This is part two of our series designed to help organizations gain efficiencies, cut unnecessary costs, and ultimately protect their bottom line in today’s uncertain landscape. The Kepler platform was developed to address a vast array of Machine Learning (ML) use cases in the CPG industry, and its easy-to-use Automated Data Science Workflows ensure all CPG leaders, and their teams, no matter the department, can leverage the benefits of advanced ML within days — not weeks. Below, we outline how to leverage Kepler’s capabilities to overcome challenges brought by demand fluctuations and new consumer shopping patterns with the level of adaptiveness required to stay afloat in the market today.
As COVID-19 continues to affect not only our health but also our economy, the highly competitive consumer packaged goods (CPG) market is experiencing a unique set of challenges. Our previous piece on the retail industry highlighted the fact that nearly half of Americans will be cutting back on “unnecessary” spending right now, a fact that will undoubtedly have lasting financial impacts on CPG companies, as well.
Given that the CPG industry encompasses products deemed “essentials” as well as “non-essentials,” companies are experiencing sharp increases in certain products, while witnessing stagnant or declining sales in others: some products fly off the shelves, while others remain stagnant. More specifically, recent consumer data reveals that companies are experiencing demand surges for highly specific product categories, including: shelf-stable or frozen foods, household cleaning products, and even at-home hair dyes.
According to recent data collected by Nielsen, buying patterns exhibited by Americans reveal that this demographic has moved to the “pantry preparation” COVID-19 buying phase in which – due to self-isolation and restricted movement – sales in the following “health and wellness” products have grown significantly. See the below chart for a full illustration.
With tighter household budgets and a greater desire to achieve wellness, consumers are adding a significant strain to specific goods. Adding to the complexity of nuanced shopping patterns, CPG companies have to overcome the challenge of forced store closures and reduced hours. For many products, brick-and-mortar retail was the primary channel for generating sales and revenue prior to COVID-19.
A relevant report by BCG, distributed last year at this time, urged the importance of CPG organizations to improve their ecommerce capabilities quickly, in an effort to catch up with new consumer behaviours. Further, the report urged CPG leaders to reduce reliance on brick and mortar stores to effectively increase year-over-year numbers.
Evidently, CPG leaders need to ensure that the online sales number is much higher in digital channels to stay afloat in 2020, and likely beyond.
The good news? If the shift to optimizing and improving end-to-end ecommerce capabilities wasn’t already in full swing, there are ways that Machine Learning can not only get CPG companies caught up to frontrunners, but also prepare them for the coming years.
Currently, CPG businesses are faced with challenges that pivot towards one extreme or the other, experiencing a sudden spike in demand for products, adding unforeseen pressures to their supply chain to ramp up production; or, a decrease in demand and sales, resulting in an overstock of non-durable, short-lived goods and a significant lag in sales. This is sure to affect companies specializing in items that fall outside of the realm of what’s dubbed as a “pandemic pantry”. Items encompassing confections, luxury beauty and alcohol are readily considered non-essential, where, a Bain & Company article reports sharp declines in demand for these.
That being said, it’s important to remember that new challenges also breed an opportunity to refine and improve tactics, for both mid-term and long-term growth. As McKinsey outlines in its extensive COVID-19 report for industries, optimism about a sudden return of demand is dangerous: leaders should be equipping themselves with the tools they need to face a protracted downturn. With this in mind, CPG leaders in all business units should think strategically about how AI investments today can help drive short-term efficiencies and bottom line value, while also aligning with long-term strategies for growth.
With a massive acceleration towards online shopping among consumers, CPG leaders can make use of fresh data to better understand product demand and adjust production and supply accordingly. The ecommerce avenue is crucial for CPG businesses to tap into, allowing them to reach new categories of customers, including those who previously made up a small fraction of online shoppers. For this reason, sharpening understanding of product demand through new distribution channels is crucial for CPG producers seeking to strategize for today’s quick wins, and tomorrow’s long-term customer loyalty.
Keeping these challenges and opportunities in mind, Kepler’s advanced Machine Learning and Deep Learning capabilities can help you extract essential insights from your data to achieve efficiency-driving and cost-cutting outcomes, through the following use cases:
How can Kepler’s Automated Data Science Workflows help you leverage advanced ML for personalized product offers?
By using customer transaction data, CPG businesses can boost their online sales through personalized product offers, with the aim of providing customized product recommendations to online shoppers to maximize their potential for conversion.
To maximize personalized product offers, Kepler’s recommender system workflow leverages data in tabular form of customers’ purchase histories.
How does Kepler’s powerful recommender system workflow function? First, the Kepler user creates a model using historical data of customers’ purchase histories, namely, items previously purchased or the ratings given to those items. Note that within this creation process, Kepler’s automated functionality ensures that the workflow can be used by anyone on your team, with no previous ML experience required. Once the model is built, you receive insights that provide the most relevant recommendations for segments of your online customers, through the discovery of patterns in your dataset. These patterns consist of items each customer is likely to purchase, which would speed up the conversion process, from browsing to checkout.
How can Kepler’s Automated Data Science Workflows help you leverage advanced ML for demand forecasting?
With the help of historical sales data, CPG leaders can accurately predict and prepare for short-term and long-term demand for their products, allowing them to adjust their pricing accordingly, as well as limit costs and maximize sales. Kepler’s Time Series Forecasting workflow allows you to better anticipate and prepare for a surge or decline in demand for your product, providing you with a clearer picture of how your sales will evolve over a given period of time.
Using tabular data, you can leverage this workflow for demand forecasting, to determine the popularity of existing products which can in turn help you make more informed decisions on the sales patterns of new products. Accurate forecasting results in reduced inventory of excess stock, accurate fulfillment, and a consistent ability to meet ongoing customer demand.
How can Kepler’s Automated Data Science Workflows help you leverage advanced ML for optimized customer segmentation?
A prominent use case affecting CPG businesses is customer segmentation, in which customers are identified and categorized in order to hone in on their preferences and needs.
To achieve customer segmentation, Kepler’s clustering workflow accurately targets the most viable customers for your products. Clustering for this workflow involves selecting the metrics that you want users to be clustered on. Metrics include factors like annual income, days since last purchase, number of purchases, net revenue, etc. Based on the results, users can try to identify a common theme or characteristics among the users that fall under each cluster.
Using tabular data, clustering allows you to bundle data of similar “groups” and identify segments of the population based on specific demographic characteristics, such as age, location, previous purchases, or spending frequency. Through your trove of customer data, you can create more precise customer segments by suggesting clusters of attributes that would not be readily apparent without Machine Learning. Kepler’s clustering workflow empowers you to create highly specific buyer groups to target consumers with the right products at the right time.