Global direct-selling company Oriflame is a catalogue-based sales company with customers in 60 countries worldwide. The company's business model relies on sales representatives to get the catalogue out to customers.
Offering a diverse portfolio of nature-inspired beauty and wellness products to so many different markets, Oriflame sought to tailor its catalogue to provide the products and offers best suited to each audience, a move that would also enable it to forecast for the business more accurately.
Oriflame brought in a machine learning solution using Microsoft Azure which was able to determine the discounts to offer in each market, which products should be featured on the front and back covers and what price they should be retailed at.
Jakub Orsag, IT Research and Development (R&D) Manager at Oriflame, said: “We wanted a solution that could predict, for example, the impact of selling a lipstick with a 20 per cent or 40 per cent discount, the difference it will make to the company’s sales, the impact it would have in different countries, and so on. We were looking for an out-of-the-box service that could help us build a solution even without data scientists."
“When determining which cloud would be most suitable for us to create our solution, Microsoft Azure emerged as a winner, as it already provided a service called Azure Machine Learning, and we just needed elementary knowledge to determine which rules or data we needed to input,” Orsag added.
The IT team at Oriflame ran various algorithms through Microsoft Azure DevOps and were able to create a solution that consisted of three main modules: automated anomaly detection, product forecast and merchandising plug-in. Using these, the company can now predict how it can achieve the best margins
Orsag said: “In some markets, Oriflame only had only a couple of people covering the entire demand and forecasting process of the catalogue, employees were quite eager to have the process automated, as it was saving them time. The story was different in other markets, where there were more people working on forecasting, and their predictions were already quite precise."
"However, the team realised that by letting the machine do the regular forecasting, they would be able to focus on tasks with added value—for ad-hoc campaigns, for example. We envision that the machine is suggesting, and employees are just checking and supervising. It is almost like waving a magic wand.”
In future, Oriflame hopes to train its machine learning solution to predict the individual impact of new products and new campaigns.