This is an English adaptation of a FoodBud historical article originally published on December 4, 2022.
A company’s ability to make money quickly may depend on timing and its ability to capture an opportunity. Its ability to make money over the long term depends more on whether it has a competitive barrier, a core capability, or what investors often call a moat.
As competition intensifies across sectors, more companies are finding profit harder to earn and are paying closer attention to whether they have defensible advantages. This article uses Warren Buffett’s moat framework as a starting point, then applies it to foodservice and SaaS businesses.
Buffett’s framework can be summarized into six broad types:
Intangible assets include brand, patents, legal licenses, and know-how. Know-how refers to specialized experience, methods, or “recipes” that are not widely understood in an industry.
Switching costs are themselves a barrier when users find it expensive or inconvenient to move. A mobile phone number or WeChat account may look easy to replace, but the real cost is getting everyone who contacts you to remember the new one. Moving from Apple to Android also involves habit changes and data migration. Online-document and cloud-storage apps naturally benefit from this type of moat.
Technical barriers become real moats only when the technology is difficult and time-consuming to build. Examples include Qualcomm and Intel in semiconductors, GE Moto in engines, and Google in search algorithms.
Network effects mean that as the ecosystem grows, each participant gets more value. There are single-sided network effects, such as WeChat and QQ, and two-sided network effects, such as Taobao, Didi, Focus Media, and Lianjia. Platforms with strong network effects often display winner-takes-all dynamics. This is why Meituan Waimai and Didi used heavy subsidies to scale quickly, cross the network threshold first, and build dominant positions.
Scale effects usually mean that the larger the supply side becomes, the more value users receive. Didi is more convenient when there are more drivers. Baidu Search becomes more useful when it indexes more content. Businesses with network effects often also have scale effects, such as Ctrip and Meituan Waimai, but the reverse is not always true. Intensive agriculture, for example, has scale effects without necessarily having network effects. Scale effects often require capital or technology to break through quickly.
Cost advantage matters most when price is a primary purchase driver, and it is more common in categories with limited product differentiation. Galanz’s microwave price war in China’s appliance sector is one example: it used manufacturing scale to reduce costs and set market prices close to rivals’ factory-gate prices, forcing many competitors out.
China’s manufacturing rise over the past three decades was partly driven by a sustained, large-scale labor-cost advantage. Today, China faces competition from manufacturing migration to Southeast Asia, where labor costs can be lower.
At its core, a corporate moat is about finding or creating the key competitive resource in an industry, then trying to control it.
Applying Buffett’s framework to restaurant companies shows why foodservice is structurally difficult.
Switching costs are weak. Diners switch preferences easily; taste fatigue and emotional drift are normal.
Network effects generally do not apply. Restaurant chains are not platform businesses and do not usually have single-sided or two-sided network effects.
Scale effects are limited for most operators. Foodservice is a distributed-store business, and most restaurants rely on fresh or short-shelf-life ingredients prepared on demand. That differs from industrial production, where centralized batch processing and scaled procurement can be more readily applied.
Technical barriers are also weak. Foodservice is more craft than technology. There may be some resource-based barriers, but relatively few resources qualify as critical and defensible.
Restaurant locations are usually not exclusive. This differs from cinemas or KTV venues, where a shopping mall may support only one operator in the category.
Ingredient supply can create a resource moat for some brands, especially those emphasizing product quality and origin. Examples include Gutian Daoxiang, a Chinese fast-food rice chain that highlights Wuchang rice from Northeast China; Ningji, a fast-growing hand-shaken lemon tea brand built around perfume lemons; Baheri and Zuo Ting You Yuan, which use yellow cattle beef for Chaoshan fresh beef hotpot; and Fengmao Skewers, which uses Sunite lamb from Inner Mongolia’s Xilingol grassland.
These companies choose relatively distinctive core ingredients and move upstream to control or secure supply. That gives them some moat value. The strength depends on scarcity and difficulty of substitution. Perfume lemons, for example, have seen strong demand because of lemon tea’s popularity, which has accelerated domestic cultivation technology and planting-area expansion. As output rises, that resource barrier is expected to weaken.
Building a moat through resource control has a high threshold. Operators need deep product-development and ingredient research, must identify scarce or hard-to-substitute inputs with origin or cultivation barriers, and for chain businesses must still balance supply capacity with store demand.
Restaurant companies that pursue resource moats often move upstream. This reflects a broader industry pattern: upgrading in the tertiary sector often moves into the secondary sector, while upgrading in the secondary sector often moves into the primary sector. The author speculated that traditional agriculture could contain significant opportunity over the next 10 to 20 years as more capital, research, technology, and higher-skilled talent from the secondary and tertiary sectors enter agriculture, forestry, animal husbandry, and fisheries.
Cost advantage is a high-threshold moat and is usually more relevant to primary and secondary industries. It is harder for service industries such as restaurants. One notable exception cited in the article is Mixue Bingcheng, the tea-drink chain known as the “Snow King,” which was then preparing for an A-share listing.
Based on Mixue Bingcheng’s prospectus, its 2021 revenue structure was 72% ingredients and 16% packaging materials, together close to 90%. The key product categories included:
From the financial statements, Mixue Bingcheng looked less like a made-to-order tea-drink chain and more like a food supply-chain company serving fresh tea drinks and desserts.
Given Mixue Bingcheng’s low-price strategy of RMB 6-8 per store SKU, the author estimated that if terminal gross margin averaged 70%, then based on RMB 10.3 billion in 2021 supply-chain revenue, terminal sales across more than 20,000 stores nationwide would have reached RMB 34 billion. The prospectus disclosed RMB 21 billion in 2021 tea-drink store sales. Using unofficial data that ice cream sales and tea-drink sales were roughly 1:2, the combined estimate would be RMB 31.5 billion, around 10% below the RMB 34 billion estimate. The article therefore judged that Mixue Bingcheng’s terminal product gross margin was roughly 64%-68%.
The prospectus disclosed RMB 10.3 billion in 2021 revenue and RMB 2.5 billion in profit, implying a 25% supply-chain profit margin. Using a simple calculation with an RMB 8 terminal SKU price and 68% terminal SKU gross margin, the single-SKU terminal cost would be RMB 2.56, including ingredients and packaging. The actual supply-chain cost would then be RMB 1.92, including procurement, storage, production, warehousing, and delivery logistics to stores nationwide.
The article argued that a new challenger in this price band would need a cost structure at least 20% lower to break Mixue Bingcheng’s cost-advantage moat.
Mixue Bingcheng’s cost advantage protects franchisee profitability at the RMB 6-8 price band and helps defend the market against third-party entrants.
That cost moat comes from two areas.
First, the company focuses SKUs and core ingredients, builds its own production bases, and sets up factories near raw-material origins. This helps secure supply and product quality, improves bargaining power with upstream suppliers, and lowers cost. Producing near origin also reduces raw-material transport loss and procurement cost.
Second, it has built warehousing and logistics bases and, through logistics partners, a transport network that broadly covers China. Its dense store-opening strategy further improves delivery efficiency and reduces distribution cost.
Low-cost production supply chains plus efficient, low-cost delivery supply chains form Mixue Bingcheng’s cost-advantage moat.
The article also speculated on two possible ways Mixue Bingcheng could strengthen that moat: moving further into primary agriculture, such as sugar, dairy, and fruit cultivation or farming; and developing automated tea-making equipment to reduce store labor costs, extend opening hours, support smaller stores, and potentially create future equipment revenue.
Although most restaurants cannot build a cost-advantage moat, the article identified an efficiency opportunity for some Chinese restaurant companies: prepared dishes.
Prepared dishes include two tracks: 2C and 2B. 2C prepared dishes target home dining, stay-at-home consumption, one-person meals, and health-and-convenience occasions. 2B prepared dishes give Chinese restaurant operators a way to improve store output efficiency.
Restaurant efficiency is often measured through labor productivity, space productivity, and product productivity. Traditional Chinese restaurants rely heavily on on-site preparation, so chef productivity has a major impact on store cost and profitability. Different dishes require very different chef time because of differences in ingredients, processes, and taste requirements.
The core value of 2B prepared dishes is raising chef productivity for the dishes with the lowest kitchen efficiency. The article expected that over the following 3-5 years, dishes with many preparation steps and long cooking times would increasingly be converted to prepared formats, potentially improving overall back-of-house efficiency in Chinese full-service restaurants by 50%-200%. It excluded Michelin restaurants and high-ticket fine dining from this generalization.
The author viewed the traditional front-store/back-kitchen workshop model of Chinese full-service restaurants as entering an industrial-upgrade window. Restaurants with many low-efficiency dishes could capture an early efficiency dividend and improve profit by upgrading prepared-dish capabilities first.
The hard truth is that many restaurant categories lack strong moats. But one moat remains available to most restaurant companies: brand.
Among intangible assets, patents and legal licenses are generally unavailable as barriers for restaurants. Brand, however, is the one moat most foodservice companies can build. Many established and emerging brands in different restaurant segments have already done so.
Brand building is not instant. It is a system, but the article highlights several priorities: strong positioning, effective use of new-media traffic platforms, building private-domain traffic pools, operating those private channels well, and deepening customer interaction and relationships.
A brand can increase revenue and lower marketing cost. It builds consumer recognition, increases trust, supports product premium, reduces customer choice cost, brings more traffic and sales, and lowers marketing investment.
A brand can also reduce operating cost. It may help secure better rent and lease terms from landlords and better supply and service pricing from suppliers.
A brand can attract stronger talent, lowering effective labor cost.
Brands can be copied in appearance, but they cannot be replicated. For restaurants, brand may be the most broadly available moat.
The article then turns to SaaS. From the second half of 2021 into 2022, China’s SaaS sector cooled sharply, and many SaaS companies struggled in a cold and crowded market.
Many founders describe B2B SaaS as slow work, requiring long-term thinking and a commitment to doing difficult but correct things. The article argues that those statements are only useful if operators ask what is slow, what the long cycle is for, and what the difficult-but-correct work actually is.
One answer is moat building. SaaS is slow because building a moat takes time. The long cycle should be used to build the company’s long-term moat. Building that moat is the difficult but correct work.
SaaS means Software as a Service, so the article separates the moat discussion into software moats and service moats.
For software, brand can work as an intangible asset, but it takes time, reference customers, and meaningful share in the target market.
Patents are more relevant to foundational software and hard-technology products. Industry application software and general-purpose software usually find it difficult to rely on patents as moats.
Licensing can work in certain sectors, such as government, public utilities, healthcare, state-owned enterprises, or industries requiring specific qualifications. It generally does not work for other vertical or general software.
Switching costs in SaaS come mainly from two areas.
The first is data migration cost. For non-private-deployment SaaS, the portability and migration cost of customer data assets matter. Customer-owned asset data should be migrated without loss because SaaS data ownership belongs to the customer. Static data, such as member identity attributes and relatively stable tags collected by stores or entered by members in a CRM system, can also usually be migrated without loss. Dynamic data is different. Tags generated from continuously refreshed behavioral insight, such as product-preference labels used for precision marketing, require long accumulation and correction. These are difficult to migrate or reproduce in a new system, even if transaction records are exported. SaaS products that create, apply, and prove business value from such dynamic data can raise switching costs.
The second is functional migration cost. Standard SaaS functionality rarely creates high switching costs because feature homogenization is common in China’s software market. Custom development is different. Large customers and key accounts often request custom functionality. Many Chinese SaaS companies accept some customization. Heavy customization raises the customer’s barrier to leaving. This creates a tension: the more standardized the SaaS product, the more efficient R&D becomes, but the easier it is for the customer to leave; the more customized the product, the less efficient R&D becomes, but the harder it is for the customer to switch.
PaaS or low-code capabilities that enable efficient customization can therefore strengthen a SaaS company’s moat.
Technical barriers are limited for most industry application or process-driven management software, unlike operating systems, databases, engineering software, or AI algorithm products. However, leading companies can work with major customers and industry associations to establish application software standards and certification systems, indirectly increasing participation difficulty.
Network effects exist in some infrastructure software. Tencent Meeting and DingTalk benefited from online classes, online meetings, and remote work, which brought many users into their ecosystems. Some information-management software can also have network effects. For example, if someone shares a mind-map file, others may install the corresponding software to use it. Vertical industry applications generally do not have network effects unless the product supports upstream and downstream collaboration, such as procurement-platform applications.
Scale effects are clear in standard software. The more installations and subscriptions a SaaS product has, the more R&D cost can be amortized. New features used by more customers can also improve activity and retention.
Cost advantage has become more important as capital pulled back and markets changed in the prior one to two years. SaaS cost pressure mainly comes from sales productivity, R&D productivity, and customer acquisition cost, or CAC. Between two SaaS companies in the same vertical, serving the same customer group with comparable products, the one with better productivity and lower CAC has stronger competitiveness.
The article defines service mainly as SaaS operations, customer success, and value-added services derived from SaaS. Some operational services can also exist independently of SaaS.
For services, brand can again be a moat, but it requires time, reference customers, and target-market share. Patents do not apply. Licensing can apply in certain sectors such as government, public utilities, healthcare, state-owned enterprises, or qualification-gated industries.
Switching costs are harder. Services built on industry know-how, professional staff, methodology, or SOPs can be advanced, but they are hard to defend because people, methods, and processes can be copied or transferred.
Marketing agencies and consulting companies often face consultants leaving to start their own businesses. Their main moat is brand, such as Ogilvy among international 4A agencies or Hua & Hua among well-known Chinese marketing companies.
Agency-operation companies, including Tmall TP operators and food-delivery platform operators, also face peer competition or clients bringing operations back in-house because the industry lacks strong moats.
Technical barriers are also weak for most services based on know-how, methodology, SOPs, human creativity, or human execution. RPA-based BPO services and AI outbound-calling services were emerging, but the article argued they did not yet present enough technical depth to create strong barriers.
Network effects are possible in supply-demand matching services or SaaS-plus-service models. Flexible labor services are one example: retail and restaurant stores create labor demand, while staffing companies and training schools supply workers, with SaaS or an app enabling matching. This is a typical two-sided network-effect model.
The article also cites Dayu U Service, a company using SaaS to provide nationwide on-site service engineers for IT companies and electronics brands, covering installation, training, repair, and inspection. This is another SaaS-plus-service model with two-sided network effects.
For network-effect businesses, the key is to break through the network threshold quickly and create an accelerating flywheel. If capital or resources are limited, the article recommends focusing on the region with the strongest supply or largest demand, concentrating resources to break through one side first, then replicating and expanding.
Service businesses can also have scale effects in three areas.
First, direct service productivity. In a call center, higher call volume and a more diverse customer base make it easier to maintain all-day response quality, optimize scheduling, and improve labor efficiency and service cost.
Second, indirect enablement productivity. Professional service systems often include training, knowledge management, and quality-control teams behind the front line. As the front-line service team scales, these back-office costs are spread more widely.
Third, case-practice effectiveness. A larger and more layered customer base increases the applicability and reuse of best practices and cases. This is similar to the “brand and performance” concept in marketing and to the software logic that frequently used functions produce higher development efficiency.
The first two mainly affect service cost and have limited customer-perceived value. The third is more customer-value oriented and can have some moat effect if the company breaks through the threshold and communicates that value effectively.
Cost advantage in operations and services has historically relied on labor, process, and knowledge. Labor cost remains a major pain point. Many methods can reduce labor cost temporarily but do not create durable cost moats. Recruiting junior staff and training them intensely may lower cost in the short run, but competitors can imitate the model or hire away the trained staff. Over time, income and market value tend to converge. The same applies in consulting: if a company trains management trainees into strong business consultants, many may later join competitors for better pay.
A more sustainable path is automation through RPA, though the article notes this was mainly limited to basic services and widely used in scaled BPO outsourcing.
The article closes with two exploratory thoughts on where service and operations moats could emerge.
The first is the evolution from human work to automation. RPA replacing basic BPO work is one example. The author divides operational services into three levels:
RPA automates the first level. The next likely upgrade is analytical service intelligence. In recent years, BI, BA, and insight products have grown. As analytical AI develops decision logic and capability closer to human reasoning, “BI + AI” could create closed-loop analytical AI for business operations. Today’s expert experience, best practices, analytical models, and human decision-making could become inputs for AI-driven decision computation and algorithms. Companies that validate these models and produce results first may build future moats.
Creative services, including creative concepts, design, copywriting, and planning, may also become more automated. The article notes that in 2022, generative AI in the United States had become a hot area, with AI-made paintings and animation reaching a professional-looking level. Marketing ideas, visual design, corporate VI, copywriting, and planning might in future be produced quickly after a simple instruction. The author calls this a possible “wisdom upgrade” of operations and services, and suggests it may have technical-barrier potential.
The second thought concerns the relationship between service providers and clients. Traditional service and operations businesses usually sit opposite the client in a Party A/Party B relationship. The article asks whether a new moat could be built by reconstructing that relationship.
B2B can be viewed one level deeper as business-person-to-business-person. Serving an enterprise means serving people in different roles inside that enterprise. If a SaaS or service provider can form a shared-interest community with the client’s operators, with benefits designed around job, position, career, and social roles, then those client-side people may become part of the moat.
Companies attack through innovation and defend through moats. The article’s final view is that downturns may be the best window for companies to think about and build their moats.
Note: IPO, prospectus, financial, and forward-looking figures in this article are historical references from the original 2022 source.