AI helps Land O’ Lakes churn data into sales

When it comes to selling products, nothing quite gives an edge like knowing exactly where to target your sales efforts — and at what price. Thanks to the confluence of sophisticated data strategies and artificial intelligence, insights like these are fast becoming reality.

Like many businesses, Fortune 500 dairy giant Land O’ Lakes is constantly seeking to optimize its pricing, better target sales, and forecast future demand. For more than a year, data analytics and AI have been key to its efforts.

Beginning in the fourth quarter of 2017, Land O’ Lakes began to deploy elements of its Data to Value program, which brings AI techniques to bear on the company’s sales and marketing efforts. The project leverages data analytics tools and disparate data sources to achieve insights into the company’s profitability, sales call success factors, and commodity markets that Jeremy Dumond, director of business intelligence at Land O’ Lakes, says were previously unattainable.

“The world of data is growing exponentially, and the challenge of boiling it down to actionable insights that drive profitable growth gets more challenging each day,” Dumond says.

As the company’s investment in data and technology grows, the demand for provable returns on that investment is growing right alongside, Dumond says. The company’s customer base is evolving rapidly, too. That makes delivering the right product at the right price at the right time a constant challenge for Land O’ Lakes, which boasted $14.9 billion in revenue in 2018.

“Acquiring new customers at prices that enhance our bottom-line profits was the ultimate mission for the project — not a new mission by any means, but we utilized a new capability to accelerate our results in the marketplace,” Dumond says. “Due to the efforts and output of this project, we have greater confidence than ever that our salespeople are equipped with the right products at the right prices for the right customers.”

The project has won Land O’ Lakes a CIO 100 Award in IT Excellence.

Optimizing sales with data

The company’s dairy food business serves a wide array of customers, including chain restaurants, independent restaurants, and delis, the diverse nature of which makes setting prices a challenge. Even a minor shift in pricing can translate to significant changes to the bottom line.

To tackle pricing optimization, Land O’ Lakes partnered with an external AI vendor. But for challenges unique to Land O’ Lakes’ agricultural business, such as propensity modeling and commodity forecasting, the company leveraged internal expertise.

Propensity modeling is all about targeting sales more efficiently. Given Land O’ Lakes’ footprint, the company has a massive number of potential sales targets for its limited sales resources. The dairy sales team needed a way to identify prospects that had the highest propensity to buy at the highest volumes.

On the forecasting front, the company’s risk management team needed better insights to drive decisions around hedging, inventory strategy, and spending.

Dumond says the company’s leadership was very receptive to doing things differently, but it was still essential to get everyone on board with the proposed changes. This meant helping everyone to reach a common understanding of the project’s possibilities and limits, and ensuring that the company wasn’t pursuing “tech for the sake of tech.” Data to Value was a coordinated effort between Land O’ Lakes’ IT, sales, marketing, and finance units.

“The modeling itself was very complex, and we explained it to the degree that inspired confidence in the approach, but the real selling point was the simple math behind the improvement in our marketplace execution success rates and how that translated to profit,” Dumond says. “Growing our business by x drove y in incremental profit, which far exceeded the necessary investment in technology and was thus worth the effort.”

Change management breeds success

In fact, Dumond says selling the improvement potential of the model to the internal sales organization was the project’s biggest challenge.

“There were big increases in ‘odds of success’ and ‘size of prize’ compared to our previous strategies, which themselves had been quite successful,” Dumond says. “It was a big bet to lean into the model recommendations and we did so gradually, but in a statistically significant way, to prove it as we went and built credibility with the selling team.”

It would have been easier to make the change if the company had been struggling, Dumond says, but the company was growing, which made it more challenging to change the approach. But the credibility of the business leaders coordinating with his team help facilitate the process.

The development team started by taking an inventory of available data, identifying what it already had and what it needed to drive the necessary insights. Ultimately, the project made use of a diverse set of internal and external data, for which the company needed data ingestion capabilities beyond its traditional ETL tools, so it turned to open source Apache Nifi.

Dumond’s team leveraged an agile project environment with weekly check-ins to address the inevitable challenges that emerge in the flow of data and the model iterations.

In the end, Data to Value has helped Land O’ Lakes improve its success rate on sales calls by more than 40 percent, Dumond says, and the average operator is buying 12 percent more than they did two years ago. The company has added sales resources in places it didn’t previously cover, as the model has shown where it can win in any market, not just where it’s won previously. It’s also led Dumond’s team to expand the model to other channels and products.

“Building a marketplace-backed project plan, in partnership with trusted business leaders and a commitment from the top of the organization, are foundational to success,” Dumond says. “Keeping the lens of execution in mind, with a corresponding respect for the ‘possible,’ helps to keep everyone grounded in the value realization curve and moving at the right pace.”