Shelf Auditing: Why AI Visibility is Transforming FMCG

09/10/2026

Introduction: The Harsh Reality of 'Share of Shelf' and Display Scoring

Visit any supermarket in Hanoi or Ho Chi Minh City, and you will see shelves in a state of chaotic disarray. Competitor products tower over yours, while your stock is hidden behind packaging. This is a phenomenon every FMCG distributor knows, yet few can accurately measure its frequency. Share of shelf is not just a statistical figure; it is a brand's survival metric at the point of sale. If products are not present in the right position and proportion, sales drop immediately, often without you realizing the cause until the next quarterly sales report is released.

The core issue lies in the display scoring process. This is the final step in the value chain from factory to shelf. If this step is weak, all previous marketing efforts are wasted. Currently, most companies are still relying on inefficient manual or semi-automated processes. We will dissect this process, examine the fatal mistakes managers often make, and identify where AI Visibility (through AI Vision technology) can intervene most effectively.

Mistake 1: Relying on 'Human Eyes' and Outdated Excel Reports

This is the most common error, particularly in small and medium-sized enterprises. The process typically looks like this: a supervisor walks around the store, takes a few photos with a personal phone, then sits at the office to 'visually inspect' and fill out an Excel file. A single photo can take 5 to 10 minutes to analyze manually if detailed SKU-level requirements are needed.

What are the consequences? First, the speed is too slow. Information about whether products are being overshadowed by competitors reaches the marketing department too late to take action. Second, human error. When tired, staff may miss an out-of-stock SKU or misjudge the display area ratio. The gap between shelf reality and reported data can be as high as 20-30%. In a chain with thousands of stores, the risk of staff arbitrarily editing data or even 'fudging' reports to show progress is significant. You are making strategic decisions based on distorted data.

Mistake 2: Assuming Photos Are Enough, Ignoring Input Data Validation

Many companies believe that simply collecting photos is sufficient for processing. They overlook a crucial technical detail: the quality and authenticity of the input data. In practice, staff may screenshot images from the internet, submit old photos to report, or take pictures from obscured angles where the camera cannot recognize barcodes or product names.

The lack of an automated validation layer makes the entire process fragile. If the AI system receives a fake image, it will produce incorrect results, and management will waste time on manual cross-checking. This is where AI Vision (AI Visibility) technology shines. Rather than being just a simple face or text recognition tool, specialized retail solutions like AIVISION's VTraks focus on detecting duplicate, fake, and screenshot images right at the input stage. If an image does not meet standards, the system rejects processing or issues a warning, ensuring that all data entering the dashboard is 'meat' rather than 'garbage'. This saves hours of work for the operations team.

Mistake 3: Manual Scoring and Lack of Consistency in Planogram Criteria

Even with high-quality images, scoring planogram (shelf layout drawing) compliance often suffers from discrepancies due to inconsistent criteria. Staff member A might give a score of 8/10 because the stock is sufficient, while staff member B gives 5/10 because the position is incorrect. This subjectivity undermines the ability to compare performance across stores or regions.

The solution lies in fully automating the scoring process. With an announced accuracy of 99.97% on vtraks.com, current AI solutions can process each image in less than 1 second. The system compares the actual image with a pre-defined standard planogram and provides an objective 'Pass' or 'Fail' result. There is no more debate about human perception. Data from millions of stores is aggregated into a single dashboard, allowing managers to see a real-time overview of Share of Shelf. When an area with abnormally low compliance is detected, the sales team can intervene immediately, rather than waiting until the weekend.

Trade-offs and Unresolved Issues in the AI Vision Process

Implementing AI Visibility is not a magic wand that turns everything green. There are trade-offs you must accept. Initial costs for system integration and training staff to take photos correctly still exist. Furthermore, AI Vision only solves the 'seeing' and 'measuring' steps. It does not automatically arrange products on the shelf. Humans still need to go on-site to move products, apply price tags, and handle expired stock. AI is a decision-support tool, not a cleaning robot.

In addition, the quality of the original planogram is a critical factor. If the planogram you input into the system is incorrect or does not fit the actual store layout, the scoring results will reflect that discrepancy. Companies need to continuously update and refine planograms based on feedback data from the AI system itself. This is a continuous improvement loop that requires close coordination between marketing, sales, and IT departments. Currently, no solution fully automates the 'action' step after a discrepancy is detected. Integrating with sales management apps or staff time-tracking software to automatically generate tasks for the sales team remains a gap that needs to be filled. However, gaining accurate, fast, and honest data through AI Vision has already solved 80% of the transparency pain points in traditional retail channels. This is a solid foundation for any growth strategy to stand firm.

There is more here than one article can hold. Keep reading on the AIVISION blog, look at our display scoring solution, or get in touch.