Artificial intelligence and merchandising10 августа 2020
It’s not a secret that merchandising is an essential tool of the modern trading and marketing system. According to statistics, the ideal implementation of merchandising standards gives 7 to 15 percent increase in sales, and maintaining the sales area in perfect condition directly affects customer satisfaction and aesthetic pleasure from being in the supermarket, stimulates re-purchase and growth of average check. Merchandising is the 10 KPIs of the routine and accurate work of field staff, each hour of which costs money. This means that retailers are increasingly looking at solutions for optimizing routine work and automating simple processes.
Among trendsetters and merchandising innovators, for example, is Walmart with a recently announced system based on artificial intelligence, which makes it possible to monitor goods on shelves in real time.
But no matter how sophisticated retailers are, no matter what rebranding they come up with, key merchandising KPIs have been and remain the basis of sales:
— if the product is not on the shelf – it is not for sale
— if the price tag does not match the product – customer loyalty is reduced
— chaos on trading shelves, incl. caused by super-deals and discount offers, leads to the fact that the retailer loses the loyalty of the next flow of customers if he does not have time to display products.
In a long chain from producer to consumer, there is an important link – the merchandiser. And while robots have not replaced this position, such as the Schnuck chain of American grocery stores with the Simbe Robotics Tally robot-merchandiser, the merchandiser is still responsible for the quality standards and appearance of the trading shelf.
The evolution of merchandising
Over the past 10-15 years, the merchandiser has evolved from a courier with a folder of papers filling in checkmarks in paper questionnaires into an advanced user of software products on tablets and smartphones. Even 10 years ago, merchandisers and sales representatives visited retail chains with cluttered paper cases and order forms. Distributors’ offices were filled with operators who took orders, digitized tons of paper, formed orders and built logistics routes. Sometimes even weeks passed from the moment of receiving the order to delivery.
Another area of responsibility of the merchandiser is to follow the display of goods criteria and compliance with the planogram. Such reconciliation took place manually according to the printed pictures: the merchandiser using a tape measure checked the length of the trading shelf with his products, controlling whether the competitors moved the brand by 5-10 cm.
The advent of mobile computers has begun an era of automation for merchandisers. The first-generation devices were specialized handheld computers (computers, not smartphones) with push-button interfaces for data input. In fact, the merchandiser, as before, recorded data, but now not on paper, but in an electronic device. So it was no need to enter data from paper reports into control systems.
A breakthrough step in the work of the merchandiser was the emergence of a smartphone, which made mobile data transfer, determination of geolocation and photographing possible. This launched the second generation of systems. SFA systems (Sales Force Automation) appeared, which were installed as applications on smartphones. Now the results of the work of merchandisers were transferred via mobile networks in real mode. The location of the employee and the schedule of his movement became available. As a confirmation of his work, the merchandiser could now take photographs of shelves and attach them to his report, which he, as before, had done manually. The progress was obvious – reporting speed increased, control over the work of the merchandiser also increased, and the price of the system decreased: almost any smartphone with a camera and geolocation was suitable for work.
Similar solutions works in Ukraine:
But there were also some problem areas. There were terabytes of photo material and, as it turned out, that all SFA solutions have the same problem. Photos are stored on servers and only 10-15% are viewed, analyzed and participate in strategic decisions.
The development of computer vision based on machine learning and progress of smartphone cameras have brought third-generation systems to the market. This technology has radically changed the work of the merchandiser. Now the report about the visit was made not by him, but by a machine, which, using the photo, analyzed the availability of goods on the shelf, compliance with planogram and price tags. From the merchandiser was only required to visit the point of sale and take a photo. Accordingly, the speed of the merchandiser grew, and the requirements for their qualifications dramatically fell. There were even crowdsourcing-based services that set tasks through applications, for example, for random people to go to the nearest supermarket and take a photo. The rest of the work was done by the algorithm. Despite the demand for such technology from customers, it did not immediately conquer the market.
Key Image Recognition Retail Players
Training the machine to recognize the image of goods on the shelf until recently was a very difficult technical task. Only two companies in the world successfully managed it: Trax (https://traxretail.com/) and Planorama (https://planorama.com/), which divided large customers around the world. Medium and small customers, however, could not afford such a service. Only for the initial training of the neural network you had to pay tens of thousands of dollars. At the same time, Trax mainly worked in America, and Planorama – in Europe. This year, Trax became a global leader in this market by acquiring Planorama. At the same time, technology did not stand still and the cost of machine learning began to decline, which made possible the emergence of small local companies that began to offer similar services for much less money and at the same time adapted their services to the needs of local customers. In Poland, this is eleader.biz, in India – BIZOM (https://bizom.in) and retail-scan (https://www.retail-scan.com), in Turkey – vispera.co, in Russia – e.g. SmartMerch (http://smartmerch.it/). In Ukraine, the startup Picsell (https://picsell.ai/) occupied this niche.
What is going to be in the future?
… further development of smartphone cameras and application functionality for merchandisers. The latest models of smartphones begin to work with a three-dimensional image, which means it will be possible to measure the availability of goods in the depth of the shelf.Well, fourth-generation systems will eliminate the human factor completely. Cheaper cameras and connecting them to 5G networks will enable continuous monitoring of shelf conditions. The need for control visits will disappear completely and the number of necessary merchandisers should be reduced by several times. In general, the prediction of futurologists that artificial intelligence will take away work from a person relates exactly to a specific area – merchandising.