ConAgra’s Quest to Gain Actionable Insights from Social Media Data

by   |   August 28, 2013 1:56 pm   |   0 Comments

Freddy Guard of ConAgra Foods

Freddy Guard of ConAgra Foods

CHICAGO—Measuring how American shoppers consume media and marketing messages is an enormous challenge for Freddy Guard, director of advanced analytics at ConAgra Foods, Inc.

Watching television is not a singular activity anymore, for example. Seventy-seven percent of consumers use another device while watching TV, Guard said, adding that 90 percent of shoppers use multiple screens to make a single transaction. Understanding how these intersecting activities correlate to brand awareness and sales is essential to measuring the effectiveness of ConAgra’s marketing efforts.

Related Stories

Four approaches for integrating social media data with enterprise systems.
Read the story »

Five questions social media marketing analytics can answer.
Read the story »

Podcast: The rise of real-time marketing in a data-driven world.
Read the story »

Operations executives: soft skills indispensable for analytics pros.
Read the story »

Guard was among more than a dozen speakers at the Institute for Operations Research and the Management Sciences (INFORMS) Midwest Conference on Aug. 21 in Chicago, attended by more than 100 analytics professionals.

In his presentation, “Evolving Marketing Analytics for Impact,” Guard detailed how analytics are making a difference at $18 billion ConAgra, North America’s largest food manufacturer. ConAgra brands, such as Healthy Choice, Hunt’s and Reddi Wip, can be found in 97 percent of American homes. Of the company’s 40-plus consumer brands, 25 are $100 million businesses and two are closing in on $1 billion.

In the past dozen years, media channels have diversified, creating many more consumer touch points and “complicating the path to purchase,” said Guard, who heads a group of 10 in marketing analytics. Across ConAgra, there are some 40 other data analysts.

Among these new channels are social media platforms like Facebook and Twitter, which are extremely popular and influential among Millennials, those born between the early 1980s and the early 2000s.

“Millennials today have $200 billion in direct purchasing power,” Guard said.

From an analytics perspective, this means there can be as many as “20 or 30” media elements in any particular campaign. Complicating matters further, product managers want to know if their marketing activities are having a long-term, brand-building impact, Guard said.

Of course, ConAgra’s leaders also want quick answers to these questions, delivered in time to make a business decision.

“The major gripe we hear is, ‘I wish the analysis came quicker,’” Guard said.

Facing these requirements, Guard’s approach has been to work with his line-of-business colleagues to shift the focus of their analytics efforts. Instead of simply confirming something, he said analytics should “yield something new” and “drive action.”

Organizational Changes to Support New Approaches
This reorientation required a number of changes. For starters, the data analytics organization had to insert itself into the planning calendar for marketing campaigns “and get a seat at the table” with executives, Guard said.

“The biggest change we have made is bringing together the medial effectiveness pieces together,” Guard wrote in a follow-up email. “We have started integrating the pre-market results from the copy testing results (likeability, breakthrough, branding, etc.), the in-market communications effectiveness and brand health tracking (change in brand perceptions and purchase intent, etc.) along with our marketing mix models (media effectiveness in driving volume) to present a comprehensive look at our campaign performances.”

Moreover, Guard said he strives to communicate these findings to his business partners “in a business context, in an easy to digest 15-20 pages at most, rather than 100-page decks for each study separately.”

“I would say across the board we have become more informed and have moved our marketing ROI in the right direction,” Guard wrote.

A second change: ConAgra’s advertising agency partners were tasked to gather some types of data, such as social media, “and synthesize it for us in a way we could use it for analysis,” he said.

But here Guard implied CPG companies like ConAgra still needed to up their in-house game in the brave new world of social media. “You need to get skills and experience with new, unstructured data types,” he said.

Another change was the approach to automation and models.

Given the shifting nature of brands and consumers, Guard said it is unwise to try and fully automate processes to collect and analyze data, which assumes a static environment. Instead, he recommended “selective automation,” adding that it was equally important to work with flexible modeling platforms. (Both Guard’s group and ConAgra finance use Tableau Software’s data visualization product.)

Like several other speakers at INFORMS, Guard emphasized the need for data analysts to get beyond data collection and modeling, and immerse themselves in the business–its brands, products and consumers.

Not only will this help every analyst “measure what matters,” it will encourage data professionals to build alliances with business owners and learn to communicate with them in their own language, he said.

Advice from Veteran Practitioners
Navigating the complexities and politics of the business organization, and keeping data projects on track and relevant, was a recurring theme at the half-day event, co-hosted by the University of Chicago’s Graham School. Appropriately, this topic was the explicit focus of the final session of the day, a panel discussion featuring four seasoned data executives.  Their recommendations included:

1. Communicate clearly to business leaders. “Speak the language of the business,” said Suzanne Fogel, chair of the Department of Marketing at DePaul University’s Driehaus College of Business. “Show how the analysis will “make a difference for strategy and outcomes.”

2. Foster internal relationships. Don’t neglect personal relationships and “connections across the organization,” said Kim VanderSchaaf, vice president at BMO Harris Bank, where she leads the commercial, business and small business banking analytics team. These relationships lead to trust, so that when an algorithm is developed, the business leader will believe in it, even if he or she doesn’t understand all the underlying math.

3. Use narratives to give meaning to statistics. “Storytelling, without complicated numbers, is important,” said Jan Gollins, principal and founder of the Delta Modelling Group. He added: “Use information to help them, not beat them up.”

4. Lay the groundwork for analytics that can change assumptions. Since some of the most-powerful models can challenge long-held beliefs inside the organization, it’s a good idea to ask for permission to have that conversation, said Steven LaVoie, founder and CEO of ArrowStream, a supply chain technology company.

“It’s refreshing to have an intellectual debate, to say, ‘Let’s figure this out together,’” LaVoie said. This technique also makes the business leader a part owner of the data project, and therefore invested in its success, VanderSchaaf said.

Ellis Booker is a freelance journalist based in Evanston, Ill. Email him at Follow him on Twitter: @ellisbooker.

Tags: , ,

Post a Comment

Your email is never published nor shared. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>