{"schemaVersion":"1.0","type":"Article","title":"How the FMCG Industry Can Embrace Change in the AI Era","description":"The mindsets, capabilities, and organizational models the FMCG industry must rebuild in the AI era.","author":{"name":"Zhao Bo","alternateName":"赵波","profile":"https://xinjignxiaozhaobo.com/en/about/"},"publisher":"Zhao Bo (赵波)","language":"en","publishedAt":"2026-03-16T00:00:00.000Z","updatedAt":"2026-03-16T00:00:00.000Z","topic":{"name":"Organizational Change","url":"https://xinjignxiaozhaobo.com/en/topics/organization/"},"tags":["AI","FMCG","Organizational Change"],"translationKey":"how-fmcg-can-embrace-change-in-the-ai-era","canonical":"https://xinjignxiaozhaobo.com/en/how-fmcg-can-embrace-change-in-the-ai-era/","markdown":"https://xinjignxiaozhaobo.com/en/how-fmcg-can-embrace-change-in-the-ai-era.md","json":"https://xinjignxiaozhaobo.com/api/articles/en/how-fmcg-can-embrace-change-in-the-ai-era.json","translation":{"language":"zh-CN","canonical":"https://xinjignxiaozhaobo.com/zh/how-fmcg-can-embrace-change-in-the-ai-era/","markdown":"https://xinjignxiaozhaobo.com/zh/how-fmcg-can-embrace-change-in-the-ai-era.md"},"citation":"Zhao Bo. “How the FMCG Industry Can Embrace Change in the AI Era.” 2026-03-16. https://xinjignxiaozhaobo.com/en/how-fmcg-can-embrace-change-in-the-ai-era/","copyright":"Copyright © 2026 Zhao Bo (赵波)","usagePolicy":"https://xinjignxiaozhaobo.com/ai-policy.txt","contentFormat":"text/markdown","content":"**Author** | Zhao Bo **Reviewed by** | Wang Hai **Layout by** | Liu Zhen\n\nHello, everyone. I’m Zhao Bo from New Distribution. My sincere thanks to all the brand owners, distributors, retailers, and service providers from across China who traveled all the way to Chengdu despite your busy schedules to attend the 11th CFC China FMCG Conference and the 6th China FMCG Distribution and Retail Conference.\n\nToday, I’d like to discuss a topic with you—“How Should the FMCG Industry Embrace Change in the AI Era?”\n\n**FMCG is likely to be one of the first industries in which AI is adopted at scale and delivers tangible results.**\n\nThis judgment is not something I dreamed up while sitting in an office. During the Spring Festival holiday, I shut myself away in a hotel and spent nearly every day experimenting with AI—studying AI agents and exploring how to make AI do real work rather than merely chat with us. The further I went, the more strongly I came to realize one thing: **AI is no longer merely a “tool that can talk”; it is evolving into a digital employee, digital assistant, and digital system.**\n\nFor the FMCG industry, this is not a minor opportunity. It represents a structural transformation.\n\nMany people still see AI as little more than a way to write copy, create PowerPoint presentations, or generate a few images. But I want to emphasize that these surface-level capabilities are not what truly matters. What matters is that **AI is rewriting consumer decision-making, reshaping how brands connect with channels, and transforming the internal operating logic of enterprises.**\n\n\n\n**Why I Say FMCG Is on the Front Line of**\n\n**AI-Driven Transformation**\n\nLet me start with the conclusion: **FMCG is inherently well suited to AI adoption.**\n\nWhy?\n\nBecause FMCG has several defining characteristics.\n\nFirst, high frequency.\n\nConsumers make numerous FMCG purchasing decisions every week. Beverages, snacks, alcoholic drinks, condiments, and household and personal-care products are purchased every day.\n\nSecond, low average transaction value.\n\nMost purchases involve relatively small amounts, so the cost of making a wrong choice is also low. Consumers do not spend half a month researching these products as they might when buying a house or a car. In many cases, they decide within seconds.\n\nThird, low involvement.\n\nA large proportion of FMCG consumption is not based on careful deliberation but on rapid judgment. Whichever product is more convenient, easier to reach, or feels more like the “right answer” in a given situation is more likely to be purchased.\n\nFourth, strong dependence on consumption occasions.\n\nFMCG purchases are closely related to time, place, weather, mood, physical condition, holidays, and social relationships. Whether you stayed up late, exercised, drank alcohol, hosted guests, took care of your children, or simply happened to pass a convenience store can all affect what you choose to buy.\n\nTaken together, these four characteristics create precisely the kind of environment in which AI excels.\n\nAI’s greatest strength is not creating demand out of thin air, but **matching, recommending, evaluating, and ranking options amid vast amounts of information**. FMCG, meanwhile, is an industry characterized by extremely frequent decisions, exceptionally high data density, and very rapid feedback. A consumer may receive a recommendation, make a purchase, repurchase, or churn on the same day, immediately creating a closed feedback loop.\n\nTherefore, once AI enters the FMCG industry, it will not penetrate gradually. It will quickly overturn many things we have long taken for granted.\n\n\n\n**What AI Is Changing**\n\n**Is Not Merely Efficiency, but Consumer Decision-Making Power**\n\nHistorically, the core logic of the FMCG industry has been “people looking for products.”\n\nBrands advertise, channels distribute products, retail outlets create displays, sales representatives drive sell-in and sell-out, and consumers compare options, make selections, and place orders themselves in front of the shelf.\n\nBut a shift is taking place in the AI era: **decision-making power is moving from “people choosing products” to “AI choosing products on their behalf.”**\n\nThis transition has already begun, although many people have yet to truly recognize it.\n\nToday, the consumer decision journey no longer begins only at the shelf, in the store, or in an e-commerce search box. Voice assistants, chatbots, recommendation engines, automated replenishment systems, and platform algorithms are making an increasing number of preliminary judgments on behalf of consumers.\n\nImagine this scenario.\n\nYou drank too much last night and feel unwell this morning. You ask AI: What would be suitable for me to drink right now?\n\nUnder the traditional logic, you would work it out for yourself.\n\nUnder the AI-driven logic, AI would provide a much more specific recommendation based on your physical condition, historical preferences, and situational needs.\n\nI gave an example at the conference. For someone like me, who pays close attention to uric acid levels, gout, and liver health, AI would certainly make different recommendations than it would for a young woman. For me, soda water, turmeric-based beverages, and fresh lemon drinks are about much more than taste. My physical condition, functional needs, and purchasing decisions are interconnected.\n\nThis is the most fundamental change in consumer decision-making in the AI era:\n\n**In the past, brands addressed groups of people. In the future, they will address specific individuals, one by one.**\n\nAs a result, several highly visible changes will emerge in the FMCG industry.\n\n**1\\. The Experience Will Be Rewritten**\n\nIn the past, consumer insights were primarily based on audience profiles.\n\nMen, women, people aged 18 to 25, white-collar workers, students, mothers, late-night snack consumers, and fitness enthusiasts.\n\nThat will no longer be enough. The future requires **individual profiles**.\n\nEven when people are drinking water, one person may want to quench their thirst, another may be socializing, another may be trying to lose fat, another may want to ease a hangover, another may be concerned about uric acid, and yet another may simply want something photogenic.\n\nAI does not understand you as part of a “broad audience segment.” It understands you based on your specific state at that particular moment.\n\n**2\\. Pricing Will Be Rewritten**\n\nIn the past, price was a label—9.9, 12.9, or 19.9—displayed on the shelf, with everyone seeing the same thing. In the future, price will increasingly become a variable.\n\nPricing and promotional cadence will become increasingly intelligent across different channels, consumer groups, times, and triggering conditions. What consumers see will no longer be merely a static price tag, but the outcome of dynamic optimization and negotiation.\n\n**3\\. Loyalty Will Be Rewritten**\n\nIn the past, brand loyalty often meant loyalty to a trademark, advertising, or habit.\n\nIn the future, consumers are more likely to be loyal to a system.\n\nWhichever system understands me better, saves me more time, and consistently helps me make choices that are “almost always right” is the system I will be more willing to remain within.\n\nI later distilled this judgment into a single sentence:\n\n**In the AI era, consumers’ fundamental expectations of FMCG brands will no longer be limited to affordability and usability. They will expect three things—you understand me, you help me choose, and you do not overstep my boundaries.**\n\nWhoever strikes the best balance among these three points will be more likely to win consumer mindshare in the next stage.\n\n\n\n**The Real Question Is Not “Whether to Adopt AI”**\n\n**but What Problem You Are Actually Trying to Solve**\n\nOver the past few years, many enterprises have been discussing digital transformation and AI.\n\nBut I have consistently felt very strongly about one thing: **many enterprises do not lack tools; they lack problem awareness.**\n\nOver the past decade, many enterprises have spent tens of millions—or even hundreds of millions—on digital transformation, yet achieved disappointing results. Why? It was not because they lacked systems, reports, or tracking points. It was because they failed to think clearly from the very beginning:\n\n**What problem am I actually trying to solve?**\n\nIf the problem is not clearly defined, more data will not help. If the model has not been properly conceived, the system will only grow more complex. If the right use case has not been identified, adding more capabilities will only create greater confusion.\n\nThe same applies to AI.\n\nMore capabilities are not necessarily better. A more expensive model is not necessarily better. A larger system is not necessarily better.\n\n**The most important thing is to start with the problem.**\n\nFor example, if you are a distributor, is your most critical problem slow inventory turnover, weak sell-through at retail outlets, low salesforce productivity, high accounts-receivable risk, or noncompliant in-store displays? You must first identify where your pain point lies.\n\n  * Identify the problem first, then define the model.\n\n  * Set the objective first, then match the data to it.\n\n  * Close the loop for one use case first, then replicate it across additional use cases.\n\nThis sequence cannot be reversed. If it is, you will almost certainly end up with a pile of beautifully presented nonsense.\n\n\n\n**Why I Keep Saying That**\n\n**Firsthand Business Intuition Is the Most Valuable Asset**\n\nThrough my recent experimentation with AI, I have developed one particularly strong impression: **AI is powerful, but it has no firsthand, embodied sense of reality.**\n\nIt possesses enormous knowledge and even has access to most of the world’s publicly available information, but it does not know what warmth or cold feels like. It does not know pain. It cannot sense a shift in the atmosphere at a social gathering. Nor does it know whether a business owner who says, “Let’s revisit this later,” is hesitating, being dismissive, or simply short of money.\n\nPeople possess these senses; AI does not.\n\nThat is why I keep saying that what is truly valuable is not merely data, but the kind of firsthand business intuition you develop on the front line.\n\nYou have seen customers frown.\n\nYou know why frontline sales representatives cannot move a product. You know that when a small-store owner says they will not place an order, what they really fear is being unable to sell the stock. You know that a brand may appear to have respectable sales while failing to establish any genuine consumer mindshare at the retail level.\n\nThis tacit knowledge is often difficult to articulate, but you can sense it.\n\nAnd it is precisely this knowledge that determines whether AI can generate real value.\n\nFor experienced industry professionals, therefore, the greatest opportunity is not to compete with AI in computing power. It is to rapidly convert your awareness of real-world scenarios, problem orientation, and business understanding into capabilities that AI can invoke. When you identify a genuine problem and immediately use AI to solve it, that process itself constitutes product development.\n\nDo not begin by trying to build a massive system.\n\nBuild a power drill first. Build one small capability first. Solve one high-frequency problem first. That is enough.\n\n**How Enterprises Should Implement AI**\n\n**I Believe It Should Be Done in Three Stages**\n\nFMCG enterprises can implement AI in three stages: **the crutch stage, the organ stage, and the system stage.**\n\n**1\\. The Crutch Stage**\n\nStart by treating AI as a tool.\n\nUse it to write copy, create PowerPoint presentations, consolidate reports, categorize public sentiment, diagnose retail outlets, provide basic assortment recommendations, and issue inventory alerts.\n\nAt this stage, do not mythologize AI, and do not attempt to restructure the company from the outset.\n\nThe objective is simple: make people willing to use it, make its recommendations reviewable, and make its benefits measurable.\n\n**2\\. The Organ Stage**\n\nNext, transform AI from an add-on tool into an “organ” embedded within your business processes.\n\nExamples include automated replenishment recommendations, out-of-stock response recommendations, sales-route optimization, trade-spend recommendations, alerts for anomalous retail outlets, and reconciliation and write-off assistance. At this point, AI is no longer merely producing reports. It is beginning to enter your daily operational workflows.\n\n**3\\. The System Stage**\n\nOnly after that do you reach the system stage.\n\nAI begins to span five levels—consumers, points of sale, cities, the supply chain, and headquarters—creating a genuine closed operating loop.\n\n> At the consumer level, it identifies consumption occasions and assesses needs.\n>\n> At the point-of-sale level, it creates store profiles, optimizes SKUs, and recommends replenishment.\n>\n> At the city level, it coordinates channels, forecasts sell-through, and improves trade-spend efficiency.\n>\n> At the supply-chain level, it forecasts demand, reallocates inventory, and coordinates production scheduling.\n>\n> At the headquarters level, it provides omnichannel insights, allocates resources, and manages risk.\n\nAt this stage, AI is no longer merely saving time. It begins to influence how the entire company operates.\n\n**AI Implementation Requires Principles; Not Everything Can Be Automated**\n\nI do not advocate connecting AI directly to core transaction processes and allowing it to execute actions automatically from day one. The risks are too great.\n\nThe approach I favor is to evaluate first and then implement progressively.\n\nTo determine whether AI should handle a particular task, I believe at least three factors must be considered:\n\nFirst, **how automatable is it?**\n\nCan the rules be clearly articulated? Can the boundaries be clearly defined?\n\nSecond, **how fault-tolerant is it?**\n\nHow costly would a single mistake be?\n\nThird, **how high is the rate of change?**\n\nIf the rules change every day—one way today and another tomorrow—the task is not suitable for rigid automation.\n\nIn addition, I believe every enterprise using AI must clearly define four things:\n\n**The task contract, acceptance criteria, constraints, and feedback loop.**\n\nWhat do you want it to do, and what do you not want it to do?\n\nWhat constitutes a correct result? What must it not touch? How should deviations be corrected?\n\nIf these things are unclear, AI will inevitably “hallucinate” and go off track. It is not deliberately trying to undermine you; you simply failed to define the boundaries clearly.\n\n**For Distributors and Sales Representatives, AI Means**\n\n**Not Minor Adjustments, but a Rewriting of Their Roles**\n\nI do not think many people have yet taken this point seriously enough.\n\n**Let’s Start with Distributors**\n\nIn the past, a large portion of distributors’ profits came from price spreads, information asymmetry, and localized relationship advantages.\n\nBut once AI enters the industry at scale, channel transparency will continue to increase. Where products are sold, how much has been sold, who has faster sell-through, who is carrying excessive inventory, and who is executing well will all become increasingly visible.\n\nThis means that a distributor’s true value in the future will no longer primarily be, “I know more than you do.” Instead, it will be:\n\n> I have stronger fulfillment capabilities.\n>\n> I provide better retail-outlet services.\n>\n> I can manage regional market operations.\n>\n> I can enable data collaboration.\n>\n> I can help brands execute more granular localization strategies.\n\nDistributors will therefore most likely evolve into several distinct roles:\n\nRegional fulfillment service providers, retail-outlet operations service providers, and data collaboration service providers.\n\nThose that can upgrade in these directions will have greater room for growth.\n\n**Now Let’s Look at Sales Representatives**\n\nThe impact on sales representatives will be even more direct.\n\nIn the past, many sales representatives relied on pounding the pavement, drinking with clients, personal relationships, and experience.\n\nThese things will not disappear entirely, but they will no longer be enough.\n\nFuture sales representatives will function more like people who execute and fine-tune strategies within an algorithmic framework.\n\n> Routes will be recommended by the system.\n>\n> AI will provide order recommendations.\n>\n> In-store displays will be recognized in real time through smartphone photos.\n>\n> AI will create profiles of retail outlets.\n>\n> AI will recommend customer-management strategies.\n\nA good sales representative in the future must therefore be able to do three things:\n\n  * **Interpret data**\n\nUnderstand the store profiles and recommendations provided by the system.\n\n  * **Use tools**\n\nKnow how to use SFA, BI, image recognition, and data assistants.\n\n  * **Translate**\n\nTranslate data and recommendations into language that business owners, store operators, and customers can all understand.\n\nPut simply, the core competitiveness of future sales representatives will no longer be physical effort alone, but the ability to collaborate effectively as “human + tools.”\n\nIn Closing:\n\n\n**AI Will Not Replace FMCG Professionals**\n\n**It Will Redefine Value Creation**\n\nI have never believed that AI will suddenly eliminate everyone working in the industry.\n\nBut I am absolutely certain that **AI will rapidly compress repetitive work that creates no value, low-quality communication, and low-level decision-making.**\n\nIt will force us to answer several questions again:\n\n> What exactly do we understand about consumers better than others do?\n>\n> What kinds of products are we better at creating than others are?\n>\n> In which scenarios are we better able to provide superior solutions?\n>\n> Do we make money through relationships or through capabilities?\n\nTherefore, whether you are a brand owner, distributor, or retailer, I believe the most critical priorities ahead are very clear:\n\n> First, build a solid foundation of product capabilities and data infrastructure.\n>\n> Second, enter AI-driven decision touchpoints as soon as possible instead of clinging solely to traditional shelf-based thinking.\n>\n> Third, rewrite marketing and pricing systems, moving from indiscriminate mass coverage toward individualized matching.\n>\n> Fourth, drive organizational upgrades so that AI genuinely enters the daily operating cycle.\n\nThis transformation has no finish line.\n\nIt is neither a one-off transaction nor something that ends after implementing a single project.\n\nIt is more like an infinite game. The winners will not be those who shout the loudest today, but those who start earlier, iterate faster, and continuously turn real problems into real results.\n\nSo I will close with the same point:\n\n**The key question is not “whether to adopt AI,” but “how to take the first step as quickly as possible.”**\n\nAs we embrace AI, we must not lose sight of the most fundamental and straightforward principles.\n\nProduct strength is the foundation, supply-chain capability is hard currency, and long-term trust is the most valuable asset.\n\nAI will not create these things for you.\n\nBut those who learn first how to use AI to amplify them will have a greater chance of winning the next stage."}