Zooming out, one can see that forecasting, when done right, is just another piece of the retail process structure. Inventory and assortment planning, macro or micro space planning, behavioral clustering, and workforce optimization are also among the functional components of an efficiently planned retail infrastructure. Having positioned itself as a ‘unified retail planning solutions’ provider, RELEX Solutions headquartered in Helsinki, Finland shines light across the wide spectrum of retail-centric processes with its technology offerings. RELEX’s clientele ranges from retailers that pertain to seasonally volatile and demand-driven sectors such as fashion, grocery, home and gardening, building merchants, and consumer electronics to name a few. Even players in the e-commerce space can equally leverage the company’s solutions.
Multi-model Approach to Forecasting
“At RELEX, we live and breathe forecasting. We have specialized in forecasting for supply chain planning and optimization for over a decade and have gradually expanded into using forecasts in other business domains such as space and assortment optimization. At present, our SaaS solution runs about ten billion forecast calculations daily, and 100 billion a week, for our customers,” writes Mikko Kärkkäinen, Group CEO at RELEX.
Their approach to forecasting embraces both automation and machine learning—the crucial elements that cater to today’s retail market’s imperative need for speed in arriving at actionable insights. With RELEX’s Business Rules engine, retailers can automate demand forecasting routines by identifying external events such as seasonal changes, promotions, trends, or weather. Additionally, RELEX’s supply chain-optimized, columnar in-memory database enables retailers to perform calculations at the most granular level. Analysts can inspect and analyze data in real time in accordance with store, region, product group, Stock Keeping Unit (SKU), sell-by-date weather, and event.
Operational excellence—in merchandising, supply chain, and store operations—is required for retailers to survive in today’s extremely competitive market
Relex can carry out the demand forecasting of 1,000 stores and 1,000,000 SKUs two years into the future in less than 60 minutes.
Machine learning, having garnered immense popularity across industries, is widely employed and ‘insanely promoted’ by many other forecasting solution providers in the space. But untamed machine learning-based forecasting that hasn’t been subjected to checks (via a comparative approach) can easily go wrong. The sector is not short of stories pertaining to failed applications of machine-learning-based supply chain forecasting. Decisions made on errors that are not rooted out early can have disastrous consequences. The very term implies that machines need time and trials to ‘learn from errors and feedbacks.’ These over-fitted, shoehorned machine learning models tend to produce unstable forecasts with extremities in their values. As a matter of fact, many of RELEX’s newer customers had to learn it the ‘hard way’ through previous implementations of machine learning-backed forecasting solutions from other providers.
RELEX considers machine learning-backed forecasting as just another tool whose algorithms find best use in analyzing the impact of different weather variables on the sale of items to update sales forecasts. RELEX also combines to its forecasting approach, the proven methods of time series forecasting and regression-based forecasting. As a result, analysts can run multiple forecasts for the same scenario using a different model for each and attain predictable degrees of errors in the forecasts. RELEX’s UI makes it intuitive to compare relative accuracies of forecasts. The company goes on to imbibe more flexibility into the forecasting technique by leveraging in-memory computing (IMC) that works upon RELEX’s in-memory database so that any snapshot can be picked from any moment and compared with any other or with the present. For instance, the accuracy of a demand forecast made at the time of product stocking can be compared against one made at the time of starting sales. Furthermore, causes for discrepancies in levels of accuracy can be zeroed in by checking for factors such as weather, or ‘trends.’ Addressing the retailer’s essential need to generate forecasts for newly-introduced products, the company has built an auto-reference model which can automatically identify the best available reference product for any new item. With all these arsenals for demand forecasting, even a small team of three-five analysts can run a retail operation with thousands of stores, and can even work based on several time and season horizons.
For any solution/product offering, the implementation and ease of use are just as important as its robust capabilities; and RELEX realizes this postulate in its entirety. The company maintains that no coding or consultancy is required to use RELEX’s offerings which can adapt ‘on the fly’ within weeks. Moreover, dedicated project managers with vast expertise in SCM are assigned to clients. By embracing agile principles, RELEX takes clients through a brief current state analysis followed by the technical implementation that typically lasts for 8-12 weeks. Additionally, the offering can be integrated with over 30 known ERP systems. The company also offers a certified interface for data supply from SAP Retail systems. As a policy, RELEX charges zero upfront fee or incentives payable until visible value is attained by the client. To give a preview of the benefits to their potential client, RELEX usually starts projects by running their historical data through the company’s systems. And it is not surprising to see how the company has in its sleeve some of the most reputed names in the retail industry as their clients, including Rossmann, WHSmith, Morrisons, Coop Denmark, and AO.com to name a few.
Tuko Logistics is a Finnish grocery wholesaler and supply chain service provider, who delivers approximately 25,000 products from 1,600 suppliers daily. The company needed a system that enabled it to plan its capacity and workforce, handle purchasing process and optimize purchasing costs, inventory and stock values. In partnership with RELEX, Tuko was able to see improvement in the forecast accuracy within two-three weeks. The company saw 40 percent storage cuts with 8-10 percent decrease in the inventory. Tuko was also able to cut down the number of planning staff by half.
"For any solution/product offering, implementation and ease of use is just as important as its robust capabilities; and RELEX realizes this postulate in entirety"
Impressive Track Record, Promising Future
As logistics scientists and experts in the realm of Supply Chain, Kärkkäinen founded RELEX along with Johanna Småros and Michael Falck who today, leads the company as CMO and President, North America, respectively. Since its humble beginnings in 2005, Relex has grown tremendously to establish an intercontinental presence with offices in the Europe and U.S., and strategic partnerships that extend all the way to South Africa. In the last couple of years, along with key acquisitions, RELEX has nearly tripled its revenues by entering the North American market and adding over 90 new customers worldwide. RELEX recently closed a second round of funding from global growth equity investor Summit Partners who continue to harbor firm belief on the company’s promising roadmap. “Operational excellence—in merchandising, supply chain, and store operations—is required for retailers to survive in today’s extremely competitive market. Unifying these core operational processes enables retailers to really step up their game. We are working to establish RELEX as the clear leader in the unified retail planning category,” concludes Kärkkäinen.