Kellogg exec on AI uses cases, implementation, and ‘culture change’

In this era of evolving technology, organizations must be highly adaptive to succeed. A Statistics report reveals that before the pandemic over 4.7 million people in the U.S. were working remotely at least half the time — a percentage that has since increased. And fully 75% of people using digital channels for the first time indicate that they’ll continue to use them when things return to a post-pandemic “normal.”

Kellogg is among the brands adopting strategies for flexible, more versatile work. For example, the company’s journey to build analytics solutions has yielded tools that can identify indicators of consumer change and the profile of a shopper by demographic. Kellogg also now leverages data to target campaigns at consumer bases and route retail sales, automating and optimizing search and inventory for ecommerce.

“We’re using AI across our entire business — from creating supply chain efficiencies to identifying the best flavor combinations for new foods to focusing marketing on key cohorts and identifying which content will resonate most with them to identifying outlet execution opportunities,” Monica McGurk, Kellogg’s chief global growth officer, said in a conversation with VentureBeat. “Every function and business unit at Kellogg prioritizes use cases based on factors ranging from their business impact to their ability to advance our technology to their ripple effect from the standpoint of change management … Ultimately, adoption of AI is about a whole new way of working.”

Use cases

On the supply chain side, Kellogg says it employs AI to send the materials and products to the places at the right time — and for the right cost. One of the company’s technologies constantly reviews disparate sources of data relating to various demand signals. When it senses a disruption, or even a pattern that might lead to a disruption, it provides a recommendation on how to avoid it.

“The challenges across supply chains, in any industry, have not changed much in decades … [With this AI-powered solution,] we’re able to make adjustments days to weeks in advance to prevent an out-of-stock issue from ever happening. We piloted the technology in 2020 with different brands, including Cheez-It, and saw meaningful improvements in service,” McGurk said. “That said, it’s not a silver bullet. For example, if you have no capacity, machine learning cannot fix that underlying problem — but it can help you make the most of what you’ve got.”

According to McGurk, the pandemic heightened the importance of investments Kellogg has made in AI in recent years. Among the challenges it and other consumer packaged goods businesses faced were maintaining supply chains and predicting granular sources of demand as local shutdowns impacted access to channels and eating behaviors. Kellogg was also forced to find ways to support its customers even when physically entering a store to take an order wasn’t possible, as well as ensuring its brands were showing up “in force” across the ecommerce channels that surged during lockdowns.

“With the pandemic came unexpected datasets we’re using in new ways — for instance, as we look to keep the consumers who tried or rediscovered Kellogg foods during the pandemic,” McGurk said. “No one knows exactly what the future will hold, but I have every expectation we’ll continue to use AI in new ways to solve both old and new challenges. Plus, the capabilities we’ve now built into our core processes are here to stay.”

Kellogg’s adoption of AI has been largely a success, enabling the company to reduce waste in its supply chain and boost sales. The company has launched pilots across a range of use cases, including dynamically optimizing the routing of its store-level salesforce in the U.S. to focus efforts on outlets with the biggest upside. In the U.S., India, and other markets, it’s tapping machine learning to predict online out-of-stocks up to 15 days in advance, enabling Kellogg to trigger reordering or steer promotional dollars away from certain foods. And the company has started using AI to analyze the demographics and preferences of geographic areas in Mexico to personalize its point-of-sale marketing in individual stores.

“Personalization isn’t necessarily easy to accomplish for a business of our size, but that’s where AI comes in. The benefit of AI is that it will help us create the personalization people crave — at scale,” McGurk said. “Most of our application of AI and machine learning has been linked to a hard business case or specific marketplace problem we are trying to solve. That ensures we are focusing our resources on something that will matter to our actual business performance.”

Learnings

On the subject of ethics and responsible AI, McGurk says Kellogg follows laws and regulations around data privacy and consent and makes sure its partners do the same. While she notes that local regulatory frameworks governing the use of AI are still “quite mushy,” McGurk says Kellogg is internally focused on understanding where possible risks and unintended consequences might be so it can get ahead of them. For example, the company takes steps to prevent bias in designing and training algorithms, particularly in situations that require “human empathy” from automation.

“We’ve made it a priority to train employees and partners who engage in things like data-driven marketing,” McGurk said. “[We want them to] be aware of local regulations and our own policies, certifying their compliance.”

To enterprises embarking on their own digital transformations, McGurk recommends avoiding an “If we build it, they will come” mentality. Avoiding falling behind requires getting the necessary data in order and identifying scenarios that can provide practical application and near-term business value.

According to IBM, AI enterprise adoption challenges include limited expertise and a lack of tools for developing models. While over 90% of businesses told the company in a survey that their ability to explain how AI arrived at a decision is important, more than half cited problems getting there, including biased data. Still, IDC predicts businesses will overcome these blockers to spend $77.6 billion on AI in 2022, up from $24 billion last year.

“Gain momentum by proving what you can do. Build your tribe of evangelists,” McGurk said. “Implementing AI requires culture change: a willingness to test and learn — and do it with speed. No matter what the corporate culture, training internal users — the ultimate customers for these algorithms — how to be better consumers of data and work with data science talent will be paramount.”