Analysis of Shelf Display for Major Tobacco Manufacturer
Business Challenge
Solution
The pilot project was launched successfully due to sufficient number of renders to train the model.
Users can log into the application and confirm the completion of tasks without an Internet connection. If there is no Internet connection, the online map is not available, and outlets can be selected from a list. Photo recognition results are downloaded as soon as network connection is available.
The web application allows downloading a database of retail outlets and editing them. The functionality of creating and loading a planogram and planolisting is available.
The mobile application can open photos and view the results for each SKU, including layout’s compliance with the planogram. The interface is intuitive and user-friendly. Outlets can be found both on the map and by name; there is sorting by name and distance. The outlets can be searched only by full RCS. Users cannot view detailed history of previous visits.
The web interface provides the same functionality as the mobile application. It is possible to check the planogram and SKU directly in the photo. However, there is no reporting (dashboard) in the web interface as part of the pilot project.
Highlights
100+
merchandisers
226
SKU (including competitors)
5 000
outlets
70 000
photos analyzed monthly
Result
It takes on average 15 seconds to obtain results.
Since using Goods Checker, new products have been launched. The ability to recognize them was available from the date they entered the market. It took 2 weeks to train with a new category; the recognition quality was 95%. It took 2-3 days to train with a new product (cigarettes), the recognition quality was 96%.
Following the pilot project, the number of analyzed photos increased from 3% to 100%, and compliance with planograms improved from 60% to 90%.