Gamefly saves 100s of hours with Cloudera Impala & Hadoop

GameFly is the number one video game subscription service in the US. Members can choose from more than 8,000 titles for console or handheld gaming systems or select the latest movies available for BluRay and DVD.

The analytics team knew they could mine true business value from GameFly’s big data—if only they could find more time to analyze it.

After implementing a “magic solution” of Tableau Software and Cloudera Impala on a Hadoop data warehouse—the team is saving more than 100 hours each week and accelerating insight by weeks. Even better, the team’s near real-time, multichannel insight into customer behavior has led to strategic insights leading to improvements including a 5x increase in new membership trial participation.

“A pretty big time sink”

After a company-wide reorganization in February 2013, the GameFly analytics team found itself inundated with new reporting responsibilities.

“After the reorganization, our team became responsible for both analytics and reporting—we inherited all of the reporting from the rest of the company,” says Andy Calver, Director of Business Analytics for GameFly.

Calver estimates that the team spent 80 percent of its time—or 128 hours each week among the four-person team— simply running reports for users.
Most of those hours were spent on the mechanics of gathering data, with very little time left for analysis.

“We probably spent five percent of our time looking at the data before we had to send it off and deal with the next request,” he says.

The team began to investigate how to provide GameFly users with real-time insight into the company’s large volume of data. They envisioned a solution that also would allow them to spend more time analyzing the data for business advantages, rather than simply running queries and delivering reports.

They decided that their answer would be a data warehouse that would function as the company’s business intelligence repository. On top of that, they envisioned a data visualization tool that would connect directly to the data warehouse and provide fast, intuitive data visualizations directly to business users.

"It worked out of the box"

The analytics team decided on Cloudera distribution of Hadoop (CDH) for its data warehouse. The next step was to find the business intelligence solution to sit on top.

The team considered Tableau 7.0 but assumed it would not work with Hadoop. They were deep into testing other solutions when they gave Tableau a second look.

“We were working with another vendor and kept waiting for them to come up with a solution to connect to Hadoop,” says Calver. “On a whim, I went back to the website and found the new release—Tableau 8.0.”

This release included a number of new native connectors, including Cloudera Impala, Apache Hadoop, Hortonworks, and more. The team downloaded a free trial of Tableau Desktop.

“It worked right out of the box,” says Tom Jankowski, Software Engineer for Data Analytics at GameFly. He recalls that a technical question during the trial period turned into a positive experience for the team.

“We submitted a technical support ticket on the second day of our two-week trial period. And we had our answer within 24 hours. We thought, ‘Well, these guys are obviously going to support us going forward.’ That was a big, big part of our decision to choose Tableau,” says Jankowski.

The team watched free training videos from the website to learn how to use Tableau.
“It's really easy to get started. There were so many resources—you can learn the whole program within a week,” says Rich Ung, Analyst for Gamefly.

Today, all four members of the analytics team are authoring visualizations using Tableau Desktop. They use Tableau Server to publish their interactive data visualizations. The group uses both data extracts and live data connections for its visualizations.

“We are using Tableau with Hadoop via Impala to analyze hundreds of millions of records in a matter of seconds. It’s so fast,” says Calver.

“Using Tableau to connect Impala has allowed us to have live connections on big data sources. It’s cutting-edge,” Jankowski adds.

“That's the direction we believe everything is going to go. I would consider us early adopters of that kind of architecture,” says Calver.

“We’re looking at data to optimize everything”

“I can't see a future that doesn’t include something like this—a visualization tool on a powerful database—that allows you to instantly figure out what's going on with your company. You don't have to wait until month-end or year-end to know how you're doing,” says Calver. “It seems like that's just an inevitable future for all business.”

The analytics team is using Tableau, Impala, and Hadoop to create a number of different data analyses for teams within the company. From spot-checking customer recommendation, investigating customer trends and preferences, to identifying channels that result in the best subscribers.

“We’re looking at our data to just generally optimize everything,” says Ung.

Teams may use vizzes to see where the company should adjust inventory— increasing when rental queues hit a certain point or liquidating surplus titles. GameFly also uses vizzes to improve individual customer-level segmentation.

“We build customer profiles so we can see what genre of games they use, their rental frequency,” says Sam Ho, Analyst, Gamefly.

“It’s part of our longer-term goal to develop a complete view of the customer. We’re working on looking at demographics from a regional point of view, things like that. That’s something we’re working towards, definitely,” says Calver. "If we can figure out how a customer uses their account, we can give them a better overall experience."

An Unanticipated Win

“We are always digging into the data to determine how effective things are,” he says. “We take answers to the business teams before they can bring the questions to us.”

The team recently created a visualization to understand its customers’ lifetime value. While the team was looking at this data by timeline, they noticed an anomaly.

“We noticed that there was a period of time where that customer group had a much higher total lifetime value,” says Calver. “When we dug into that, we saw that was the time in which we offered a 30-day free trial offer.”

“We were able to not only visualize the data, but—once we noticed this—we were able to dig in using Tableau and confirm our suspicions by looking at it ten different ways. And we could do that rapidly because it was a live connection to our data so we could visualize it instantly,” he explains.

The team used its visualizations to convince stakeholders to change its previous offer, a 10-day free trial for new subscribers, to the more impactful 30-day free trial offer.

The team reintroduced the 30-day offer in the beginning of October.

"It's just faster and easier"

The analytics team at GameFly is seeing a number of benefits from its adoption of Tableau, including:

• Freeing 100 hours each week. The interactive, automated reporting the team has built in Tableau has had a significant effect on how they use their time. “We were spending the bulk of our time on the reporting and not as much time on the analytics,” says Ho. “The opposite is true, now.”

  From the 128 hours the team estimated it spent on reporting tasks prior to Tableau, the time has been cut down to less than 20 hours.

• Speeding data insight by up to a week. Thanks to the Tableau workbooks built on live data connections, teams like the digital business group can see and understand data as often as the data is refreshed—rather than waiting until the team was able to run the reports.

  “They're able to get data immediately without having to ask us first,” says Calver. “The only downside is that now it's changed expectations, so they always want everything all live.”

  The team also notes that visualizing data rather than presenting it in table or spreadsheet form adds speed of understanding.

  “There is a real difference in how quickly the user can understand what’s happening with the story through visualization versus a giant table of numbers where you have to try to find the outlier. It’s just much faster and easier,” says Ung.

• Spending time on more impactful—and interesting—tasks. “The magic combination of Tableau and our big data Hadoop solution just make our lives simpler and better. We can actually do data analysis as opposed to just reporting!” says Calver.

  He notes that it’s far more fun to spend time “digging into something that interests you” as opposed to simply running data queries and emailing numbers out to coworkers.

Overall, the team agrees that Tableau—coupled with the Hadoop data warehouse—has incited a culture change across GameFly.

“It’s changing the way that people absorb data,” Calver says. “Instead of coming up with one or two big things to ask the analytics team, they go in and look at all of the little things that they need to know about.”

Learn more about how Tableau and Hadoop can work together

Want to learn more about how Tableau and Hadoop can work together? To find more case studies, user stories, news and info, visit our Hadoop resource page.