Discussions about Luckin Coffee have been polarized from the start. Supporters have remained optimistic, especially when the Q3 earnings report announced "per-store profitability" last November, which was met with a very positive response from the secondary market. Detractors are equally steadfast, citing the unsustainability of the cash burn and numerous doubts surrounding its rapid IPO. Now, a short seller report seems to substantiate these concerns.

First, this is not a short seller report researched and analyzed by Muddy Waters itself. On the evening of February 1st (Beijing time), Muddy Waters tweeted that they had received an anonymous 89-page report claiming Luckin Coffee is a fraud, and attached the full report. Muddy Waters conducted a preliminary assessment and deemed it credible. I downloaded the report for learning purposes, as it addresses many of my long-standing questions about Luckin.

On the evening of February 3rd, Luckin responded positively via the SEC, firmly denying the short seller report. The details are widely available online, so I won't repeat them here. However, in my view, this somewhat brief response is very "A-share" in style—confident in tone but lacking evidence.

1. Reasonable: Indeed, all orders are placed online, so key sales data can be verified through system-level third-party payments, and technical due diligence is easy. Luckin could prove its innocence by opening its systems. 2. Lacking evidence: The announcement did not directly rebut any of the specific allegations in the short seller report, including operational data, business model, or capital maneuvers. It only made general denials without explanation.

So, I was somewhat disappointed with Luckin's response. However, this means there is value in continuing the analysis and discussion. I welcome your corrections and feedback.

I have a coffee-drinking habit, having tried everything from simple Nescafé instant to pour-over. Now I have two fixed habits: at home, I brew with instant cold brew powder like San顿半 and add hot milk; outside, if I need to meet someone, I go to a café like Starbucks; if I just need a takeaway, I usually choose Luckin or convenience store coffee. There are exceptions: in Shanghai, I prefer niche chains like Peet's, Manner Coffee, etc.

This represents a typical consumer group like me, for whom coffee serves both functional and enjoyment needs. For all coffee drinkers, I strongly agree with the report's categorization (page 69 specifically refers to "Chinese consumers"), which can be roughly divided into the following types:

1. Space: Those who want a space for gatherings, meetings, or work while enjoying a cup of coffee, e.g., Starbucks, Costa, Pacific Coffee. 2. Style: Those who want a refined, flavorful coffee, e.g., % Arabica, Starbucks Reserve, or specialty coffee shops. However, I believe Reserve stores in China are not yet specialty coffee; people go there more for the first category and to buy exquisite accessories. 3. Beverage: Those who want a caffeinated drink, which could be coffee, tea, or just juice. This group typically orders mochas or sweetened lattes, and may also like milk tea, e.g., 一点点, Coco. 4. Functional use: Those who want a caffeinated drink to stay awake and energized, usually coffee addicts. This group also goes to convenience stores like FamilyMart, 便利蜂, or fast-food chains like KFC, McDonald's. 5. Coffee lover: Those who want a very high-quality coffee, true enthusiasts, the smallest niche.

Based on this categorization, let me dissect the core logic of the report and my views.

-01- Starting with Luckin's Positioning A successful consumer brand must determine its positioning from Day 1. Value positioning and price positioning are vague, but audience positioning is concrete and usually defines the former. If a company understands the real needs and core pain points of a specific group, it can certainly win customers and get them to pay. So, what is Luckin's audience positioning?

I believe it has always been the fourth category: functional users. Functional needs do not mean zero taste preferences. When Luckin burst onto the scene in 2018, it made Starbucks tremble. On the surface, it did take some customers from Starbucks, but in reality, Starbucks is too expensive—more precisely, its price-performance ratio is low.

For those with functional needs who do not want to lower their taste to instant coffee, and when convenience store fresh-brew coffee outlets are not yet widespread in most Chinese cities (except Shanghai), Luckin found a narrow window of opportunity. To quickly capture this group's mindshare, it used pervasive advertising and sustained aggressive subsidies (the report uses "cut-throat," which is apt).

So, price-performance and functional needs are Luckin's tags, consistent with the footnote under its logo: "Professional coffee, fresh style."

This discussion of positioning naturally leads to the core conclusions of the short seller report. Let's look at them:

1. Luckin's positioning for functional users is wrong because this market is actually very small and slowing, and young people with caffeine needs are largely satisfied by tea shops, making competition fierce. 2. Luckin's positioning and marketing have created a group of pseudo-coffee addicts who are highly price-sensitive and have no brand loyalty. So if Luckin raises prices, sales will drop immediately. 3. Luckin's profitability depends on the economics of each store, and per-store profitability basically depends on sales and costs, where sales = volume x price. Since costs cannot be significantly reduced in the short term, the key indicators for profitability are product volume and price. 4. Luckin has falsified both volume and price per store, and actual data is far below historical financial reports, so stores did not actually achieve profitability in Q3. To offset the inflated store profitability, Luckin deliberately increased marketing expenses recorded at headquarters, and the two amounts are almost equal. 5. Not only are stores currently loss-making, but both core indicators are declining quarter over quarter, making future profitability even more distant. Therefore, this is a fundamentally broken business. 6. Finally, the report makes six warnings, including a piercing analysis and speculation on management's flashy capital maneuvers and the dark backgrounds of key directors, concluding that the company faces significant credit risk overall.

In short, the report argues from two major aspects:

  1. "Fraud": key operational metrics are falsified, including other product revenue and marketing expenses. 2. "Fundamentally broken business": not only current losses but also future losses.

It draws two important conclusions:

  1. Wrong positioning: the niche market is not large enough, actual demand is insufficient, and competition is fierce. 2. Business model invalid: key metrics—volume and price—are far below expectations, making profitability unattainable.

Having discussed positioning, let's move to the market: Luckin, Starbucks, etc., are in the fresh-brew coffee market.

-02- The Fresh-Brew Coffee Market The chart below is from a January article by 深响 titled "Starbucks China in Trouble," citing Frost & Sullivan statistics. The yellow bars represent per capita annual fresh-brew coffee consumption in China, showing growth of about one cup per year. Even so, by 2023, per capita will only reach 5.5 cups. Compared to other countries in 2019—Japan 103 cups, South Korea 88 cups, the US 136 cups—China's per capita fresh-brew coffee consumption has significant growth potential.

The short seller report, in the "broken business model" section, first argues that "functional users in China's coffee market have already been satisfied by tea drinks, and China is a society with a stubborn tea-drinking culture." The reasons given: 1. Asian countries that were colonized by the West have coffee-drinking cultures, hence higher per capita consumption. Other countries with functional caffeine needs, like India and China, have already been served by domestic tea drinks. 2. According to USDA and ICO statistics, China's per capita consumption of unroasted coffee beans grew at a CAGR of only 3-4% from 2015-18, indicating weak growth in fresh-brew coffee consumption. 3. According to Meituan big data, while the growth of new coffee shops has slowed or even declined, the number of offline tea shops is growing rapidly.

I remember when 一点点 opened its first store in Beijing, I, being sensitive, urged a friend to call and inquire about franchising. The answer was that the milk tea market in Haidian District was already saturated (and I actually believed it). As a veteran enthusiast of both milk tea and coffee, I think the author knows too little about China's beverage market and has logical problems. My reasons:

1. Although an earlier coffee culture positively promotes the coffee market, it does not explain why countries like Japan have both the highest coffee consumption and relatively high tea consumption. The report includes a chart showing this clearly but does not explain the contradiction. I think this actually suggests that China's total caffeine intake from tea and coffee combined has significant room for growth, not just potential. In 2017, per capita total intake was 81mg, vs Japan 148mg, South Korea 90mg, and the US 162mg. 2. Slower growth in per capita unroasted coffee bean consumption does not directly equate to actual fresh-brew coffee sales performance. The reason is simple: there is a long, unmeasurable processing chain from unroasted beans to fresh-brew sales, and statistical gaps can occur at any stage. Moreover, with fresh-brew coffee currently only accounting for 16% of the overall coffee market in China, I need to know how much of the coffee bean sales are directly used for fresh-brew coffee shops. If per capita consumption of such beans grows less than 5%, it might indicate that the fresh-brew market Luckin faces is indeed limited. However, the proportion of fresh-brew coffee sales is likely to continue growing, gradually approaching the levels of Asian countries like Japan (where coffee beans and fresh-brew coffee together account for nearly 50%). 3. The Meituan data on tea shops includes all tea-related beverage shops, most of which are actually milk tea shops. Based on my understanding of many milk tea enthusiasts, our demand for milk tea is not for functional caffeine intake. Drinking milk tea and coffee are not in conflict and do not cannibalize each other.

So, the rise of tea shops does not mean that people with caffeine needs have stopped drinking coffee. Finally, the slowdown in fresh-brew coffee shop growth may not be due to "not enough demand" as the report claims, but could be due to poor management, rising rents, and other reasons.

Therefore, China's fresh-brew coffee market has potential for sustained growth, and the report does not provide strong evidence that China's coffee market or fresh-brew coffee market lacks growth momentum.

-03- Suspected Fraud Now let's look at the more important issue: whether key data such as volume and price are falsified. The report roughly says the following, and I'll only highlight the most important first three points:

  1. In Q3 and Q4 2019, per-store product volume (sales) was inflated by at least 69% and 88%, respectively. 2. Items per order fell from 1.38 in Q2 2019 to 1.14 in Q4. 3. The average price per item was inflated by 1.23 yuan.

I found Luckin's Q3 2019 earnings PPT and cross-verified it. Page 16 indeed shows product volume per store per day and provides Q4 guidance.

The short seller report analyzed video footage from a random sample of 620 stores over a total of 981 complete store-days in Q4, meaning each store was monitored for an average of 1.58 days.

The report selected store samples based on store markers in the Luckin app:

  1. Selected 38 representative cities out of 53 covered by the app (covering 96% of stores). 2. Excluded 851 store-days with defects, including equipment failure, missing records over 10 minutes, or being asked to leave by Luckin staff, with an exclusion rate over 40%. 3. City selection roughly respected the proportion of total stores in each city; the top five cities were Beijing, Shanghai, Guangzhou, Shenzhen, and Hangzhou. 4. Time selection: 572 weekdays and 409 weekends. Store establishment: 82% before Q2 2019, 13% in Q3, and 5% in Q4.

From this, they derived:

  1. Average daily orders per store: 230 orders/day. 2. Multiplying by the above (1.14 items per order) gives 263 items per store per day.

I have not systematically studied statistics, and I hope friends with relevant experience can point out the strengths and weaknesses. Here are some of my thoughts:

1. Unreasonable time weighting: Over 60% of Luckin's stores are in office complexes, with the top five cities averaging over 65%. Busy hours are concentrated on weekdays, but the sample only has 58% weekdays. A more reasonable approach would be to first calculate the ratio of weekday to weekend sales over a month and then weight the time selection accordingly. 2. Questionable sample store selection: According to the earnings PPT, Luckin had 3,680 stores by Q3, but the report sampled from 4,507 stores listed in the app at the time, meaning 827 new stores (18% of total) were added. Q3 added 717 stores (16% of total). We know new stores need a ramp-up period to reach stable operations. From this perspective, how should stores be selected appropriately? Should geographic location be considered? I remain skeptical about the rationality of this selection. 3. Monitoring duration per store is too short: As mentioned, each store was monitored for 1.58 days. With an average daily operating time of about 11.5 hours, each store was monitored for only 18.2 hours. I hope friends can point out what needs attention or improvement here, because if any of the above doubts hold, the sample selection becomes biased, and the conclusions derived from multiplication will have larger deviations. 4. The absence of order volume data is the biggest point of contention: The report's second point calculates items per order, multiplied by the actual order count to get total items. This formula is fine, but if items per order is also unfair, then the first point's conclusion is even more suspect. Moreover, I could not find any historical order volume statistics in Luckin's official earnings PPT or financial reports, so the report's claim of quarterly decline in order volume cannot be verified. Thus, order volume is the most important key indicator that Luckin needs to explain.

Next, let's see how points 2 and 3 were calculated. The report analyzed 25,843 receipts from 10,119 customers, covering 2,213 stores in 45 cities, and found:

  1. Items per order fell from 1.38 in Q2 2019 to 1.14 in the survey. 2. The average price per item was only 9.97 yuan, 12.3% lower than the 11.2 shown in the Q3 earnings PPT.

Here we need to verify two things: first, whether the survey conclusion is reasonable; second, whether there is actual fraud. Let's look at the first. With over 20,000 data points, the margin of error is 1% and the confidence level is 99%. Regarding sample selection, the report compared the sample composition with actual stores.

I found two points worth doubting:

  1. The sample overweights T3 cities, but third-tier cities are not Luckin's focus. 2. The sample overweights stores in shopping mall settings. As mentioned earlier, office workers are Luckin's positioning and main market.

If the sample selection is biased, the data only reflects the consumption behavior of customers in this sample, with an error of less than 1%. I hope friends can point out the exact problems or confirm the accuracy of this sampling.

Even if we assume the data from the 20,000+ sample is true and credible, with 1.14 items per order and 9.97 yuan per item, it does suggest Luckin may have falsified data.

Although Luckin's financial report does show Q3 price per item at 11.2, I did not find the items per order data; only the total items per store per day. To get items per order, one needs the daily order count per store. Therefore, the report estimated this number as follows:

The estimation logic is based on back-calculating from delivery orders, which are derived by dividing delivery costs by the fulfillment cost per order. In the financial report, I can find total delivery costs, but not the per-order fulfillment cost. I clicked on the first annotation of "coffee_detective calculation" in the "Source" of the chart above. Aha! I found this: "All numbers were company reported except 'delivery expense per delivery order' was assumed based on http://tech.163.com/19/0326/08/EB6BUEJK000999C0.html."

So, the key figure of per-order fulfillment cost comes from a March 2019 column by NetEase Tech titled "Luckin Mystery." The data is based on an interview with an unnamed investor, Mr. H, who said, "...SF Express delivery costs are 8-10 yuan per order based on market research, which is twice the cost of coffee." I have two objections:

  1. The report assumes fulfillment costs of 11 yuan for periods before Q2 2019, which is inconsistent with the source's interview (8-10 yuan).
  2. According to news from May 8, 2019, Luckin had begun seeking cooperation with Ele.me and other delivery providers, so from Q2 onward, it is reasonable to believe that per-order fulfillment costs may have dropped significantly, at least below 8 yuan.

Therefore, the accusation of falsifying items per order is based on an interview with an external person in a third-party report, which seems not rigorous, and the inference of fraud based on this may not hold.

In summary, we can still draw the following conclusions:

  1. If the sample collection is credible, the average daily price per item per store is indeed falsified, with the real price at 9.97 yuan. 2. Although the survey shows 1.14 items per order, there is no strong reason to believe this number is declining each quarter. 3. Since the decline in order volume cannot be proven, the report's future order volume predictions lack data credibility.

-04- Profitability Discussion Having discussed the key indicators, we can finally return to the fifth of the six conclusions: whether stores are profitable, the conditions for profitability, and whether future profitability is sustainable, given that costs are unlikely to drop significantly in the short term.

We have broken down the core revenue indicators and controversies, and costs can be estimated from the financial report, so the store-level model emerges. Below is the store-level break-even chart from the Q3 earnings PPT, which shows the previously announced per-store profitability indicator: Q3 store profit margin of 12.5%.

In contrast, the short seller report's statistics, based on 263 items per store per day and a price of 9.97 yuan, show an actual store profit margin of -28%. The difference between positive and negative is indeed huge.

At the same time, the report rigorously conducts a classic sensitivity analysis, with price on the x-axis and daily sales volume per store on the y-axis.

Based on this sensitivity analysis, there are several interesting findings:

  1. Assuming daily sales volume is only 263 cups as the report claims, the store breaks even when the average ticket price reaches 13 yuan.
  2. If we use the Q3 financial report's 444 cups, the store breaks even when the average ticket price exceeds 11 yuan, which roughly matches the financial report.
  3. Similarly, if the average ticket price is 10 yuan, the store would need to sell 900 cups to approach break-even, which is nearly impossible, hence the report's conclusion that future profitability is even more unattainable.
  4. However, in the long run, under the current "buy two get one free" pricing strategy, if the final price per cup stabilizes at no more than 16 yuan, say 15 yuan, the store only needs to sell 200 cups to break even.
  5. The conclusion is that the store profitability model is highly sensitive to unit price, and according to the first law of demand in economics, price levels significantly affect store sales.

In short, store-level profitability is theoretically possible.

However, from the official price of 11 yuan to 16 yuan is nearly a 50% increase. Can consumers accept that? This brings us back to the second conclusion discussed earlier:

Are consumers highly price-sensitive, and what does the price-sales curve look like?

-05- Drinking Luckin or Coffee? Look at the "Price" As discussed earlier, Luckin's audience positioning is consumers with functional needs, and it has always been so. They are unwilling to completely give up taste for low prices, and their pursuit of price-performance is increasing. Therefore, on one hand, this group's needs will not be satisfied by the rise of tea drinks; on the other hand, this group is gradually converting from instant coffee drinkers, a process that is irreversible and will continue to grow in the future.

What I am ultimately interested in is: What price or price range is acceptable to this group (including me)? This largely determines the authenticity of Luckin's current and future profitability.

Unfortunately, the report does not provide deeper analysis on this, which is the most regrettable part.

Finally, the report discusses many other topics. If you have the chance, you can read it carefully. For example, although it does not directly analyze price, it uses app activity and retention to argue why most of Luckin's current users are extremely price-sensitive, so when Luckin tries to raise prices, sales will decline further. It also analyzes the sales proportion of products other than coffee, further supporting the core conclusions above.

Returning to the title of this article, do we want to drink "Luckin" or "coffee"?

My view is that we want to drink coffee. And when we go to Starbucks, we want to drink Starbucks.

This is the fundamental reason why Luckin and Starbucks are not in the same competitive arena, at least for now. It is also the fundamental difference between me and the report's author. Moreover, in China's fresh-brew coffee market, there are not only Starbucks and Luckin, but also convenience store and fast-food coffee, as well as emerging chains like Seesaw, Fisheye, Greybox, Manner Coffee, etc. Some have lower prices than Starbucks, most have higher quality, and each has its own positioning.

Luckin faces intense competition, but it is not from tea shops; it is from peers.

In the second half of last year, Luckin launched 小鹿茶 and vending machines, and expanded its snack categories. In my view, Luckin coffee shops are gradually evolving into Luckin "afternoon tea" shops. Whether this business model can succeed requires more operational data to verify.

Source: 我思锅我在 (ID: angelplusdevil)