You’ve never heard of them—and that’s by design. Siloam Springs, Arkansas-based Simmons Foods (Simmons) is solely devoted to the business-to-business poultry market. The time Simmons doesn’t spend branding its products to the consumer market can be used to better serve their business customers’ needs.
Simmons employs more than 6,000 people across its Poultry, Pet Food, and other business segments; the company reported revenue in excess of $1 billion in 2010. Simmons wanted to make better use of its data in order to save money through improved processes and better demand planning. Today, the company has saved millions of dollars, reduced time-to-answer by days and increased its analytics output by a factor of ten while also improving quality.
Silos Are for Feed, Not Data
Since 1949, Simmons Foods has grown and processed top-quality poultry for its customers. Since those early days, the company has expanded into other segments but its main focus remains on “the protein business,” as it’s called within the industry.
Like many other organizations within the agriculture and manufacturing industries, capturing potential savings is deeply important.
“The margins are very, very thin. And it's a tough competitive environment,” says Dan Boyce, Director of Demand Planning for Simmons. “What we do is fairly simple at the heart of it: we turn corn into chicken protein. And the more efficiently we can do that, the better we can make money.”
While this may sound simple, the reality of making a profit in this industry is complex. Even something seemingly unrelated, such as gasoline regulations requiring ethanol blending in summer fuel blends, can affect the business—because ethanol is made from corn.
“When production of summer fuel blends comes on, our feed prices go up,” Boyce explains. “So how do we model all of this in such a way that we can understand when and how best to start buying and storing and blending our feed to get the absolute best value out of that calorie of corn and turn it into an ounce of chicken in the most efficient way possible?”
As the head of Demand Planning, Boyce is responsible identifying areas where the company could reap additional value. Part of the problem his team faced was simply accessing the data.
While Simmons invested in Oracle E-Business Suites for its back-end data, Boyce’s team still found itself chasing reports and data housed in disparate data sets—typically Excel spreadsheets maintained by multiple groups. Occasionally data sets would exceed 100 million rows, but usually the sources were smaller.
“Those data sets were spread all across the company internally as well owned by a lot of our external data partners,” he says. “The environment was very siloed—very old school.”
Boyce notes that this can be typical within the agricultural and manufacturing industries. “In traditional agricultural businesses, people don’t always think from a data perspective. They don't think about how their data is structured and their consistency and building it in such a way that it can be used effectively.”
Boyce estimates that six people—his four-person team plus two other resources—spent between 30 and 50 percent of their time simply completing “data grunt work” as he calls it. “We would get data sets and run reports and tweak the results to get ready to start the analytics process.”
“It Was Just That Difficult”
The Demand Planning team was approached by the Poultry segment with a request: provide more analytics rather than simply straight reporting. In order to do so effectively, the team needed to overcome the results of years of “data mavericks” across the company.
“When people fend for themselves, from a data perspective, everyone is going to create reports and views and publication schedules that meet their group's needs,” Boyce says. “As soon as you go from group to group, none of those reports or productions or schedules of publication are compatible with each other—even though they might be using the same base data.”
Often, they would be asked to work with these “maverick” reports that had been set up months or years prior to answer different questions. “We would try to make integrated views from these reports and it just wasn’t sustainable from a maintenance perspective,” he says.
Another challenge to the team was the fact that their primary analytics tool was Microsoft Excel. Or, as Boyce puts it, “Excel and a jackhammer—we pretty much used Excel and brute force.”
The entire process was so difficult that the team was able to complete a planning cycle only once each month. “We used to crunch through all of our data once a month, because it was just that difficult and that many people were involved in the process.”
Answering ad hoc questions was equally difficult. Boyce estimates that these sorts of requests could take days to close. And if a similar question came up a few months later in a slightly different format? The team would have to start all over again.
“We’re Not Willing to Drop $100,000 Just to See if Something Will Work”
Boyce set out to find an analytics solution that would be compatible with the Oracle back-end solution.
“I just went out and began with internet searches and phone calls to my network,” says Boyce. “I came up with a list of ten or so options.”
Boyce spent time reading message boards, downloading free product demos where possible, and continuing to research. Eventually, he narrowed his choice down to three options—Oracle Business Intelligence Enterprise Edition (OBIEE), QlikView, and Tableau.
“Many enterprise solutions are expensive and require extensive implementation and ongoing systems support,” says Boyce. “So I questioned, ‘Is this cost-effective for us?’”
Part of Boyce’s concern with enterprise solutions’ price tags was the initial investment required to run a proof of concept.
“We're just not willing to drop a few hundred thousand dollars just to see if something will work for us,” he says. “We'll commit to a long-term spend. But I have to prove it works first, internally. And your enterprise-level products just don't allow that (for little or no cost)—not with a company our size.”
He decided against QlikView from QlikTech because he found more positive online comments and grassroots support for Tableau.
Another important decision factor in Tableau’s favor was the fact that Boyce was able to download a free demo of Tableau Desktop and build visualizations using his company’s data.
“It's a different conversation internally to say, ‘I've tested this and it works,’" Boyce says.
“We Said, ‘We’re Going Back to Excel;’ They Approved the Purchase the Next Day”
During the 30-day free demo period, Boyce built visualizations using a core set of data, extracted the workbooks and then published them in Google Docs for internal Poultry segment users who accessed the workbooks using the free Tableau Reader.
“People realized that they could get so much more information out the Tableau version of our data. I requested a spend for two full licenses of Desktop, but just like any other corporate environment, nobody wants to spend money. The day before our trial versions were about to run out, I sent an email to the person who was supposed to approve it saying, ‘We're going back to Excel starting tomorrow.’ And then they approved the purchase the next day,” he remembers, laughing.
In order to ensure that they were getting the most out of the investment, the team took advantage of Tableau On-Demand Training videos and the Tableau Community Forum.
Boyce also attended a Tableau Fundamentals and a Tableau Advanced Classroom Training event in Atlanta. He feels he gained a lot from the training, but even more from the exposure to other Tableau users’ perspectives.
“Interacting with other companies that were involved in that in-person training was valuable,” Boyce says. “Being able to turn to the guy from a totally different type of company and say, ‘What types of things are you doing and how are you solving this problem?’—that gave me a lot of insight.”
The team continued publishing extracted workbooks using Google Drive for approximately a year.
Boyce’s team decided to use Tableau to visualize data for one large business unit at a time. "The first topic we tackled was the Demand department—which makes sense, because we’re the Demand Planning group. We gave the organization a good way to visualize how demand has changed over time, how we've reacted as a company from a financial and a production standpoint to those changes,” explains Boyce.
As management began working with Tableau visualizations, word spread and soon Boyce’s team was fielding requests for help from other departments within the company.
“After Demand, we did Inventory, and we're starting to look at Finance and Production,” says Boyce. “It turned from a push into a pull. And as soon as that happened, we went and deployed Tableau Server,” says Boyce.
Today, a core group of up to eight people author data visualizations consistently and the team is in the process of training an additional five people to produce visualizations. Simmons manages access to its data using Active Directory for authentication.
Approximately 30 members of the Simmons Poultry segment management team regularly consume Tableau visualizations, often using mobile devices such as iPads.
The Demand Planning team has obtained a few mobile devices to make sure that their visualizations are optimized for consumption on a smaller screen. “We’ve changed a few things, created another tab or two in order to simplify a visualization,” he says. “It’s a relatively easy process.”
“Once You Build It in Tableau, It’s Evergreen”
Simmons is enjoying significant returns from its Tableau investment. For example, instead of pulling and analyzing data once per month, the team now updates its data on a daily basis.
“It's an easy process to duplicate day after day using the Tableau data management process,” says Boyce. “Having a daily snapshot is just so much more effective than having a monthly one.”
He points out that because the team has pre-loaded its dashboards with the constraints, the timeline and anticipated demand, they are able to answer ad-hoc questions far faster.
“When somebody asks a ‘what if’ question, it's literally a 10- or 15-minute exercise to answer the question rather than trying to spend days re-creating data that's unsustainable,” says Boyce.
The team is saving even more time with Tableau’s ability to automatically update existing reports with new data.
“Once you build it in Tableau, it's evergreen, as long as that data's being kept up,” says Boyce. “I always say, ‘All you have to do is hit F5 and wait for a couple seconds and there's your updated answer to whatever question you had three months ago.’”
This frees up time to increase both the depth and the quantity of analytics that the team produces.
“Before, we could just do the bare minimum with our time, and now we've got the flexibility and the creative time to really do some deeper analysis,” Boyce says. “I’d say we've improved our output by a factor of ten.”