When AI Starts Deciding for Consumers, FMCG Becomes the Front Line

Author | Zhao Bo. Review | Wang Hai. Layout | Liu Zhen

In the FMCG industry, what AI is truly changing isn’t one or two “tools” — it’s the consumer themselves: who makes the decision, how they are persuaded, what price they see, what journey they take, and ultimately who they stay loyal to.

In the long piece below, I’ll stand entirely in the shoes of an FMCG practitioner and, starting from the consumer, systematically break down five dimensions that AI is rewriting.

Why FMCG

is becoming the front line where AI rewrites consumer behavior

If AI is reshaping every industry, then in the consumption space, FMCG is almost the first to feel the change. The reason isn’t the technology itself, but the structural characteristics of FMCG consumption.

I. High-frequency, low-deliberation, heavily context-driven: the category most suited to being “taken over” by AI

FMCG products share several typical characteristics:

Extremely high decision frequency: purchases happen daily or weekly, far more often than durable goods like appliances, 3C electronics, or cars.
Very low per-purchase ticket price: the sense of loss from a wrong purchase is minimal, so consumers don’t invest much mental effort.
Extremely shallow decision depth: choices are mostly habitual, picked up on impulse, rarely compared at length.
Highly fragmented scenarios: supermarkets, convenience stores, e-commerce platforms, forward warehouses, community stores, vending machines, brand mini-programs, livestream rooms… the channels and touchpoints are too complex to track with the naked eye.

What does this mean?

For the consumer, FMCG decisions were always a “good enough will do” affair — they are naturally willing to have the decision simplified and automated.

For the algorithm, FMCG is a training ground with massive samples, continuous feedback, and rich data — every day brings new behavior, new orders, new A/B test results.

Under this structure, AI naturally starts helping consumers accomplish two things: first, remembering “what you usually buy”; second, guessing “what you might want this time.”

And so a key question emerges: on high-frequency, low-involvement FMCG purchases, is the consumer really choosing actively — or simply going along with the “default option” the system hands them?

II. When the industry discusses AI, it’s mostly about “efficiency”; the real transformation is hiding on the “human” side

Today’s industry discussions about AI mostly revolve around:

More accurate demand forecasting, reducing stockouts and overstock; smarter replenishment, improving supply chain turnover; more precise targeting, better-looking ROI; automated content generation, lower creative costs.

These all matter, but they remain stuck at the enterprise’s point of view. What truly determines the industry’s future shape are the following fundamental shifts happening to the consumer.

Who is making consumption decisions on their behalf? How are they being influenced and persuaded? What has price become in their eyes? Is the consumption journey they take one they explored themselves, or one that was pre-choreographed for them? Is the object of their loyalty a brand, or an entire algorithm-driven experience system?

Once the answers to these questions are quietly rewritten by AI, many of the marketing, channel, and branding methodologies we’re familiar with will lose the premises they once rested on.

The Shift in Decision-Making Power:

From “People Choose the Goods” to “AI Chooses the Goods for People”

Of all the changes, the most profound — and the most easily overlooked — is that “who makes the decision” has undergone a structural migration.

A new role takes the stage: the AI agent enters the consumer’s shopping scene

The old logic was: a person has a need → they search and browse on their own → they compare on a shelf or a page → they decide what to buy.

Now, the chain increasingly looks like this:

A person expresses a vague need in natural language: “Buy me some more of what I usually drink,” or “Help me restock the tissues and laundry detergent we normally use at home.”

The AI system does the “dirty work”: checking order history, looking at current prices, comparing promotion strength, calculating delivery time.

The system offers one or two “suggested options,” and the consumer only needs to tap “confirm.”

On the surface, this is just a UX optimization; in essence, it’s a redistribution of decision-making power:

The consumer goes from “choosing on the shelf myself” to “choosing from the options the system provides.”
The starting point of the decision shifts from “I suddenly thought of something I need to buy” to “the system reminded me it’s time to buy.”

For FMCG, because a large share of purchases are already repeat buys and routine restocking, this “decision delegation” happens with almost no resistance.

E-commerce platforms’ “buy again with one tap” and “frequently bought list”; platform or brand apps’ “restock reminders”; smart speakers and voice assistants that let you say “order the milk and cat food I usually get”; the “frequently bought lists” in community group-buying and forward-warehouse apps.

In these scenarios, the consumer’s active deliberation gradually recedes to the background, while AI’s “presets” and “recommendations” become the true starting point of the decision.

Decision engagement declines, while decision efficiency and “fit” rise rapidly

For the consumer, this change has clear benefits:

Search cost drops dramatically: no more flipping through category pages one by one — just confirming the system’s proposed option.
Risk of choosing wrong is reduced: the system references historical preferences and review reputation to filter out the most unsuitable options.
Time becomes more valuable: limited attention can be spent on more important, more complex consumption decisions.

But at the same time, the depth of the consumer’s engagement in decision-making keeps declining.

Many purchases turn into reflexively clicking the same option, with almost no comparison anymore. Whether a new brand or new SKU gets a chance to enter the consumer’s field of view now depends on whether the system is willing to give it an exploration slot. The shelf is no longer a linear display — it’s a small algorithmically filtered slice of the world, presented as a recommendation page.

Practitioners need to realize: you are less and less directly persuading “that person” — you are persuading “the system that hands them the option.” If you’re not on the recommendation list, for a portion of consumers, you’ve already “exited the competition.”

Decisions get “better,” but feel “less like your own choice”: a new paradox

AI genuinely makes many decisions objectively better:

Comprehensively weighing price, delivery, reputation, ingredients, and other dimensions; continuously learning from user feedback and gradually adjusting recommendations; avoiding “impulsively ordering something you’ll never use a second time.”

But the subjective experience can become:

“It feels like everything just happened naturally, but I’m not quite sure how I was persuaded”; “I know the recommendation fits me well, but I also vaguely feel like I don’t have as much agency anymore.”

This brings a long-term challenge: FMCG brands must use AI to improve decision quality while also, at key moments, giving consumers enough of a sense of “making up their own mind.” This is both an experience-design problem and a psychology problem.

The Rewriting of Experience:

Redistributing Personalization, Emotion, and Attention

If the shift in decision-making power is a structural-level change, then on a day-to-day, felt level, what consumers experience most intensely is that the experience itself has changed.

From crowd personas to individual personas: everyone is walking their own “exclusive script”

In the traditional marketing era, we were used to sorting people into categories: stay-at-home moms, office workers, students, middle-class men…

These are coarse-grained segments based on demographics, occupation, and family life stage.

In the AI era, high-frequency consumption categories like FMCG are giving rise to a different way of segmenting.

For example, the same “young female white-collar worker” gets split, inside the system, into countless micro-personas. Office workers who love sweet drinks but are wary of sugar; single renters who frequently order late-night delivery; fitness enthusiasts who often buy sports nutrition products.

It can go even further down to the specific individual — someone who likes small packaging, tries new flavors, and is unusually price-sensitive at the same time.

The result: everyone sees a different category ranking on the e-commerce homepage; a brand’s content and messaging to them differs too; the same brand looks completely different on different people’s phones.

The consumer’s subjective experience is: “How does it always show me what I want to see?”; “It seems like every push notification and recommendation has become more targeted”; “The longer I browse, the better the system understands me.”

Attention is being redistributed here. It’s shifting from one-off impressions bought with mass traffic to long-term, multi-touchpoint, multi-dimensional personalized companionship. The relationship between brand and consumer changes from a series of brief encounters into a storyline that keeps being written.

Emotion-driven content and advertising: from “grabbing attention” to “tuning emotion”

After AI got involved in content production, FMCG advertising is undergoing three layers of change:

First layer: a leap in content production efficiency

Massive numbers of poster, short-video, and copy variants can be generated quickly; completely different stories and visual languages can greet different demographics and interest circles.

Second layer: precisely triggering “emotional buttons”

The system doesn’t just know what you “like to drink” — it knows what “moves you.” Some people resonate more easily with childhood-nostalgia content; some are more sensitive to health-anxiety messaging; some care intensely about social identity and trend labels.

AI tests different emotional narratives, finds the one that best drives your clicks and conversion, and keeps reinforcing it on you.

Third layer: the boundary between content and advertising disappears

Many “trend-seeding videos” and “lifestyle content” pieces are, behind the scenes, brand stories and placement strategies AI helped generate.

What the consumer sees is content that suits their taste perfectly, without necessarily realizing they’re being guided down a highly engineered persuasion path.

For FMCG products, this emotional impact matters especially. Low unit price and low cost of trial-and-error give emotion-driven factors more weight in the decision; a bottle of drink, a bag of snacks, a tube of lipstick gets loaded with emotional value like “treating yourself, fighting anxiety, expressing an attitude.”

In other words, advertising no longer just occupies your eyeballs — it occupies your emotional channel; behind the scenes, AI keeps optimizing which storyline, which color scheme, which music, and at what moment of your day, pulls you into that consumption path.

The algorithmization of shelves and packaging: design tamed by data

In offline stores, the most intuitive experience for consumers is that the shelf seems to “understand me better and better.” New products get refreshed at a fast pace, packaging style always lands right on current aesthetic trends, and information layout happens to follow reading habits.

Behind this is the fact that packaging design is no longer a spur-of-the-moment creative call — it has been folded into continuous testing and iteration.

Copy A vs. copy B; warm-leaning vs. cool-leaning main color; functional information emphasized vs. emotional information emphasized; click-through rate, conversion rate, and browsing dwell time on the e-commerce side, sales volume and shelf-position performance offline — all of it becomes input for the next round of design optimization.

So what the consumer faces is no longer the crystallization of a single designer’s inspiration, but a version filtered by algorithms and data as “most likely to make people buy”; the shelf as a whole is itself the output of a continuously trained model.

For brands, creativity is no longer an unquestionable “art” — it has to accept being quantified and tested. At the same time, collaboration between design teams and data teams becomes a key capability shaping consumer perception.

The Intelligent Turn in Pricing and Promotions:

Consumers Step Into the “Algorithmic Pricing Era”

If the changes discussed so far are relatively “soft,” then on the pricing and promotion front, AI’s impact on the consumer is “hard.” The same bottle of drink, for the same person, may show a different price at a different time and in a different scenario.

I. Dynamic pricing: price shifts from a “label” to a “variable”

In the traditional retail era, prices were relatively stable. Changing a store’s price tags had a cost, so they couldn’t be changed all day long; brands and channels set promotion calendars together, with a relatively fixed rhythm; consumers could rely on “remembered historical prices” and “comparison shopping” to make judgments.

Once AI is layered on, prices start to display the following characteristics.

High-frequency micro-adjustments: prices are automatically fine-tuned based on real-time sales, inventory pressure, proximity to expiration date, weather, holidays, and time of day.
Multi-scenario divergence: the same SKU presents a whole “price cloud” across online and offline, different cities, and different channels.
Model-driven optimum: the goal shifts from “maintaining a certain list price” to “using the most suitable price structure to achieve the overall optimum in profit and turnover.”

The consumer’s direct impression: “How did the price quietly change without me noticing?”; “The original price no longer has any reference value, since there’s some kind of promotion running at every moment anyway”; “I have to buy on the promotion’s rhythm rather than my own.”

This change is especially pronounced in FMCG: short shelf life, high sales volume, and strong demand elasticity give dynamic pricing enormous room to operate. A large number of promotion labels mask the algorithmic price adjustments underneath, making it very hard for consumers to form a stable judgment.

II. Personalized promotions: everyone lives in their “own price universe”

Before AI got involved, FMCG promotions were more like a form of “public welfare.” At the same supermarket, at the same time, every consumer faced the same promotional poster and the same discounted price.

Once personalized promotions appeared, the situation changed: the system first builds you a “price persona” — how price-sensitive are you? How loyal are you to this brand? Do you tend to buy only when there’s a discount?

Based on that persona, different consumers are then handed different combinations of coupons and discount strength. Some get 11% off, some get 21% off; some receive a coupon for 50 off orders over 99, while others are stuck long-term with only 20 off orders over 199.

For the consumer:

Upside: “there’s always some discount that fits me pretty well” — promotion hit-rate is high; they feel “taken care of,” which boosts satisfaction and repeat-purchase intent.
Hidden change: price is no longer a “public piece of information” openly displayed on the shelf, but rather a “private experience” mixed in with one’s personal behavioral history; the uncertainty of “did I overpay” rises, but because the cost of figuring it out is too high, most people just settle for “whatever, close enough.”

This turns the FMCG pricing system from a transparent market mechanism into a semi-transparent algorithmic mechanism.

III. Promotion rhythm gets algorithmized: consumers’ price expectations get systematically reshaped

Before AI, promotions followed a few fixed rhythms: major shopping festivals like Spring Festival, National Day, Double 11, 6.1; category-specific, periodically fixed “monthly/quarterly big promotions.”

After AI, promotions become “an entire year stitched together from countless micro-promotions”: new small campaigns are continuously generated based on real-time data, and whatever tests well gets amplified immediately. If a campaign performs well, it’s automatically extended or replicated; if not, it’s shut down quickly.

And different people see different rhythms — certain shopping periods get reinforced for some people, while for others there are simply more small promotions in everyday life.

Consumers are gradually trained into a new mindset: “there’ll always be another promotion coming anyway”; “if there’s no promotion I won’t buy, and if the discount isn’t strong enough I won’t buy either”; “I only buy when the system decides it’s ‘time for me to buy.’”

In other words, AI isn’t just adjusting price itself — it’s reshaping the consumer’s entire mental model of “when it’s worth buying.”

The Restructuring of Relationship Structures:

From Brand Loyalty to Experience Loyalty

Once decision-making methods, experience pathways, and pricing mechanisms have all been rewritten by AI, the relationship between consumer and brand naturally can no longer be as simple as the traditional “see the ad — form a preference — repeat purchase.”

The object of loyalty shifts from “a name” to “an entire experience system”

In many FMCG categories, consumer loyalty to a brand originally started out emotional: childhood memories, advertising stories, spokesperson image, social identity, and so on.

After AI gets involved, another dimension is layered on top: Does this brand/platform keep remembering me? Does it “think things through for me” at every key moment? Does it “step up immediately” when something goes wrong?

Consumers gradually form a new kind of “experience loyalty” — loyalty to the brand or platform where “I always buy the right thing, quickly, and comfortably.” Loyalty to “this system always knows roughly what I’ll need next time, and has it ready in advance.”

The result is that a specific SKU may become more replaceable, while the replaceability of the whole experience system actually decreases. What brands compete on is no longer just product strength and advertising power, but also the smoothness and thoughtfulness of the entire AI-driven experience chain.

Psychological expectations are upgraded: personalization and intelligent experience become “table stakes”

There was a time when personalized recommendations, smart customer service, and consistent cross-channel experience were seen by consumers as pleasant surprises, bonus points.

But as more and more leading brands and platforms do this better and better, consumer expectations have shifted.

They now default to assuming you should remember their history; default to assuming you should deliver the right content and offers at the right moment; default to assuming you should have an always-responsive online service channel.

Once this expectation becomes the baseline, any brand that reverts to “one template for everyone” or “no personalized communication” will look behind on experience.

A brand image with no memory capability will be seen as “doesn’t understand me”; a channel without smart service capability will be seen as “inefficient” and “inconvenient”; touchpoints that are heavily homogenized and don’t offer personalized interaction will be seen as “noise.”

The direct implication for FMCG practitioners is that “intelligence and personalization” is no longer an edge innovation you can slowly experiment with — it’s gradually being written into the consumer’s mind as a basic infrastructure requirement.

Between “being understood” and “being calculated against”: trust becomes the new key variable

As AI deeply embeds itself in consumers’ lives, a subtle but crucial psychological tension emerges.

On one hand, consumers enjoy the feeling of being understood, remembered, and taken care of; on the other, they also vaguely worry that “am I being seen through too clearly” — even being exploited or manipulated.

This tension exists in FMCG too — you’ll always see a drink discount when you’re thirsty, coffee and functional drinks when you’re up late, electrolyte drinks after a workout.

From a shopping-experience standpoint this is genuinely nice, but if even your emotional swings, sleep schedule, and social relationships can be precisely leveraged by advertising, many people will feel an uncomfortable sense of “being surveilled.”

This means that in the AI era, when handling the consumer relationship, FMCG brands must answer two layers of questions at once.

The functional layer: are you genuinely using AI to help them make better decisions and get products that suit them better?
The values layer: to what extent do you respect their pace, boundaries, and right to know, rather than just thinking about selling a bit more?

Truly sophisticated brands keep reinforcing a certain feeling in their narrative.

“We use AI to make things easier and more reassuring for you, not to squeeze a bit more out of you”; “you can choose more personalization at any time, or you can choose to keep it simple”; “you can turn off certain recommendations at any time, or tell us not to push this type of content anymore.”

Once the consumer believes this, AI stops being just a cold efficiency tool and becomes part of the brand’s promise — a promise to use technology responsibly.

In Closing

From Studying the Consumer to “Studying the Consumer Plus the Algorithm”

Before AI, when FMCG people studied the consumer, they mostly cared about who this person is, where they are, what they want, and why they buy from me instead of someone else.

After AI, the questions to answer have grown by half again:

How does the layer of algorithm behind them understand them? Who decides what they see and what they don’t see? Who sets their price universe and promotion rhythm? Who is quietly rewriting their decision path and experience expectations?

You’ll find that part of the real competition has already shifted from “between brands” to “between brands and algorithms.”

You need to win the consumer in the classic sense, and you also need to win, in this new sense, the entire AI system that carries their experience — the platform’s recommendations, the channel’s ranking, the internal logic of proprietary systems.

And for the FMCG consumer, the overarching trend brought by AI can be condensed into one sentence: decisions become easier, experiences become more personal, prices become smarter — but without quite realizing it, they’ve also handed over part of their power to “make up their own mind” to an invisible algorithm.

For today’s FMCG practitioners, the question truly worth continuing to ask may be this: in a world where “the algorithm keeps getting smarter,” are you also making your own brand more human?