If you are a tech-savvy digitization enthusiast, you would know about recommenders and how it has the potential to change the way you work with suggestions and insights. However, few people know how much work goes into building a highly effective recommendation engine and the real struggles data scientists face to achieve these results.
Building a Recommender from scratch, based on the hunch that it will become one of the most sought-after inventions of all time, is a story worth reading. Preeti Menon, acing to perfect a technological advancement that’s relatively new and picking pace, shares her story and struggles with making Ivy Recommender one of the most accurate and actionable recommenders currently available in the market.
Conceiving the future
Our product strategists, as usual, conceived this idea ahead of its time and took it forward. Our consumer goods industry knowledge and our passion for making our solutions better and smarter helped us transition from a Unified Commercial Platform to an Intelligent Route to Market. We could create a working model with 95% accuracy in 1 and a half years.
Perfecting the Algorithm
From the beginning itself, our teams had a focused mission, idea, and vision. We created an almost perfect algorithm that would work for companies that varied largely in size and offerings. Despite all ups and downs while creating the solution, we believed that an intelligent factor would change how consumer goods companies focused on their customers.
With data scientists and developers on board, the team customized the shortlisted algorithm to suit the specific needs of our niche. Quick optimization helped us to meet and ace industry standards of recommender performance. As development progressed, the team also realized that the usual static model wouldn’t work, so we had to recreate a dynamic model that utilized results from previous operations and reprocessed it to learn, thereby increasing the accuracy of the results of the AI&ML engine.
Improvising the Performance
The team hoped to provide our customers with at least a 2% revenue boost by promoting same-store sales. This ROI was a bet any CIO from the CPG world would have gladly taken. With time, our team achieved 5% the revenue growth with our dynamic algorithm for a handful of customers, and we’re hoping to see an even better number.
With years of data, our solution curates smart insights and actionable recommendations that simplify field agents’ sales strategies, allowing them to achieve their KPIs faster while helping manufacturers generate more revenue by implementing our solution.
Born digital CPG companies have been randomly collecting data, while legacy companies never had an opportunity to look at their data at a single location. Over the years, we have helped some large conglomerates collect and store data in a single data mart to identify and retrieve actionable insights. Our team obtained constant support and recommendations from our customers, who readily helped us incorporate valuable inputs to make their recommender more robust and powerful.
Futuristic Use Cases to benefit customers
With access to historical data, Ivy Recommender can guide sales teams with recommendations that are tried and tested at stores of the same segment based on demographics to help produce assured results. These recommendations include Next best SKUs, Must Sell Lines, in addition to providing training and development for field force agents.
Let’s conclude with a quote from Preeti, “Intelligent recommendation systems are yet to take the CPG world by storm, and we are already building a product promising scalability and assured success.”
Learn more about Ivy Recommender here