How Deep Learning Supports Growth-Driven Forecasting


Consumer goods companies are transitioning into an era with unprecedented requirements for speed, scale and intelligence.

As their lives become more digital, hectic and overscheduled, consumers are making purchase decisions based on an increasingly vast network of inputs, influencers and options. As a result, consumer goods companies must find ways to forecast customer needs, build products accordingly, and not just have a presence, but thrive, through every shopping channel.

Of course, this is easier said than done. Many CGs are struggling to establish modern agility based on a foundation of outdated analytics, ineffective marketing and lopsided relationships with retail partners. They're often incredibly data-rich, but because the data lives in silos, in many cases they're actually quite data-poor when it comes to accessing real-time unified insights that drive modern success ― the reason for having data in the first place.

Sourcing Beyond Traditional Limits
Traditional advertising strategies that once were used to drive sales and provide a strong ROI for CGs have become less influential as consumers turn to social media networks, online search, and even smart assistants to inform product evaluations.

In order to resonate with their target audiences, CGs must not only develop an organic, authentic understanding of what consumers are looking for, but also what factors influence their decisions. They must gain this understanding across all possible touch points if they hope to influence buying behavior and forecast accurately.

As mentioned, staying relevant and competitive in the marketplace requires access to the right data. But collecting, tracking and making use of this information requires a bit more than a standard spreadsheet.

Enter AI
Fortunately, artificial intelligence makes it easy to analyze data collected from a gamut of data sources spread across various disciplines — syndicated, loyalty, social media and omnichannel. The impact of AI is expected to be so meaningful that Accenture expects potential profitability rate increases at an average of 38% across industries by 2035.

The union of massive computing power and endless data sets has led to the development of a class of machine learning algorithms (the foundation of AI) called deep learning. Deep learning models develop nuanced perspectives using layers of connected nodes (like neurons in a human brain) that can detect incredibly complex patterns.

As an example of its value, deep learning is the backbone in facial recognition and self-driving car technology. Deep learning has the unique ability learn from a mistake just as humans do, and then apply corrective measures so that the mistake doesn’t recur down the line. CGs can unleash the power of AI to redefine the way they source and manage consumer data and create incredible new efficiencies.

Becoming a 'Tastemaker'
One particularly exciting frontier for AI-driven consumer insights is demand forecasting. AI with deep learning capabilities allows CGs to dive far deeper into their analysis,  transitioning from a SKU-location level to understanding buying behavior at the household and shopper level. The AI approach to forecasting leverages data sets across retailers and channels and delivers insights about consumer engagement, brand preference, price sensitivity and lifestyle choices.

Using AI to forecast marries historical data, such as sales from past promotions, with granular, forward-looking elasticity models that are far more effective at predicting lift than traditional elasticity models. Using deep learning to develop more dynamic forecasting makes it more accurate and, ultimately, far more valuable. In fact, early internal studies suggest AI-powered forecasts reduce traditional statistical errors by 25% to 50%.

In doing so, CGs establish themselves as influencers in their industries, the "tastemakers" who don’t just respond to consumer preference, but who meet the consumer at every pivot, ready and supportive of their new whims.

Retail Partnerships Become 'Primary Intelligence Units'
Here’s the real kicker, though. AI doesn’t just make owned data more valuable; it makes data from across the retail partner ecosystem more actionable.

AI-enabled solutions do this by allowing CGs and their retail partners to analyze massive amounts of shared information through a single platform. Any manufacturer account manager can easily collaborate with retail category managers on a unified, standardized system to develop targeted assortments that match what shoppers want. It’s a powerful way to promote ongoing, collaborative innovation in product development, distribution and promotions.

The CG landscape isn’t getting any less competitive, and every company is looking for its edge. Fortunately, as fundamental questions of competence and competition are answered by innovative leaders, greater efficiency and intelligence will become commonplace. To do so, CGs must break down silos within their organizations, and also between themselves and their partners. They must become increasingly proactive, deliberate and data-driven.

CGs are already seeing tremendous benefits from leveraging deep learning to create more resilient, growth-oriented businesses and products and strategies that cater to the exact, evolving requirements of their consumers.

About the Author
As senior vice president of product strategy for Symphony RetailAI, Sy Fahimi is responsible for product strategy, direction and execution.