The Consumer Packaged Goods (CPG) industry is experiencing rapid growth, as consumer demand for everyday products, such as food, beverages, household items and toiletries, have remained high, even in times of financial crisis (with rising inflation and worldwide supply chain issues). This means retail execution is more important than ever. To illustrate, the global CPG market will reach $2,460 billion by 2028, representing a Compound Annual Growth Rate of 3% from figures in 2022.
![The global CPG market will reach $2,460 billion by 2028](https://static.wixstatic.com/media/841b36_08c3e3e450af4124984ebf7a1136e774~mv2.png/v1/fill/w_980,h_1038,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/841b36_08c3e3e450af4124984ebf7a1136e774~mv2.png)
The CPG space is also fiercely competitive. For example, CPGs must defend their profits from new players entering the market, and the increasing prevalence of Retail Private Labelling. For example, since Q2 2022, Retail Private Labelling profits in stores like Aldi and Target have increased by an average of 2%.
Therefore, for CPGs to maintain growth and edge ahead of their competition, brands must gain richer and more accurate store-level insights faster and more cost-effectively.
The shop floor represents the final frontier in the sales process. So, companies invest heavily in Retail Execution (REX) tools such as Image Recognition (IR) technology to deliver the comprehensive store insights they need to generate more sales.
Theoretically, one photo file uploaded to an image recognition tool can reveal multiple KPIs such as planogram compliance, pricing, and promotional information, etc. Image recognition is also capable of delivering these results in a short time frame (typically in a few minutes, if not seconds).
![](https://static.wixstatic.com/media/841b36_693352b4d70a43ae9411ea6447aed9d5~mv2.png/v1/fill/w_980,h_1025,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/841b36_693352b4d70a43ae9411ea6447aed9d5~mv2.png)
Unfortunately, however, this isn't the reality with CPG's experience with traditional image recognition. A recent survey found that 49% of CPG brands feel that their data and insights are not leveraged to the fullest extent. This may be due, in large part, to the unreliableness of the technology, as planning accuracy was ranked as retailers’ second biggest priority for improving their brand in 2023.
The drawbacks of traditional image recognition are caused by the laborious process of manually acquiring images and annotating data, which is time-consuming and often yields inadequate amounts of data. In addition, once the IR is developed, environmental factors like lighting and narrow aisle widths in certain store formats can impede data accuracy, hindering CPGs' abilities to make the necessary improvements in their retail execution strategies.
![Traditional Vs Synthetic Image Recognition Methods](https://static.wixstatic.com/media/841b36_aeff61d8ffec4006bb5c143a3819f3fa~mv2.png/v1/fill/w_980,h_572,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/841b36_aeff61d8ffec4006bb5c143a3819f3fa~mv2.png)
To counteract these pitfalls, new generation image recognition technologies leveraging synthetic data, such as Neurolabs' ZIA, can help CPGs gain confidence in their data once more and help them utilise insights more effectively. Read on to find out how.
Gaining better CPG data and insights starts with optimising retail execution processes
Customer experience is central to all successful REX strategies. Therefore, CPG brands must ensure that products are readily available, shelves look orderly and well-presented, and their goods are competitively priced. For more information on meeting customers' high expectations on the shop floor, read our post on building the 'Perfect Store' strategy.
Unfortunately, CPGs using traditional IR technologies face some challenges. Namely, these technologies rely on real-world data to be developed, making them vulnerable to disadvantages posed by adverse retail environments, such as poor lighting and layout, which can impede data accuracy.
Gathering large amounts of real data to train IR learning algorithms can also be a resource-intensive and time-consuming process that may be prone to human error.
Enter Synthetic Image Recognition. With synthetic image recognition, CPGs can benefit from more reliable and accurate data, leading to greater efficiency in auditing processes and, ultimately, optimal guided decisions in store execution that lead to increased sales. The following section will explain how this can be achieved through our solution.
How synthetic data generation provides better data-driven insights for CPGs
Synthetic image recognition uses computer-generated digital twins of the products, generated directly from product packaging labels or artwork, to train machine learning algorithms. It then analyses shelf images to identify the products within them and their proper placement.
![](https://static.wixstatic.com/media/841b36_e1c5aa56c1374a43a77e90cc83eb3dc9~mv2.png/v1/fill/w_980,h_572,al_c,q_90,usm_0.66_1.00_0.01,enc_auto/841b36_e1c5aa56c1374a43a77e90cc83eb3dc9~mv2.png)
With this in mind, here are some examples of how synthetic IR can help provide better data insights for CPGs:
Improved data quality
Synthetic IR produces highly-accurate insights, streamlining the product detection process. The outcome; CPGs get better quality data (with fewer errors and more accuracy), enhancing retail decision-making.
For instance, Neurolabs’ ZIA has cloud connectivity which lets field marketing agents (FMAs) on the shop floor see real-time insights on store shelves. With ZIA data insight accuracy doesn't deteriorate over time because the data is generated using computer algorithms, which can be easily tweaked and updated to reflect changes in the real world. Subsequently, synthetic data generation also helps FMAs quickly see what issues need to be fixed, such as low stock availability.
In contrast, traditional IR data struggles to maintain real-time accuracy because the data is based on fixed features extracted from a limited set of real-world examples. Traditional IR training data can also quickly become outdated as new environmental variations emerge in the real world. As such, this explains why so many users may have had a poor experience with traditional IR as it can be slow to adapt to new retail scenarios.
Diversity of training data
Synthetic image recognition offers a significant advantage over traditional IR because it can generate diverse data sets for training algorithms. It utilises synthetic computer vision to recreate product variations, environments, and lighting conditions without human input. This means synthetic data generation is proficient in creating a greater variety of training data sets, providing CPGs with various insights to analyse and test that wouldn't be feasible with a real-data approach.
This level of functionality allows ZIA to “imagine” how an SKU would look like on various shelf environments prior to the real SKU being on the shelves.
Subsequently, the ability to create diverse training data ultimately leads to a robust, high-performance technology with high accuracy guaranteed. In addition, this level of accuracy is achieved very fast (a few hours/days) because Neurolabs’ ZIA can automatically generate high volumes of training data in hours, giving the tool a distinct edge over traditional IR.
Low complexity in integrating Synthetic Image Recognition
Synthetic image recognition technology can adapt and scale with the ever-changing needs of CPGs. This is because solutions like Neurolabs’ ZIA stores all data in the cloud and can integrate with existing Sales Force Automation (SFA) tools. As a result, it's easy for field agents to enhance their solutions with a powerful and readily available technology – namely, synthetic image recognition.
To illustrate, CPGs can update master product catalogues and artworks within their SFA. Neurolabs’ ZIA can access this data and automatically update the product recognition models in record time.
In contrast, updating SKU changes with traditional IR technologies typically takes days to complete. The benefit of this type of functionality ensures FMAs working on behalf of CPGs have the most up-to-date data sets to help them perform their daily tasks and stay ahead of competitors.
Neurolabs ZIA: enhancing data-driven retail execution through Synthetic Image Recognition
Neurolabs’ ZIA also offers dedicated customer support to ensure that CPGs and FMAs get the best insights from its synthetic data generation to ensure retail execution is optimised to the highest standard. Overall, Neurolabs ZIA offers a reliable and scalable technology, so companies can worry less about the IR data gathering process and focus more on building their brand .
If you want to learn more about how Neurolabs’ ZIA can bring the gold standard of image recognition technologies to your CPG brand, you can find out more here or alternatively, contact us for a demo today.
At Neurolabs, we are revolutionising in-store retail performance with our advanced image recognition technology, ZIA. Our cutting-edge technology enables retailers, field marketing agencies and CPG brands to optimise store execution, enhance the customer experience, and boost revenue as we build the most comprehensive 3D asset library for product recognition in the CPG industry.
Comments