“CP brands have been discussing predictive analytics for years. But what’s different about 2023 is that CP companies have become much smarter in understanding how data, including artificial intelligence (AI), can be used to inform key business decisions and enhance human experiences,” said Scott Clarke, Vice President of Consumer Products, EMEA and APAC, at Publicis Sapient. “This knowledge—together with an increasing capacity to gather large amounts of data, as well as increased computing power and algorithmic capability—affords CP brands the opportunity to do predictive analysis at scale.”

While the practice is still in its inception among CP firms, a Stanford study analyzing 30,000 manufacturing establishments across industries revealed that businesses that use tools to automate prediction achieve nearly $1 million more in annual sales compared to non-adapting competitors. CP companies that can successfully integrate data from multiple sources and automate for high-value use cases will be able to scale and benefit from this growing technology.

To start, predictive analytics means using historical and real-time data to predict how people will behave in the future—from first-party or third-party consumer data, to social media comments, to weather patterns. Either manually or automatically, companies use this data to determine many different things: what kind of advertising content customers will engage with, which SKUs should be discontinued, at what time marketing emails should be sent or how much inventory should be sent to specific retailers.

The difference between predictive analytics versus general data analytics is the power of the predictive algorithm that harnesses the power of the data to turn it into a proactive decision-making tool. 

In fact, a recent Gartner survey reveals that only 53% of marketing decisions across industries are influenced by data in general, let alone predictive analytics. The ability to use data to make decisions in a proactive, predictive manner remains a source of competitive advantage across the CP industry—and for many CP companies it has emerged as the elusive North Star.

Many remember the classic predictive analytics story from Target that went viral in 2012: The retailer sent coupons for baby products to a pregnant teenager who hadn’t publicly disclosed that she was pregnant. It’s unclear whether Target’s pregnancy prediction algorithm was that advanced at the time or it was just an interesting coincidence, but the story elevated the potential power of AI in business use cases, especially marketing.

Since then, some AI use cases have created an inflated perception of what predictive analytics can do for businesses. The root of this problem is a misunderstanding of how to implement predictive analytics. Businesses need to have the right mixture of relevant expertise and clearly defined use cases to see significant ROI from this strategy.

Thus far, many CP brands have been focused on just one aspect of predictive analytics: building up consumer data sources and figuring out new ways to integrate and store that data. While this has been crucial for an industry that previously didn’t rely on consumer data, most brands have yet to take the next step.

“More data is not necessarily the answer,” Clarke said. “The vast majority of data you capture may not be relevant to the decisions you need to make every day as a business. Defining the right business questions and use cases—and understanding how to deliver those use cases—should really be the focus.”

So, what are the top use cases for predictive analytics in the consumer products industry for 2023, and what steps should CPs take to become successful? The areas where CPs can make the most impact on their customers are predictive content, demand forecasting and consumer trends insights.

“A CP brand may want to recommend a new product to consumers, but the value propositions for different demographics might be very different,” Clarke said. “You might like a new caffeinated beverage to give you an energy boost while working out; another person might drink it to keep them awake while studying. The exact same product is being recommended, but the content being used to inspire people would be very different.”

Implementing predictive analytics allows CPs to automatically serve the optimized homepage or ad to the consumer who’s most likely to identify with it based on an algorithm that learns over time. One way CPs can gather data to create dynamic websites and customer journeys is clickstream analysis. As CPs invest in their brand.com websites, tracking certain metrics—such as where consumers click on pages, how much time they spend on different portions of the homepage and what steps they take on the path to purchase—can help determine which products and content appeal to customers.

“Understanding which clicks are intentional versus unintentional—and being able to test different content combinations for different types of users—will become increasingly important and a mandate for CPs,” Clarke said.

Some CP firms are partnering with third-party technology to deliver dynamic content, while other firms are building out analytical capabilities in-house. For example, Mondelēz partnered with an outside provider to create personalized CTV ads across devices based on data like geography, weather and other third-party information. Kraft Heinz has instead built out a CDP with AI/ML capabilities, called “Kraft-O-Matic,” to improve ad efficiency and discover other customer insights.

Whether in-house or external, it’s important that firms use a cloud-based CDP that can integrate with all different data sources and that retail partners can easily access in real time. It’s also crucial to use KPIs to embed data into daily working practices instead of continuing with legacy processes.

Another crucial post-pandemic use case for predictive analytics is inventory management and demand forecasting. In many instances, COVID-19 broke formulas and algorithms put in place to determine consumer demand. Major global events can negate even intelligent machine learning based on years of data. How can CPs use predictive analytics during “unprecedented times,” especially with a lack of consumer visibility compared to retailers?

One way to begin creating a new demand forecasting model is by incorporating consumer behavior insights from new digital channels. Rather than factoring in just the typical variables, like seasonal purchase patterns from retail data, CPs can factor in customer data from D2C channels.

Consider that 90% of the world’s data has been created in the last two years, according to the U.S. Chamber of Commerce. From weather, to social media trends, to Google search data, to first-party customer data—and second-party data from retailers—CPs have a variety of sources that, once integrated, can help to create more accurate demand forecasts. However, it’s crucial to have a scalable, cloud-based data-sharing platform before integrating new data sources. To start creating a more accurate demand forecasting algorithm, CP firms can slowly integrate external data to increase accuracy.

For example, US Foods migrated from an on-premises SQL data server to the cloud to unify internal and external data for demand forecasting. Because of the newly streamlined data ingestion, the company uses additional resources to incorporate weather and clickstream data from its e-commerce business to create more accurate forecasts. Now, according to a data scientist at US Foods, this data collaboration platform saves the company $100,000 per year.

Another CP firm, Campbell Arnott's, integrated its supply and demand planning data to decrease manual forecasting and improve overall forecast accuracy. This allowed the data team to test for erratic demand scenarios—and begin to use a weekly-based forecast—as well as increase prediction accuracy by 15%.

The first step to being able to integrate new types of data sources for demand planning is having a solid data-sharing backbone that works for external partners—and internal talent that can experiment with the data to create more accuracy for future demand.

Another growing digital frontier for CPs is predictive analytics for the research and development of new products. From fashion to food and beverage, customers’ behavior and preferences are constantly changing, and using data to predict these trends is becoming a mandate.

Many CPs are already utilizing predictive trend data through social listening algorithms and new data channels, such as “smart" devices like coffee machines, thermostats and more. Data from these channels can work together in unison to form powerful behavioral insights and help inform where to focus R&D efforts.

For example, the Keurig smart coffee maker allows users to customize their coffee through a Wi-Fi-enabled app. The coffee makers can track which types of coffee and flavor combinations customers are using, and the company can use that data to predict customer trends and develop new K-Cup products.

The AI company Tastewise uses data from restaurants, social media and online recipes to mine data to see which ingredients and foods are trending. Rather than look at surveys or industry reports, the company allows CPs to analyze real-time online content from millions of different sources.

However, in order to see results, brands have to take the next step and test these new products, styles or variations that may be vastly different from the company’s previous legacy. “Most companies want growth, and you don’t get that through optimizing what you have today,” said Simon James, International Lead Data & AI. “You have to be prepared to test new things. A zero tolerance to risk is not the safe bet that you think it is.”

Predictive analytics also come into play when deciding which products to de-list or sell. Unilever, for example, created an AI-powered tool to determine which products should be discontinued based on first-party and retailer data. For large CPs with hundreds of brands, using historical data from different sources will be crucial to determine which products under those brands are fit for a digital future.

The first step for CPs is to identify a use case for a new product and determine which data sources could be informative for that use case. Rather than reactively jumping on trends to inform SKU rationalization, CP firms should be proactively deciding which brands or product categories are ripe for experimentation—whether or not current data sources can give a well-rounded recommendation.

For CPs that aren’t experimenting with predictive analytics, there are a few important considerations to keep in mind in 2023. The first is data privacy. “If anything happens, like a data breach, you’re going to lose some of your consumers,” Papasakelariou said.

Another key consideration for CP leaders is change management for analytics technology and processes. It takes time for teams to get acclimated to new data sources for decision-making, and executives should make sure they have buy-in and provide training to account for the adjustment.

“People must be held accountable to acting on data,” Clarke said. “Teams also need to be creative and inspired in terms of coming up with the right use cases (i.e., the things that you’re going to measure and inform through the data).”

Finally, as CPs begin to use data more creatively to predict the future, companies that can easily share and view data in real time will be able to use their data more efficiently and creatively.