top of page

How Data Analytics Can Boost Any Organization’s Performance

Data analytics isn’t just the realm of high-tech industries any more. It’s making inroads into a wide variety of organizations, from financial firms to sports teams to medical institutions. One sign of the times is that New York City recently passed a bill to make the Mayor’s Office of Data Analytics a permanent part of city government, according to StateScoop.

It was against this backdrop that the Wharton Customer Analytics Initiative (WCAI) held its second annual Analytics Accelerator Challenge. Companies of all stripes submitted their marketing analytics problems for a chance to have a Wharton/University of Pennsylvania student team come up with solutions. At the end, the chosen organizations are presented with data analytics coding, models, and tools they can apply going forward.

This year, 30 firms applied and four were selected: The Barnes Foundation, a museum; Fuel Cycle, a market research firm; Hachette Book Group, a publishing company; and Reed Smith, a law firm. Of the 200 students who applied, 24 were chosen. They ranged from undergraduates to MBAs to post-docs, and hailed not only from Wharton but other disciplines at the University of Pennsylvania such as engineering and computational biology. Teams were given only four weeks to meet their goals.

Barnes: Using Analytics to Boost Traffic

An art museum famous for its French impressionist works, the Barnes Foundation is nearly 100 years old. It underwent a transformation six years ago when it relocated from its original home in suburban Pennsylvania to the heart of Philadelphia’s arts and culture district, the Benjamin Franklin Parkway.

“We’ve been very successful in our first six years on the Parkway,” said Peg Zminda, who wears several hats as executive vice president, CFO and COO. She explained that the museum has welcomed more than 1.7 million visitors — about triple the number it had at its original location in suburban Merion, Pa. — and has signed up 16,000 member households compared to about 300 previously.

“So you’re probably saying, gee, this all sounds really good, what’s the problem?” Zminda notes. It turns out that since the move, attendance has not been consistently strong as expected. And unlike some nonprofits with large endowments, the Barnes is relatively under-capitalized, so earned revenue is important and predicting visitation is crucial.

“There’s huge power in segmentation … the real action is under the hood, once you start looking at different demographics.” –Raghu Iyengar

Tackling the Barnes’ problem was the student team headed by Anuj Gupta and Catherine Bache. They decided to create a demand forecasting model. The team tested three drivers of attendance: Actions controlled by the Barnes such as pricing and special events or exhibits; competitive actions such as peer pricing and competitor institution attendance; and macro factors such as seasonality and local and national jobless rates.

A significant finding was that if the Barnes continues on its current course, it could suffer an attendance decline of about 10% a year. To arrest the decline, the museum could implement modest reductions in price that would stabilize non-member attendance while remaining revenue-neutral. Specifically, if the Barnes were to reduce its average ticket price after discounts by 8.5% (dropping the price of adult admission from $25 to $23, for example), attendance would stay stable over the following year.

The team said their demand forecasting model could be embedded in the foundation’s current reporting templates with little extra effort to use and update. Gupta and Bache handled “a humongous amount of data” in a short four-week period, yet created a model that was “implementable and simple, according to Wharton marketing professor Raghu Iyengar, a co-director of the Wharton Customer Analytics Initiative. He pointed out that data analytics can help non-profit organizations like Barnes. “Many of us think about analytics for social good. You see this in action [here],” he said.

Fuel Cycle: Helping A Client’s Business

The next company was Fuel Cycle, a Los Angeles-based market research firm. Research director Emily Burton said the company “powers some of the most customer-centric brands” — Google, Hulu, Facebook, AIG, and others. The firm asked the student team to help it understand customer behavior at Rent-A-Center (RAC), one of its clients. Fuel Cycle supplied customer data from RAC, which rents out consumer electronics, appliances, furniture and other items under flexible rental and purchase agreements.

The team used various statistical tools to investigate two main areas. One was customers’ rental rate preferences compared to the rates they actually received. Not surprisingly, most customers surveyed said they preferred lower rates. However, when the team looked at the data by segment it saw “some really interesting patterns,” student presenter Kelly Herring said. For instance, they found that older customers tended to be more amenable than younger ones to more expensive agreements.

Herring said RAC could use this information to refine its sales strategies. It could push sales of more expensive, higher-end goods to older consumers and offer lower rates to younger individuals in order to capture a greater volume of that demographic.

The team also looked into predicting when customers were likely to default — that is, fail to make payments or return the item — a major concern in the rental business. New customers were found to have higher default rates than existing or returning customers. The product itself was a factor too: Rental agreements for TVs were associated with high default rates, for example.

Customers’ income brackets were also examined. Lower-income brackets reflected a higher default rate, which makes intuitive sense because financially strapped individuals might have trouble fulfilling the contract, the team said. Yet they found some higher-income brackets with significant default rates as well.

Herring speculated that these customers might be relocating frequently, changing jobs often, or have multiple residences, which could hamper RAC’s ability to track them down. These are insights that could be applied to RAC’s real-world decision-making, which came to light through a deeper dive into the data.

“There’s huge power in segmentation,” Iyengar said. Some of the results might seem obvious at first, but “the real action is under the hood, once you start looking at different demographics.”

Hachette: Tracking Marketing Effectiveness

Stephen Colbert, Tina Fey, David Sedaris, Malcolm Gladwell, and Neil Patrick Harris are some famous authors published by New York-based Hachette Book Group. But just having authors with star power isn’t enough to sell books, according to Ryan Pugatch, its vice president of strategic technology. He said that Hachette aims to become a more data-driven enterprise.

Traditionally, “publishing has been an industry driven by gut feel,” Pugatch said. As such, Hachette’s marketers are “on a treadmill,” fulfilling a checklist of promotional campaigns or responding to the perceived expectations of the author or agent, without adequate data about which efforts are yielding the best results or the most value.

One wrinkle in tracking information is that the company’s sales aren’t direct-to-consumer but through retailers like Amazon and Barnes & Noble. As a result, “we lose the customer journey a little bit,” Pugatch said. Hachette can run a marketing campaign to drive potential customers to its website, “but then they’ll move on to Amazon and we don’t know if they actually go through with the purchase.” He also noted that although the publisher sets the prices of its e-books, the printed book prices are set by the retailers.

Analyzing 250 book titles, the students developed a model to help Hachette track sales indirectly. The variables included book characteristics (blockbuster vs. non-blockbuster; top five genre groups or not); the format (electronic, hardcover, others); the sales forecast; time events (number of weeks since publication; publication season); and marketing levers, including campaigns mounted pre- and post-release.

“Do books just sell themselves, like get [bestselling author] Elin Hilderbrand to write a book? … No. You need marketing even for blockbuster authors.” –Eric Bradlow

The team also performed what student presenter Huy Le called “a deep dive into the relationship between marketing and [web] traffic.” They zeroed in on Hachette’s Mail Chimp campaigns, tracking the number and cost of emails sent and analyzing the effects. They found that running these campaigns had a positive effect on web traffic, and that the more dollars spent, the greater the response. Interestingly, although cost had a noticeable impact, the number of emails sent did not.

Wharton marketing professor Eric Bradlow, who co-directs the Wharton Customer Analytics Initiative, said the team found a strong positive connection between pre-sale campaigns and books by blockbuster authors. “That to me was fascinating,” he said. “Do books just sell themselves, like get [bestselling romance author] Elin Hilderbrand to write a book? … No. You need marketing even for blockbuster authors.”

Reed Smith: Boosting Law Practice Efficiency

Listed among the Global 20, Reed Smith is one of the world’s largest law firms and it wants to use data analytics to improve efficiency. However, Jack Nelson, the firm’s chief of legal operations, whose team is responsible for technology strategy, jokingly pointed out that the analytics job is a bit of a “misnomer.” Not only Reed Smith but “the entire legal industry is woefully behind the curve in data analytics,” he said.

The law firm’s task for the student team was to analyze and derive insights from millions of billable time entries generated by its lawyers. Reed Smith wanted the team to figure out ways to productively organize the engagements and identify any actionable business trends. Using analytics and machine learning, the team worked with the anonymized time entries — some of which contained only partial information — to find key patterns.

Student presenter Yifan Chen said the team sought to answer questions such as, “What types of cases are bringing in the most revenue? Take the longest? Take the shortest [time]? What types of industries are they in?” They explored data categories such as industry, practice group, type of tags, activity, task, the kind of work, geography, and fees.

Some examples of findings were that the tags “Public Company” and “Divestiture” were linked with the greatest average work amounts — or how much time in total was spent on the case. The longest average work hours — how many consecutive hours were spent on the case by lawyers in one sitting — were tied to matters tagged “FCA – Criminal Fraud.” When analyzing by industry, the team found that the “Hotels and Resorts” and “Private Equity” industries incurred the greatest average work amounts.

Reed Smith could use information like this to make predictions about new cases and improve future decision-making, said Chen. For example, if the company can more accurately anticipate the time and revenue associated with a new case, it might help them determine which lawyer would be best assigned to it.

WCAI’s Bradlow said the model the students developed will be incorporated into Reed Smith’s data operations. Iyengar added that although some people still think of data analytics as “all about complex machinery,” it is ultimately practical in that it helps companies with decision-making. It can help companies decide, he said, “what kinds of methods and models working together can give you the biggest bang for your buck.”

bottom of page