When is it worth implementing an AI-based analysis of your product shelf?


The immense pace of the AI-based image recognition development has made the automated shelf analysis tools that have been around for several years now much more accessible to producers and field force agencies. Data transfer across cellular networks, the quality of images taken with smartphones, and finally the speed of processing data and returning analysis results are no longer obstacles to the implementation of technologies based on neural network deep learning. However, many sales directors and CIOs are still wondering whether or not to implement such an advanced technology.

The replacement of a manual examination of display parameters with one performed by a machine – of those basic parameters (presence, facing), of those more complex ones (share of shelf, share in the category, competitors), or even the most advanced parameters (product blocking, SKU and category proximity for the company’s own and competitors’ products) – releases huge amounts of time and personnel resources. On the other hand, automation inspires a leap into the risky field business analysis and management through goals – and all of this in real time. Just imagine all those multidimensional reports we never created due to the lack of data or for fear of their poor quality!

If it is possible to draw conclusions for a single store and employee using an intelligent display analysis through segments, teams and regions, business strategies, central promotions, trade marketing assumptions and HR policies, then we are facing a significant quality difference which many sales managers have been awaiting for years. However, it is not only the quantity, but also quality and reliability of data that are of equal importance here. The resources of collected data and the possibility of processing them later (e.g. retrospective analysis of previously unmonitored indicators) constitute a competitive advantage that may influence success in a given category or market.

Investment in an intelligent solution such as eLeader Shelf Recognition AI is a decision to enter a project that requires the involvement of an indispensable part of resources. Apart from the people involved in implementing the functionality into the system and establishing the rules and criteria (e.g. KPIs dedicated to display analysis, and no matter what technology we use, such an analysis is always worth performing), we also need some time to train the neural network to recognize products.

DOWNLOAD YOUR FREE WHITEPAPER

In order for the process to run smoothly and bring about a satisfactory ROI, we recommend conducting a basic “analysis”, which in principle requires only the knowledge of the appearance of the products – especially of their packaging labels. Below you can find basic guidelines which will help you determine whether it is worth considering automated analysis.

It is therefore appropriate to consider the implementation of eLeader Shelf Recognition AI if:

• products are sold in packaging;
• the packaging has a shape that distinguishes it from its background;
• packaging labels can be easily distinguished from those of competitors, especially if they differ within the brand and category;
• your products are sold through a modern channel and placed on shelves, e.g. according to a listing;
• you are selling products through a traditional channel using merchandising standards or planograms.

If the answer to most of the above questions is positive, then the basic conditions for the implementation of automation of shelf analysis have been met, and so the project has been given a green light. If, however, the above questions have not been answered in the affirmative, it may be better to stick to traditional methods, with particular emphasis on the possibility of introducing the perfect store strategy within the basic and easily measurable KPIs (presence of products, facing, shelf number, etc.). This will not only facilitate a more effective implementation of the strategy in the field, but will also provide a strong foundation for reliable analysis using business intelligence tools that can operate on large volumes of data coming in regularly and from many sources.

When is a manual shelf audit more effective?

• when the audited products are without packaging, loose or difficult to separate from the immediate surroundings (e.g. bread, cakes);
• when products are sold in small quantities in an unusual environment (e.g. car showroom, customized stands);
• when you are a mirror manufacturer ;).

Naturally, you should always consider changing your trade marketing strategy and introducing at least one product category with packaging that meets the criteria of qualification for an artificial intelligence analysis project.

It is worth noting here that neural networks “see” products in a way very similar to how the human eye perceives them. It is therefore a legitimate thesis that packaging features that do not allow it to be separated from the environment or distinguished from other products are a signal that something needs to be improved so that consumers can ultimately see them better on store shelves.

So what are the challenges for sales in the era of galloping digitization and artificial intelligence that is gaining new footholds in the market? Certainly, it is worth considering at least the possibility of using a technology that eliminates mistakes and errors inherent in human perception and endurance, given that at the same time the business environment and competitors are facing similar challenges and seeking solutions to them. Although deep learning technology is not for everyone, it often turns out that the apparent difficulty does not exclude the use of intelligent analysis (e.g. eLeader’s neural network – Ala works well for round bottles and bags of crisps).

If your products meet the requirements for recognition by means of a modern technology, it is necessary to assess the possibility of implementing the eLeader Shelf Recognition AI at the strategy and equipment level (it must be remembered that this is a system of interconnected vessels when it comes to functionalities like Perfect Store, RAO, or the advanced BI reporting). In order to obtain the necessary knowledge, it is best to talk to an eLeader analyst, who will help you to perform an initial assessment of the feasibility of the implementation and qualify the project for deeper analysis or recommend alternative solutions that will enable a satisfactory implementation of business intelligence and perfect store in your organization.

MORE ABOUT ELEADER SHELF RECOGNITION AI

Read also:

eLeader Retail Portal – mobile B2B marketing platform for your customers

The neural battle or artificial intelligence in sales

Drop a few pounds with Perfect Store!