First, a story. Li Hua works at a well-known FMCG company. Starting in January 2017, the headquarters used a sales management system to monitor sales reps' daily work: at least 8 hours of attendance, visiting at least 15 stores, and within work hours, taking photos at intervals as evidence. For each store, they had to photograph the storefront, shelf display, cooler, cut-case display, price tags, and more. This is the daily routine of an FMCG sales rep. Who knows how much Li Hua hates such phone app monitoring software. To get their regular attendance pay, they battle these apps daily. Every system has loopholes; as the saying goes, "The higher the policy, the lower the countermeasure." Sales reps are not pushovers, so they resort to tricks like photographing computer screens, color pages, or photos on other phones; taking a photo of one store's front and another store's shelf, and so on. These cheating methods are endless. The headquarters not only has to assign a dedicated audit clerk to randomly check uploaded data in the backend but also send out inspection teams offline to track and take evidence photos... This is just the tip of the iceberg of false data that FMCG manufacturers encounter in offline data collection, management, and expenses. It not only consumes a lot of manpower and resources but also fails to guarantee data authenticity. Such data loses much of its reference value. Manufacturers are also worried: spending tens of millions a year to develop a comprehensive sales management data system, only to end up with data full of water, and market staff complain. They say, "In the era of big data, why does my data have little reference value?"
The Power of Artificial Intelligence In recent years, AI applications have appeared in various industries, especially with the maturation of speech recognition, image recognition, natural language processing, and other related technologies, leading the implementation of AI business scenarios in robotics, autonomous driving, smart speakers, unmanned stores, smart stores, smart manufacturing, and more. Recently, New Distribution interviewed Ms. Wang Qin, co-founder of Guangzhou ImageDT Data Technology Co., Ltd. (ImageDT), to understand what new industry changes the combination of AI and the FMCG industry can bring. With 10 years of experience in the FMCG industry, Wang Qin deeply felt the difficulty of data collection in this industry. When she saw the application of AI, especially image recognition technology, her intuition told her this might be an effective solution to the data collection difficulties in FMCG. 2016 was a turning point. Wang Qin discovered that using image recognition to monitor product information on shelves and grasp the overall market dynamics was a good entry point. Coincidentally, she met two partners with technical backgrounds, and the three hit it off. In November 2016, they co-founded Guangzhou ImageDT Data Technology Co., Ltd. (ImageDT), combining technical capabilities with industry knowledge. The trio embarked on a new attempt to fully empower FMCG with AI.
(ImageDT co-founding team)
Leveraging AI to Empower the Industry FMCG industry management is relatively rough. This is also why FMCG distributors have survived for so many years. In the past, brand owners couldn't reach the terminal; the market in T1-T6 cities is complex and diverse. To sell products to every corner of the country, they rely on a vast network of distributors. In the era when distributors bought goods from manufacturers, manufacturers often didn't even have a proper system; most data was manually recorded. It wasn't until after 2000 that some decent systems like SFA (Sales Force Automation) appeared. These systems gave many brand owners confidence, at least they were no longer as blind as before. Previously, data was reported by distributors, and they didn't know if it was accurate. Now, at least they could have approximate data on how many terminal stores were in a region. However, data collection still heavily relies on sales reps, requiring them to continuously input correct data, which inevitably leads to errors like forgetting to record, recording incorrectly, or missing records. Against the backdrop of continuous deep exploration of AI technology, ImageDT found that image recognition technology can effectively solve the inherent drawbacks of traditional manual order taking and entry, making data collection faster and ensuring the authenticity of data value.
"With image recognition technology, the work of sales reps becomes simpler; they no longer need to calculate or fill in data. At the same time, the data capture is very structured, so reps don't have to worry about forgetting to fill in or missing information, ensuring that frontline staff continuously collect market photos effectively and truthfully," Wang Qin told New Distribution.
If Sales Reps Are Ineffective, Let AI Do the Data Work Although data-driven management is often mentioned by FMCG brand owners, very few companies truly achieve big data management. Many companies have adopted SFA systems, but they may use a management tool for a long time without updates. Apart from adding new features, there is no fundamental change. "The real transformation is to achieve a complete change at the operational level of sales reps, turning tasks that take an hour on SFA into just a few minutes," Wang Qin said. This is what ImageDT is continuously pushing forward.
Another important point: data authenticity. For brand owners, using software is a necessity; they must use it. But after using it, sales reps spend an hour collecting data at each store, and the data management gets is fake, forming a vicious cycle that makes data lose its original meaning.
"One of the important problems image recognition solves is helping companies understand the real frontline data," Wang Qin told New Distribution.
Image recognition helps companies build a trust relationship. Sales reps take photos to give the company real shelf data. After receiving the data, the company can guide overall optimization and rectification based on the data, ultimately achieving a climb in overall store product data, forming a positive improvement loop. For sales reps, they only need to take photos, and AI automatically generates a report for them, showing which shelf displays are insufficient and which need restocking, making immediate rectification clear.
If Expenses Are Hard to Control, Let AI Handle Verification Market expense verification is another headache for brand manufacturers. All brand owners have to hire large inspection teams year-round to monitor market staff behavior. The manpower and travel costs for inspections are a huge expense every year. In the past, for a promotional activity, they had to keep receipts and photos, organize them monthly, print activity materials, and mail them to headquarters for review. During this process, there are issues like lost receipts, missed photos, and long reimbursement cycles. ImageDT applies image recognition technology to expense verification. After an activity, all materials can be uploaded to the expense management system by taking photos. AI image recognition can both prevent fraud and improve the efficiency of the verification process.
"From the most basic execution monitoring for clients, to the fastest generation of data reports, to data analysis, and even to marketing strategy and strategic formulation, ImageDT can assist clients," said Wang Qin, co-founder of ImageDT. "ImageDT is not just a platform product; we also hope to do a client deeply and thoroughly. In fact, we hope to be a consulting company, providing one-on-one customized solutions for brand owners."
Final Thoughts: If we set aside the high-sounding term "artificial intelligence" and look from the perspective of brand owners, in the past, the information chain from brand owner to consumer was very long and fragmented. Product display, promotional placement, and customer data were difficult to manage effectively. Take promotions as an example: during a promotional activity, brand owners distribute promotional prizes, but they cannot guarantee that the promotional items and policies reach the terminal and are displayed. Distributors might take them, sales reps might intercept them, or store owners might sell them as regular products. Thus, there is a lot of waste in promotional placement. Even if there is placement, market feedback is hard to obtain because information has to go up through layers of distributors, and it often doesn't get through. So brand owners have to hire market research companies to survey how many people participated, how many saw it, and what their feedback was. But such feedback is often delayed. Also, surveys are sampled and may not represent the whole, and research costs are very high, making effective evaluation difficult. The application of AI in the FMCG distribution channel is essentially about completing the digital transformation of the channel, driving business efficiency through intelligence. Efficiency will inevitably replace inefficiency. As technology iterates and improves, it will accelerate the transformation of the FMCG industry. Many AI technology companies like ImageDT are continuously penetrating the industry, empowering brand owners, and achieving industry-wide transformation.