In recent years, the FMCG industry has invested in digital transformation, completing the initial stages of 'data online' and 'system integration.' Now, this process is accelerating into a new 'AI-driven' phase—companies are no longer satisfied with 'seeing' data but seek to make data flow in real time and actively empower business. AI is shifting from a tool attribute to a capability attribute, penetrating key links such as supply chain, channel, and marketing, driving enterprises from static analysis to dynamic intelligent decision-making. How to make AI move from 'foreseeing the future' to 'achieving growth' and truly become a business engine has become a key industry issue. On August 20, 2025, at the sub-forum of the 7th China FMCG Conference hosted by New Distribution, Mr. Qin Yafeng, founder and CEO of Beijing Super Node, delivered a sharing titled 'FMCG AI Deeply Rooted in Business Scenarios—Becoming an Efficiency Partner for Business Growth.' The following is his on-site speech (partially abridged), compiled by New Distribution for readers. From Anxiety to Value: AI Is the Only Antidote Today's FMCG industry is in a complex stage where 'informatization, digitalization, and AI-ization' coexist across multiple generations. Different enterprises vary greatly in their technological progress: some have just completed initial informatization, some are halfway through digitalization, and yet the AI wave has already swept in. This phenomenon of 'multi-track parallel' has left many enterprises anxious: on one hand, they worry about falling behind; on the other, they are unclear about how to apply it and what value AI can truly bring. In fact, anxiety itself is not scary; it precisely indicates that the industry is thinking forward. The real way out is not to follow trends blindly, but to let AI land where it can create business value. The FMCG industry is moving from the 'era of speed' to the 'era of efficiency.' The logic of enterprise growth is shifting from 'seeking increments' to 'managing existing stock,' and more critically, from 'management' to 'activation.' If we rely only on rigid control, digitalization can easily become empty rotation. Only by activating business scenarios one by one through AI can we truly release efficiency dividends. This 'activation' is not an abstract slogan but concrete changes in each link: Can channel management be more refined? Can business actions be more precise? Can investment truly reach the target audience? Can ROI be dynamically optimized? These questions are precisely the entry points for AI implementation. Four Entry Points for AI Deeply Rooted in Business Scenarios The FMCG industry chain is extremely complex, involving both collaboration between manufacturers and distributors and interaction between terminals and users. The value of AI is not to provide a universal overall solution, but to gradually connect into a system through implementable small-scenario entry points. The following practical directions have already shown clear business efficiency improvement logic. First: Intelligent Purification of Store Data In channel management, the problem of 'dirty, messy, and poor' store data has long existed. Manual entry, distributor reporting, and system crawling often bring a large amount of false and duplicate data, with a true rate of less than 75%. The intervention of AI can, through an automated process of 'verification—deduplication—labeling,' cross-check key elements such as store signs, locations, and names, and match them in real time with map POIs to generate unique IDs. This mechanism not only cleans historical data but, more importantly, embeds 'verification' into the entry point of new data, ensuring that future channel digitalization is built on a solid foundation. Compared with traditional manual audit models, this approach can reduce channel distortion to the lowest level and directly improve the reliability of decision-making. Second: Intelligent Distribution of Business Tasks Traditional SFA systems emphasize 'management,' requiring salespeople to visit stores on fixed routes, but in today's fragmented retail environment, this model has become ineffective. AI can intelligently recommend differentiated tasks based on the combination of store tags and personnel roles, making business actions more focused. At the same time, through a real-time quick review mechanism, it can verify immediately after task completion, forming a closed loop of 'distribution—execution—feedback.' This not only solves efficiency problems but, more importantly, re-activates the execution capability at the channel end. In the past, distributor teams often complained about being 'burdened by tasks,' but now they can find the most matching actions through intelligent allocation, and incentive mechanisms can also be implemented faster at the frontline. Third: Precise Reach of Marketing Investment In the FMCG industry, code-based marketing was once a standard tool, but it lacked precision, leading to resource waste. AI technology gives stores 'eight major scenario tags' (campus, community, business, cultural tourism, etc.), enabling targeted delivery combined with codes and user behavior. A certain instant noodle brand used this to achieve a 3% increase in repurchase rate within just seven days, which means AI is not only 'able to deliver' but also 'delivers accurately,' re-matching 'goods, money, and place' and significantly improving ROI. In an increasingly competitive environment, this precision is no longer a bonus but a critical watershed determining whether enterprises can escape involution. Fourth: ROI Analysis and Intelligent Optimization ROI has long been a management difficulty for FMCG enterprises. Incomplete POS data and difficulty in real-time tracking of shipment data have led to long-standing blind spots in input-output evaluation. Through AI-built analysis models, input data such as in-store, home-delivery, and personnel can be correlated with outcome indicators such as sales and shipments, assigning weights to different store types to form a dynamic ROI evaluation system. More critically, this system is not 'after-the-fact accounting' but can provide optimization suggestions, such as adjusting O2O investment ratios and optimizing cost structures, thereby helping enterprises truly achieve refined cost management. From Small Scenarios to Full-Domain Intelligence If AI applications in the FMCG industry remain at the level of 'slogans,' they will soon be eliminated. AI that truly brings value must be built on a mature data foundation. A complete marketing automation platform and full-chain data integration are prerequisites for AI to play its role. Only when data from 'in-store, home-delivery, and code' are connected can AI truly become an accelerator for marketing digitalization. More importantly, AI cannot be achieved overnight. For most FMCG enterprises, the most feasible path is to start with small scenarios: first solve specific problems such as store data, task distribution, or ROI optimization, and then gradually expand to the full chain. Through these 'verifiable small achievements,' organizational trust is accumulated, practical results are solidified, and ultimately, enterprises are promoted to complete the upgrade of full-domain intelligence. In an era of slowing consumption and intense competition, FMCG enterprises urgently need new growth levers. AI is not an external embellishment but an internal structural reshaping. What it brings is not just tool upgrades but the reconstruction of organizational capabilities. The future winners in the industry will not be the enterprises that shout AI slogans first, but those that can truly embed AI into business scenarios and continuously iterate in practice.