One of the world’s second-largest wine and spirit manufacturers, with a presence in more than 150 countries, continues to expand to new geographies every few years while increasing market share in existing markets.
In 2022, the sales for the company grew by 17%, but the brand visibility and Share of Shelf in new markets were low. The company’s existing Retail Execution solution for tracking sales and merchandising activities was clunky and time-consuming, with connectivity issues in remote areas.
The company needed a solution to simplify merchandising activities at retail stores and track KPIs closely in real time. That is when we partnered with the alcohol giant to implement our Image Recognition solution, Ivy Eye, for automating in-store merchandising activities for the company.
Need for Granular KPI Tracking & Visibility
The alcohol company tracked the retail execution KPIs using a smartphone-based data entry solution, and finishing the outlet coverage with the existing merchandising field force took time. With the company expanding into newer markets, hiring and training field force agents was a big challenge.
For in-store merchandising, the field force was tracking only shelf availability at a brand level and the share of a shelf of products. With limited field force bandwidth, granular KPIs tracking such as SKU level OSA, Planogram adherence, and Eye level product presence was difficult.
The company needed a system to seamlessly integrate with its current merchandising activities while automating all the necessary steps to achieve maximum productivity and coverage.
Automation & Real-time Monitoring with Ivy Eye
Ivy Eye, an integrated image recognition solution for consumer goods retail execution, was implemented for the leading alcohol company. The modern retail channel in India is one of the fastest emerging markets for the company, and that is where we first started implementing Ivy Eye.
- The image recognition technology of Ivy Eye automated KPI generation through image capture, which meant a single photo of the captured shelf was sufficient to track KPIs like Shelf Availability and Planogram compliance.
- The Automation with Ivy Eye helped them reduce the time spent by the merchandisers at the outlets. We cut down the time spent by merchandisers from 45 minutes to less than 20 minutes at the stores.
- The solution provided near real-time KPIs to the field users so they could take corrective actions if there were gaps in the KPI actuals vs. target.
The following KPIs were auto-enabled to capture through an image recognition system specifically trained to identify varying sizes and varieties of liquor bottles.
- OSA (On-shelf Availability)
- OOS (Out of Stock) at the SKU level
- Facings at the SKU level
- Share of Shelf
- Planogram Adherence
- SKU presence at Eye Level
- Price Tracking
- POSM Tracking
The KPI tracking was done for a select set of competitors as well. This way, the brand could compare pricing, variants, and competitors’ sales to develop marketing campaigns and strategies.
Challenges of Image Recognition Implementation
As with any computer vision technology, implementing an image recognition solution on the ground is always challenging. Challenges start from infrastructure, processes, and regulations, going all the way to adoption by the field force used to doing things manually. Here are a few tech issues we faced with Ivy Eye implementation for this leading alcohol company.
Varying Product Sizes
- The Image recognition system initially couldn’t identify identical products with minor size differences. To resolve this, Ivy developed simple technological modifications like using aspect ratio and related product size features as part of the machine training and post-processing to distinguish these products seamlessly. The feedback loop ensures that with every image captured, the system trains itself.
Store Lighting Conditions
- Image recognition systems always struggle with low lighting conditions at the outlets. However, our solution has an inbuilt lighting detector to enable Flash Feature when the lighting condition is below the threshold value.
- The system found it difficult to process images captured at an angle. To avoid poor-quality images based on capture angle and edges, we developed a feature to ensure that the phone is parallel to the shelf while capturing images. The edges had to align with the camera focus, which improved image quality.
- Ivy Eye initially didn’t work well on images captured from a long distance. The system needed images captured from 2-3 feet from the shelf. We eliminated this issue by training merchandisers and field reps and providing strict guidelines to capture the images uniformly.
Measuring Real Impact
2.8 minutes of image processing time
55% of time savings
5% improvement in OSA
Ivy eye implementation improved business visibility by tracking additional KPIs and increasing the granularity of KPI tracking. Improvement in execution resulted in a 5% improvement in OSA in 4 months of execution.