Ukrainian startup PICSELL announces the start of cooperation with HENKEL

CMS Group is successfully expanding its expertise beyond marketing communications. Four years ago, the CMS Ventures program was launched, the goal of which is the development of technology startups. The team of PICSELL (within CMS Group) recently signed a contract with the international FMCG manufacturer — HENKEL. The essence of the PICSELL product is to optimize merchandising and increase efficiency with the help of computer vision and machine learning technologies.

Implementation of the PICSELL system into the client’s workflow.

HENKEL team was looking for a solution that would help increase efficiency and optimize staff costs. The PICSELL product realized this through the transfering many routine processes to the already available computer vision technology, which scanned photos received from the field and processed them automatically. In addition, a Sales Force Automation Solution (SFA) was developed for HENKEL that helped establish the structure interaction process at all levels.

The pilot period lasted 9 months. The launch process went through stages of 100 SKUs. Each step took from 1 to 2 months. At the moment, about 90-95% of SKUs are automated. Recognition of advertising materials, additional points of sale, and control of promotional periods were left for the last stages.

The main benefits of the PICSELL and HENKEL collaboration.

1. Optimization of the working hours of merchandisers.

Firstly, there is a reduction in the time spent by the merchandiser at each particular store. Secondly, it reduces the cost of training employees. Previously, merchandisers should have known 450 HENKEL product SKUs, planograms for each point of sale, assortment matrix, shelf length occupied by product category. They had to understand with which competitors in the distribution network to compare promotional periods, prices and brand representation in terms of shelf length. The merchandiser had to manually enter all the data into the old-style system, the new staff made many technical errors in the early stages, which cast doubt on the correctness of the data. Now merchandisers use PICSELL software, where they just need to take correct photo. That is, thanks to the system, the client reduced the costs associated with staff training, and the time to fill out reports; there is confidence in the correctness of the data. Moreover, time is freed up that the merchandiser can spend either on visiting more points of sale, or working with inventory, with the design and organization of the logistics process from the manufacturer to retail.

The following steps relate directly to the introduction of such innovations in the nearest future:

2. Optimization of routes of sales representatives.

PICSELL team plans to develop a customer management system for the merchandising team, which would help not only to automate routine processes, but also to collect data on the routes of merchandisers and compare it with the real situation on the shelves. Next, an advisory model for changing the routes of merchandisers will be prepared, based on an analysis of historical data obtained at each specific point.

3. Display of goods optimization.

Based on the accumulated data on the recognition of SKUs on the shelves, the client will receive recommendations on how best to make the display of goods, which product category should be expanded, and which should be arranged differently. The system is able to show how often a particular SKU is Out-of-stock in certain networks. This is important information in the negotiations between the network and the manufacturer. The client will now be able to calculate by himself how much he "under-earned" due to the fact that the goods are not on the shelves. The integration of such systems allows to receive data much faster.

What difficulties did client encounter while implementing the PICSELL system?

Recognition accuracy, the reason for which is not good enough camera quality on the devices of sales representatives. As a result, the client had to replace smartphones in some regions. Since 2018, the devices with high-resolution cameras have become much cheaper. Now a phone for $ 100 allows you to have a camera from 12 to 16 megapixels, this is more than enough to get high-quality photos.

The automation of routine work is the future. Such solutions make it possible to respond several times faster and see the problem situation in retail with the company’s brands, to better approach the issue of finding additional opportunities to increase sales and profitability. And, of course, the main function is to increase efficiency and reduce the cost of maintaining the merchandising structure. The next step for us will be the analysis of historical data collected by the PICSELL system, and the integration of the module for generating route recommendations for each individual point or network”, — comments Vasyl Grigoruk, National Sales Manager, category "Laundry and home care products".

If we talk about global trends, computer vision technology is becoming available for integration into business. To create a neural network, or to take a turnkey solution and adapt it to the needs of the business — today it isn’t difficult. But this is not the problem for the implementation of such projects. The difficulty lies in the formation of large volumes of data sets for training a neural network, as well as checking for errors that at times reduce the accuracy of recognition and automation of the process as a whole. Such checks are mainly carried out manually and require trained personnel. On the one hand, we have experience in managing large personnel structures; if necessary, we can scale up to several thousand part-time operators in a couple of weeks. On the other hand, we have created an entire ecosystem that integrates the human resources & neural network into a single platform. Thanks to this decision, already at the initial stage we can get the recognition accuracy of goods on the shelves in the range of 90-95%, and the speed of launching pilot projects has increased significantly. Now, from the moment of signing the contract for the test project to obtaining the results, it takes less than 4 weeks”, — notes Sergii Germashev, CEO of PICSELL.