accounts then place orders in bulk. If even a small number of orders trigger contractual penalties, the model can turn a profit.
Another tactic targets made-to-order curtains, furniture, and electric-heating installation services that have not imposed regional sales restrictions, then changes the delivery address on accounts from places such as Shanghai to Tibet. The merchant has no local network for on-site measurement and cannot fulfill the order normally. A single high-value order can generate 400-500 yuan in compensation; hitting just a few orders out of several hundred produces meaningful income.

6. Employment Shifts and Misaligned Incentives Normalize Extreme Behavior

Wei Shijie said the industry often treats stay-at-home mothers and other people with abundant time online as being adjacent to black- and gray-market activity. Yuan Bu rejected an identity-based explanation and instead pointed to the post-pandemic employment structure: full-time work shifting to part-time work, unemployed university students, and slowing urbanization have produced more people seeking low-barrier income.
After major platforms entered the refund-only market in 2021, arbitrage was no longer a technical behavior limited to “a small handful of people.” Yuan Bu cited a news report about someone in a township putting goods obtained through refund-only into a supermarket for resale. The scene was “absurd and surreal,” but it showed how receiving goods for free was becoming normalized.
His sociological explanation is an inversion of rewards and penalties. Behavior that should be punished is rewarded immediately, while the platform provides protection, making non-participation look foolish. As more people follow suit, the rules shift from protecting the vulnerable to functioning as an incentive system that shapes behavior.

7. Legitimate Demand, Opportunism, and Malice Are All Embedded in High Return Rates

After the pandemic, return rates at many stores more than doubled, rising from roughly 20% to 35-40%, with apparel, sporting goods, and mother-and-baby products showing the sharpest increases. Content consumption drives impulse purchases, while return-shipping insurance and easy returns have created a “buy more, try more, return what doesn’t fit” habit. Wei Shijie considers the trend essentially irreversible.
He divides consumers into 3 tiers: high-risk users seeking to cash out; opportunists returning more because benefits such as 88VIP’s unlimited shipping coverage remove the cost; and ordinary users who occasionally forget the 7-day deadline and return items on day 8 or 9. The last group rarely creates disputes—“merchants basically agree 98% of the time.”
Wei Shijie compares unlimited shipping coverage to an all-you-can-eat buffet. Consumers would not otherwise return so much, but once the marginal cost reaches zero, they become more aggressive. Yuan Bu classifies this as opportunistic behavior, showing that the benefit itself changes the behavior of ordinary users.
The genuinely malicious cases include “buy genuine, return fake”: a customer receives an authentic product and sends back a counterfeit, sometimes buying a visually identical low-priced substitute on another platform and entering the original merchant’s returns warehouse as the delivery address so the substitute is shipped directly across platforms for the switch.

8. Platforms Can Spot Abnormalities but Are Always One Step Behind

Taobao allows an individual store to blacklist up to 200 users. Based on platform data and merchant complaints, it identifies abnormal accounts and displays risk warnings to sellers, much like WeChat warning users that a recipient’s account may be suspicious before a transfer.
Account assessments rely on combinations of signals: temporary phone numbers, registration patterns, login frequency, order volume, and complaint counts. An ordinary user might file 1 complaint a month; an account filing complaints every day and accumulating more than 500 in a month is clearly outside normal behavior.
Yuan Bu repeatedly preserves one important limitation: platforms “always have a lag.” Normal users also change addresses, so a single signal is not enough to establish abuse. Platforms often need to wait for abnormal behavior to occur, for the account to extract some benefit, and for negative merchant feedback to accumulate before they can confirm the relationship with any confidence.

9. E-Commerce Law Leaves a Blank on Whether Merchants Can Refuse a Sale

As Wei Shijie summarized Article 49 of the E-Commerce Law, once a product is listed and the consumer pays, the contract is formed. The platform has no right to rescind it; only a court can do so. Applied to the internet’s high-volume, low-value, high-frequency transactions, that framework creates enormous execution friction.
If merchants sued order by order, neither the courts nor the businesses could bear the cost. Reporting to the police generally requires losses of at least 3,000 yuan for a case to be accepted, and even exceeding that threshold may not resolve the problem. The result is that large volumes of low-value malicious behavior fall outside the reach of traditional remedies.
Wei Shijie frames the unresolved question this way: “Consumers have the right to buy, but do merchants have the right not to sell?” How to add rescission conditions for abnormal transactions while preserving the principle of good faith remains a legal and platform-practice “fill-in-the-blank” exercise.

10. Two-Sided Markets Need Consumer Credit, but Labels Are Dangerous

Wei Shijie argues that a mature e-commerce system should have 2 credit scores, one for merchants and one for consumers. Yuan Bu agrees that “credit must be bilateral,” but says the current system is plainly asymmetric: merchant credit is well developed, while the consumer side still relies mainly on internal abnormal-account detection.
Taobao once allowed buyers and sellers to rate each other. In Yuan Bu’s recollection, roughly from 2012 to 2020—or perhaps into around 2022—merchant ratings could appear in consumer profiles. But the feature had limited use cases and little deterrent effect, and was eventually removed because there was no practical application.
Rebuilding the system faces obstacles including incompatible data across platforms, privacy protection, differential treatment of consumers, concerns over big-data price discrimination, and false positives. Other than membership tiers, platforms have not introduced a market-wide consumer credit rating system and are still discussing the idea with experts and regulators.
For confirmed bad actors, a platform can require real-person verification, restrict login or purchases, and withdraw benefits such as instant refunds and return-shipping insurance. But black- and gray-market accounts may not correspond to real identities in the first place; once an old account is banned, a new one can be registered cheaply. Governance therefore remains a contest in which “the devil always has one more trick.”

11. Merchant-Side Problems Include Store Networks, Inferior Goods, and “Take the Money and Run”

Wei Shijie observed that apparel merchants are increasingly refusing to ship. Yuan Bu first explained the legitimate operating model: apparel sellers need to test designs frequently, so a SKU may carry only 30-50 units, or 100-200 at most. Once a hit sells out, replenishment can take more than 10 days, while stocking 500 units risks leaving the entire inventory unsold.
At the other extreme are malicious store networks: one entity controls hundreds or even thousands of stores and posts thousands or tens of thousands of products a day to occupy free exposure. Some networks share supply chains and customer-service teams and are essentially traffic strategies; others use low deposits to create room for nonperformance.
Yuan Bu gave an example: a store sells 10,000 orders at 50 yuan each, collecting RMB500,000 upfront. Normal fulfillment might cost RMB200,000, but the operator spends only RMB50,000 shipping inferior goods and then shuts the store. With a large supply of replaceable accounts behind it, abandoning one store carries little breach cost.
Taobao no longer supports zero-inventory fulfillment or malicious store networks and has issued rules to govern them. Yuan Bu says the core damage is “bad money driving out good”: merchants investing in brands, physical operations, and long-term businesses bear higher costs, only to be pushed out of the spotlight by mass-produced copies.

12. Taobao Followed Refund-Only for More Than 6 Months, Then Cut Volumes by 80-90%

Yuan Bu said Taobao discussed whether to follow Pinduoduo in 2021, but internal teams considered the policy unreasonable and wanted to preserve fairness. By 2023, competitive pressure had changed the calculation. Prompts such as “Would you like a refund-only?” began appearing proactively in chat, allowing consumers to end disputes before the merchant agreed.
The prompts ran for more than 6 months before being withdrawn. The platform found that high-ticket, branded, and differentiated products retained value for return and resale, making the low-value-goods logic impossible to apply wholesale. Consumers did not necessarily welcome the automatic suggestion either; some explicitly wrote, “I don’t want a refund-only,” fearing that the merchant would pursue them offline.
Starting last July, Taobao first loosened the policy for merchants with experience scores above 4.8, then expanded it to scores above 4.6 before the Lunar New Year. Where the after-sales entry once displayed “refund-only” or “return and refund” directly, it increasingly defaulted to return and refund; quality disputes also received a second confirmation step. The proactive prompting of refund-only largely stopped.
As of the point discussed in the episode, non-essential refund-only scenarios had been largely eliminated, with volumes down roughly 80-90%. The remaining cases were limited to products with “fake-crack” issues, goods unsuitable for return, fresh products that would spoil upon return, or items priced below 10 yuan when shipping could cost 10-20 yuan. Yuan Bu called the decline not modest but “massive.”

13. Restoring the Ecosystem Requires Platforms to Take Back Governance Costs

Internal observations after the loosening suggest that merchant appeals may have fallen by more than 60%, while negative sentiment around refund-only and malicious returns declined by more than half. Around Singles’ Day, the platform could intercept roughly 400,000 malicious refund orders a day. Yuan Bu stressed that individual merchants still encounter problems; governance is not finished.
Taobao also uses data shared with Xianyu to identify “malicious refund and resale.” If a user obtains a refund-only outcome on Taobao and then lists the product on Xianyu, later refund requests may no longer be supported. If the platform misses the abuse and the merchant suffers a loss, it may cover the merchant and simultaneously tighten the account’s purchase access and other benefits, including instant refunds and return-shipping insurance.
Yuan Bu highlighted the appeal of refund-only for platforms: automation reduces the costs of reviewing evidence and mediating customer disputes. If a quality complaint is handled with an immediate refund, spending on front-end product screening and compliance governance also falls. “It is a very effective way to shift costs,” but those costs ultimately land on merchants and other consumers.
His conclusion is not to abolish consumer protection, but to correct an excessive imbalance. Regulatory language has shifted from preventing “involution-style vicious competition” to comprehensively addressing it, signaling a policy retreat from the extremes. Platforms should also stop copying competitors blindly: “After 3 or 4 years, learning from it won’t help—you won’t become it.” Future competition must return to each platform’s supply, user mindshare, and high-quality growth.

Fixing the American Dream with Andrew Schulz Snipd: The AI Podcast App for Learning — with CEO Kevin Ben-Smith