AI-Era Global Marketing with 沈晨岗 of Feishu Shenno Meetsocial
Summary
沈晨岗认为，2025 marked a clearer inflection point as Chinese companies going global shifted from products to brands, though the transition was not overnight. Rising traffic costs, homogenized competition, and tougher quality and compliance standards across the media and traffic ecosystem have closed off the path of competing ever more aggressively on price and execution. Companies must move beyond immediate sales toward user research, product iteration, and brand mindshare; ultimately, what goes global is not just a good, product, or brand, but “a company with soul and vitality.”
In Feishu Shenno’s D-MES model, Digital is not a parallel capability but a multiplier on sales, product, and brand capabilities. Digitalization improves the efficiency of payments, supply chains, and delivery while also raising the speed and quality of decision-making. 沈晨岗’s competitive view is that the faster a company reads the market, the more accurately it understands change, and the more effectively it acts, the more likely it is to win.
Marketing’s biggest AI consensus is that it is important and should be deployed now, while the biggest disagreement is what it can actually solve. The controversy over McDonald’s Netherlands’ AI-generated holiday ad showed that being able to generate an image or video does not mean being able to deliver commercial creative. If content does not drive likes, registrations, orders, or purchases, what is the point of making it merely for the sake of making it?
AI has materially lowered the cost and barrier to content production, but it still struggles to independently own high-value, low-tolerance-for-error budget and media-buying decisions. An optimizer makes hundreds or thousands of decisions every day about what to keep, what to cut, bids, media, and market budgets. AI currently contributes information from the back end and is gradually moving toward the front end; handing high-risk customer-service or media-buying judgments to AI means a single error could have fatal commercial consequences.
Feishu Shenno breaks creative work into agents for requirements understanding, strategy planning, image or video production, and quality control, with humans orchestrating them into an agentic flow. A creative professional who once produced 5–10 assets a day can now deliver hundreds or even thousands in standardized scenarios. During the Spring Festival, one short-drama client received 6,000 edit concepts in a week, while daily ad spend rose from several thousand dollars to the low six figures.
After mass generation, the scarce capability is judging and improving commercial performance, not generation itself. 沈晨岗 says that “a 1% gap in quality can ultimately translate into a performance gap of more than 10%.” AI can assign hundreds or thousands of tags to creative assets, but it cannot independently create the underlying framework for judging quality; people still need to abstract knowledge, define methods, and teach the model through feedback.
A service provider’s potential moat is not a general-purpose model, but reusable infrastructure, cross-client data, and a digital loop tied to customers’ businesses. Feishu Shenno says its capabilities can be spread across more than 100,000 customers, while its systems capture more than $20M in daily advertising-spend data. Fixing a client’s website or data-collection infrastructure could produce a 10%–20% improvement, far above the 1%–2% typically fought over at the media-buying layer—the effect is “multiplicative, not additive.”
沈晨岗 expects marketing over the next 3–5 years to evolve from standalone ad buying into the digital nervous system of the enterprise. User viewing, liking, registration, purchase, and repurchase will feed back into products, orders, and supply chains, while AI lowers the barrier to global marketing. But the first-mover advantage will depend on whether organizations can genuinely rebuild their processes: “We all think it is hard; others may find it even harder.”
Deep dive
1. The core of Global Expansion 2.0 is not changing markets, but moving up the value chain
沈晨岗 divides the objects of global expansion into four layers: goods, products, brands, and companies. Goods offer only basic functionality; products create integrated value; brands add identity and emotional value; the endgame is “a company with soul and vitality” going global.
A phone assembled from Huaqiangbei components is still a commodity. Adding a white label or logo does not automatically make it a brand. Xiaomi gradually built a product brand and moved toward becoming a global company by combining R&D and product capabilities with its own ethos.
During Feishu Shenno’s first 10 years, its mainstream clients were still focused on selling commodities. 沈晨岗 believes that from 2025 onward, more companies will move from products to brands, marking the shift from Global Expansion 1.0 to 2.0.
The marketing objective is different in the two phases. Previously, the question was whether an investment could pay back the same day, week, or month. Now companies must balance sales conversion with business building and spend more time building durable competitive strength.
2. Rising traffic costs and tighter compliance are forcing companies to stop racing to the bottom
程曼祺 asked whether early-stage companies simply failed to recognize the value of marketing. 沈晨岗’s answer: when buying traffic produced immediate sales, companies naturally maximized that opportunity rather than prioritizing whether users would remember, recognize, and like them.
As traffic costs rise and commodity products become more homogeneous, relying solely on low prices, cheap traffic, and ever-finer sales execution produces declining marginal returns. A seller with annual sales of several hundred million yuan, or even more than RMB1B, and tens of millions of overseas users may still find that the old path has reached its limit.
Moving into product design and user operations requires additional investment. Companies that laid no groundwork in the early years struggle to transform at the inflection point; some manage to move upmarket, while others suffer major setbacks and “even die.”
From the perspective of media and the traffic ecosystem, requirements for product quality and compliance are also tightening. Cost competition once encouraged corner-cutting and regulatory edge cases. That route is becoming increasingly blocked, leaving companies with only harder, longer-term work.
3. Anker shows that brand capabilities must be built before the selling phase is over
沈晨岗 uses Anker as an example. With limited resources, its founding team had to begin with commodity sales, but invested in R&D as revenue accumulated, continually improving concrete product capabilities such as charging-cable connector manufacturing and power-bank performance.
The key is not to wait until after success to start talking about brand. It is to be “thinking about when to prepare for the second step while taking the first step.” Product advantages and consumer memory and affection accumulate together, eventually creating durable differentiation.
He also sees a generational shift. The first generation of entrepreneurs is constrained by existing production, sales, and market assumptions. The second generation may have received more market-oriented and advanced marketing education overseas, making it easier to start from consumer needs, launch a separate global brand, and reorganize product and marketing in a different way.
4. The cognitive shift in 2.0 is from “what do I make?” to “what does the user need?”
The commodity logic is: “I have made something; now I need to find a way to sell it.” The product and brand logic starts by understanding what overseas users need and then deciding what to produce. 沈晨岗 sees this as the fundamental difference in the starting point for sales between the two generations of operators and the two phases of global expansion.
Operators entering the product and brand phase continuously track what consumers care about, how products should evolve, and how to meet demand while generating profit. User research is no longer a one-off exercise before launch.
Anker and Xiaomi conduct extensive user research every year. Feishu Shenno also provides global digital research for companies, feeding market feedback into product evolution so they can stay as far ahead of competitors as possible.
5. D-MES presents the value chain in reverse, while D multiplies every capability
Feishu Shenno uses D-MES to assess the potential of global brands: S stands for Sales Performance, E for Product Evolution, and M for Mental Dominance. The three correspond to commodity sales, product iteration, and brand mindshare.
The sequence is a reverse description of the goods-to-product-to-brand path. The acronym is catchy, but it also presents a structure for describing corporate competitiveness backward from sales results, product capabilities, and brand mindshare.
Digital Competence is treated separately because sales, R&D, consumer mindshare, and business operations all require digitalization. 沈晨岗 says D is a multiplier, not a fourth additive component layered alongside the others.
He cites L’Oréal as an example. Its channel, product, and brand capabilities are all strong, yet over the past 3–5 years it has still invested enormous sums in connecting online and offline stores and business processes. At the top end of competition, digitalization may be the core gap supporting every visible capability.
6. Digitalization first improves process efficiency, then decision quality
The first layer of digitalization is greater efficiency within an existing framework. Starbucks links its mini-program, electronic payments, membership records, purchase history, and promotions, shortening the transaction process while allowing loyalty operations to feed user data back automatically.
Similar upgrades cover supply chains, product delivery, and product design. Tasks that once depended on lengthy coordination across teams can be completed more efficiently through digital systems.
The second layer is better and faster decision-making. A digital supply chain can make decisions more quickly and at higher quality. Meituan also relies on extensive data to allocate riders, capacity, time, and incentives, then continuously adjusts based on the results.
沈晨岗’s competitive conclusion is straightforward: among companies operating at the same level, the one that detects market and consumer changes faster, understands them more accurately, and acts more effectively is more likely to win.
7. The consensus is to adopt AI immediately; the non-consensus is what it is for
沈晨岗 uses a cartoon from a management group to summarize the current mood: “Is AI important? Very important. What can AI solve? Don’t know. When should we deploy AI? Now.” There is broad agreement on its potential value, but the business boundaries remain unclear.
His approach is not to wait for complete answers, but to run cost-conscious experiments: define the objective and expected output first, then decide what should be handled by AI, by people, or by a combination of both.
程曼祺 mentioned that McDonald’s Netherlands pulled an AI-generated holiday ad after viewers found it cheap, lazy, and lacking sincerity. 沈晨岗 sees the episode as a severe mismatch between capability and objective: a generative tool can make a film, but that does not mean it can deliver the commercial communication a brand requires.
Commercial creative must answer whether the goal is to make users like the content, like it publicly, register, place an order, or buy. Even a visually appealing piece is pointless if it does not serve those objectives: what is the purpose of making the ad?
8. Low tolerance for error in high-value decisions limits AI’s move from assistance to accountability
Basic customer service relies on general knowledge, which AI can handle relatively easily. Professional decisions such as whether to spend money on a particular media channel are different: a single error can seriously harm a client, and average accuracy cannot conceal tail risk.
沈晨岗 compares the issue to autonomous driving. Even if a system’s probability of error is lower than a human’s, regulatory and liability questions do not disappear when the consequences of one accident are sufficiently severe.
The more fundamental constraint is who takes responsibility for the commercial result. AI has difficulty assuming responsibility for high-value decisions, so high-risk scenarios cannot be handed directly to it; human review and judgment remain indispensable.
9. ChatGPT changed knowledge distribution; Claude Code showed a leap in capability
ChatGPT’s first shock for 沈晨岗 was not an immediate view of how marketing would change, but the realization that knowledge had been generalized and could be distributed in personalized form at an efficiency that was previously unimaginable.
Customer service became one of the earliest deployment scenarios because it is relatively expensive, requires comparatively limited knowledge or capability, and naturally fits a conversational format. Once work becomes complex or high-value, hallucinations and the absence of clear judgment standards quickly emerge.
He admits that the industry’s understanding has swung back and forth. At first, he thought AI would “solve problems immediately.” Later, he realized it could not simply replace knowledge work, and that commercial adoption would advance layer by layer.
He believes Anthropic’s AI coding over the past 2 years—especially Claude Code—represents a highly revolutionary change. Work that might once have required 100 engineers can now be completed by an AI coding agent in a short period and at a relatively high level. In his observation, leading technology companies globally have already handed most programming work to AI coding, with people focusing more on review, validation, and management.
10. Marketing AI is changing content production first; humans still steer strategy and budgets
Text, image, video, and creative production are the first areas to be affected. Production that was previously expensive is now cheaper, more accessible, and more efficient, making it possible to deliver large volumes of creative assets in a single batch.
Deeper data strategy is still evolving. A marketing optimizer faces hundreds or thousands of decisions each day: whether to keep an ad, adjust a bid, reduce the Google budget, or increase the Japan budget. The accumulated judgments directly change the final result.
Because spending decisions have very little room for error, AI currently works mainly in the background to organize information and support judgment, then gradually moves “from the back end to the front end,” rather than directly taking over account-level decisions.
沈晨岗 compares AI to a recent college graduate or PhD. It must learn from a mentor why a particular placement or adjustment was made, then encode that experience into a business model. Calling a general-purpose LLM cannot skip this training.
11. China may move faster on applications because its business baggage is lighter
During visits to Meta, Google, Nvidia, and multiple startups in Silicon Valley, 沈晨岗 saw AI discussed everywhere in cafés and large numbers of big-tech employees starting companies. The U.S. places more emphasis on foundation models, including embodied-intelligence models and research built on foundation models.
His observation is that China is moving faster in applications across some industries. Consumer and To B scenarios are more numerous, competitive shifts happen faster, and companies have stronger incentives to embed AI into real workflows.
AI is an intellectual capability, not an add-on button. To create value in cost, efficiency, or outcomes, companies must redesign business and service processes in the same way they did when the internet emerged.
The resistance encountered when Meituan’s restaurant-management system entered the U.S. illustrates the burden of incumbency. Local stores, payment settlement, and upstream and downstream partners are already embedded in legacy systems. Even if a new process is better, the cost of adjusting the whole system remains enormous. China’s existing business processes are generally less mature and fixed, making them easier to restructure.
12. Data fragmentation at traditional ad groups could become a structural burden in the AI era
Feishu Shenno began building its first data warehouse in 2015 using HP and Tableau systems, investing approximately RMB1M–2M at the time. 沈晨岗 says that was still “a huge sum” for the company.
From day one, the company began consolidating media, creative, ad operations, and other marketing activities into a unified system. Once AI arrived, it could layer on computation and analysis without first rebuilding the basic data infrastructure.
By contrast, customer and business data at traditional large advertising groups is often distributed across different subsidiaries that are neither connected nor free of internal competition. Client service can theoretically be integrated, but the underlying digital connections do not truly exist.
To recommend a budget, an agency must know where historical funds were spent, what worked, and what did not. 沈晨岗 believes many traditional institutions must first “break themselves apart and rebuild themselves” to use AI effectively. That organizational turn is close to impossible—like Nokia knowing the direction but watching itself fail.
13. Feishu Shenno is not rushing to sell agents because the outcome is not clearly defined
沈晨岗 first asks what an AI agent actually is. Market definitions often do not map to a clear task or commercial result and can promise only abstract capabilities such as “conducting market research.”
If the research merely searches for and summarizes competitor information, it still cannot answer what counts as acceptable, whether a client can use it, or what value it ultimately creates. Both the agent’s capability boundary and the way its results are measured remain too vague.
Feishu Shenno therefore keeps many agents inside the company for joint use with human experts. Experts turn intermediate outputs into results that clients can understand, receive, and hold someone accountable for. Claiming that the technology already enables clients to create value on their own is, in his view, “an irresponsible statement.”
14. Creative work is being decomposed into an agentic flow, enabling 6,000 assets in a week
A commercial creative project can be broken into small tasks including requirements understanding, goal identification, strategy planning, content-format selection, image or video production, logo placement, and quality control, with different agents assisting at each node.
A product manager or project manager orchestrates the agents and embeds human judgment into a standardized process. The client inputs the desired objective and ultimately sees a high-quality deliverable; the middle layer is an agentic flow combining people and AI.
A creative professional who once produced 5–10 assets a day can now produce hundreds or even thousands, depending on the degree of standardization and the complexity. Every node continues to be digitized, with results fed back to the agent for learning while humans tell it what is good and what is not.
During the Spring Festival, the team delivered 6,000 short-drama video-edit concepts to one short-drama client in a week. The client’s daily ad spend rose from several thousand dollars to the low six figures. 沈晨岗 treats continued scaling as the performance signal: “The more it spends, the better the performance must be, or it would not keep spending.”
15. Once generation becomes commoditized, quality frameworks and scaled data become the new leverage
沈晨岗 believes the scarce capability is not whether a team can produce a large volume, but whether an asset is a 1, a 3, or a 5. Beyond production cost, creative unlocks a larger media budget, so small quality differences are amplified by distribution.
His estimate is that “a 1% gap in quality can ultimately translate into a performance gap of more than 10%.” A large model can assign hundreds or thousands of tags covering colors, scenes, and elements, but that does not mean it knows how to create good creative.
AI cannot grow an underlying cognitive framework by itself. People must first abstract the knowledge and scenarios, define the methods and processes, and then let AI iterate within that framework. Talent structures will therefore need more higher-order abstract thinking.
The calculator analogy explains the shift. People can forget how to do multidigit multiplication and division or manually calculate square roots, but a bridge engineer still needs to know when a calculation tool can be used. Competing with machines on simple capabilities that have already been tooled “basically has no value.”
16. Marketing will become the digital nervous system of the enterprise
Feishu Shenno says its AI infrastructure can be reused across more than 100,000 customers, while its systems capture more than $20M in daily advertising-spend data. A single company would struggle to bear the same cost or achieve the same learning speed and sample size.
But 沈晨岗 rejects a generic promise that ROI will “increase by at least 30%.” Only after understanding a client’s current ROI, product, relevant input variables, and the causes of its problems—and matching those against the client’s own patterns—can anyone offer a conditional judgment. Treating AI as a “magic” module that produces growth as soon as it is plugged in is a dangerous illusion.
Through Meet Experience／蜜悦科技, Feishu Shenno also helps clients build websites, data-collection systems, user-data management, and operating systems. While the advertising side may fight repeatedly for a 1%–2% improvement, fixing the client’s own digital links can produce a 10%–20% gain. The value is “multiplicative, not additive.”
沈晨岗 expects AI to keep lowering the barriers to creative work and global customer acquisition, putting small and large companies on a more equal starting line. Marketing will also connect viewing, liking, registration, purchase, repurchase, products, orders, and supply chains, becoming a real-time “digital nervous system.”
His anxiety is not about the strategic direction, but about organizational execution. Designing an AI workflow is easy; getting the entire team to understand it and change established ways of working is far harder than expected. That is also where the opportunity lies: “We all think it is hard; others may find it even harder.”
His 3 reminders for new global operators are to balance survival with long-termism, prioritize data and technology from day one rather than paying a high-cost catch-up bill later, and make good use of specialist partners. Global competition is no longer a game that a small team can handle alone; collaboration can also create a multiplier effect.