Can AI Outperform Humans in FMCG Marketing? Insights from 2.4 B Users and 210% Gains

Over the past four years, FMCG sales growth in China fell from 3.1% to 0.1%, while average price cuts reached a record 3.4% in 2024, squeezing profit margins despite aggressive discounting, promotion, and distribution efforts.

Our research team spent more than six months surveying dozens of domestic and international brands—including Coca‑Cola, Procter & Gamble, Dongpeng Beverage, and Juewei Food—to produce a deep‑dive report titled “FMCG AI Marketing Implementation Guide: Seven Scenarios × Transformation Methodology × Benchmark Cases.” The study revealed a common misconception: when companies mention AI, they immediately think of copywriting, poster design, or video generation, which is only a superficial entry point.

A severely underestimated entry

Data shows that by 2025 the Chinese “one‑code‑one‑product” market will reach ¥86.7 billion with a CAGR over 22%, yet more than 70% of firms use the code merely for anti‑counterfeiting, wasting a powerful consumer‑connection channel. Fully leveraging the code creates a “five‑code‑one” ecosystem that links 2.4 billion consumers and 4.2 million retail outlets, enabling end‑to‑end traceability of who bought what, where, and what they purchased before.

Once this data is consolidated, AI can be applied in three key ways (Chapter 3 of the report):

Breaking data silos across e‑commerce, offline, and private domains to build a 360° user view.

Transforming user segmentation from an annual update to a real‑time, per‑second adjustment.

Decision‑logic as the next battlefield

Many assume AI’s value lies in cost reduction, but the case studies show that the most successful firms did not prioritize “cost‑cutting.” Instead, they replaced experience‑based decisions with data‑driven, real‑time decision logic.

For example, a snack‑chain brand previously needed a week to launch a regional repurchase campaign—exporting data, manually selecting audiences, drafting copy, designing posters, pushing the campaign, and finally reviewing results. After deploying an AI‑powered membership agent that only required a target such as “increase Wuhan repurchase rate by 10%,” the entire workflow became fully automated, delivering a strategy that outperformed manual effort by 210%.

The advantage stems from AI’s ability to analyze millions of user behavior trajectories simultaneously, precisely identifying users with recent repurchase intent and the content types they respond to best.

Similar AI logic now drives budget allocation and promotional pricing: instead of allocating budget based on last year’s channel performance, AI tracks ROI by the day (or hour) and dynamically shifts spend; promotions shift from blanket discounts to targeted couponing for price‑sensitive users and new‑product pushes for insensitive segments.

Chapter 4 and Chapter 7 of the report reconstruct the actual implementation processes of Juewei Food and Procter & Gamble, detailing technical stacks, integration steps, and pitfalls.

The deepest overlooked layer: supply‑chain and marketing coordination

Front‑end AI marketing shows quick wins, but true differentiation emerges when AI aligns supply‑chain and marketing. Historically, FMCG suffers from a “promotion‑without‑supply‑chain” problem, leading to stock‑outs or over‑stocking.

AI‑enhanced demand forecasting, which traditionally achieved ~75% accuracy using only historical sales, can reach 98% when incorporating marketing plans, weather, holidays, and competitor actions. A case study reported a 30% reduction in inventory turnover days after adopting AI‑driven dynamic allocation, freeing capital for additional turnover cycles and cutting storage costs.

At the store‑level, AI vision tools replace manual shelf checks: a store employee uploads a photo, and within seconds AI verifies product placement, stock status, and pricing. One consumer‑goods giant reduced out‑of‑stock rates by 15% using this approach.

These incremental improvements translate into tangible profit gains across the value chain.

Final takeaways

The transformation is not a simple purchase of AI tools or a brief ChatGPT training; it requires rebuilding data foundations, redesigning business processes, and developing organizational capabilities. The full 14‑chapter report offers a maturity assessment model, a roadmap for enterprises of different scales, and a self‑diagnosis questionnaire to help firms identify the most suitable AI entry point and avoid blind investment.