Despite a veritable army of visitors -- more than 20 million people worldwide -- Moviepilot.com can pinpoint the Star Wars fans among its readers and then deliver specialized content about Wookies, droids and The Force.
On a smaller scale, pet insurance provider Trupanion can identify the policies that best suit the owner of a Border Collie and the ones that don't fit the medical needs of a Persian cat.
These companies and countless more mine troves of data to better understand their customers and clients so they can offer the most tailored user experiences. It's not a luxury; a large majority of businesses use customer information to gain an advantage in a competitive marketplace. If a company isn't analyzing data, it's probably steps behind the competition.
It wasn't always like this: With limited personalized consumer feedback, companies for decades typically had to use a one-size-fits-all approach to interact with customers. But today, with customers willingly submitting the most personal information imaginable about their desires, experiences and problems, companies can pull out every needle in the haystack and react accordingly.
"We have a huge amount of data, but lots of companies have tons and tons of information and don't know what to do with it," said Phil Walden, director of digital strategy at Moviepilot. "The advantage goes to those who have a clear idea of what to do with it." For Moviepilot, the "doing" happens thanks to proprietary software that culls and analyzes user information gleaned from Facebook, Instagram and Tumblr. Others rely on vendor software to do the heavy lifting and -- if they're truly invested in maximizing customer interactions -- integrate the data with their CRM systems so they can properly respond.
But no matter the technology, many companies that use data analytics for customer experience management say it will bring a business only so far. "In addition to the technology," said T.J. Houk, Trupanion's director of business intelligence, "it's the people who understand the data and the company itself to make the most of the data. Otherwise, it's just data."
May the data be with you
Walden couldn't imagine Moviepilot's business without its ability to analyze customer data. The company couldn't provide the latest stories, photos and gossip on its members' favorite movies if it didn't know which films they liked in the first place.
When Moviepilot launched four years ago, its founders knew they couldn't create an ordinary dot-com website. People wouldn't easily discover relevant movie reviews and news from Moviepilot through Google searches, but they would see stories if the content streamed through their favorite social channels and sparked engagement. Thus, when users want to share stories or comments, follow a favorite writer or receive updates on a specific movie, they must sign into the Moviepilot website through their Facebook accounts.
For users, allowing access to their accounts means they get select content about their favorite movies and actors in their Facebook streams. For Moviepilot, it means a treasure trove of user demographics, locations and other personal information at its disposal. With the aid of proprietary software, Moviepilot sorts all this data -- indexing selected interests against the movies that similar users like -- and determines what kinds of information a user should see next.
"We find what our users are interested in and give them what they like," Walden said. "Internally, we use the data to influence our content strategy. So, say, if they like X-Men, they'll probably like certain movies and TV shows from that action genre. Eventually, the data helps us learn why they're interested in a certain thing, and that [has] helped our growth."
What Walden means is that by making sense of each user's information and interests, Moviepilot closes the gap between the data it collects and the data it creates. That created data tells Moviepilot that, for instance, certain 20-year-olds in San Francisco prefer Supergirl, while 30-year-olds in the Midwest favor Wonder Woman. So the content that Moviepilot will push onto a user's personalized website or her Facebook feed will reflect what she read in the past, and will also consider the stories her Facebook friends are reading.
Before personalized data laid bare customers' interests, a business would have sent a survey group into the field and would have been lucky to accumulate 100 honest and complete responses, Walden said. Now, personalized data from Moviepilot's users' Facebook pages -- as well as information gleaned from their Twitter and Instagram accounts -- provides thousands of detailed answers, he said, pointing to the nearly 30,000 data points already collected from Avengers fans who are awaiting a sequel that's still a year away.
The personalized data not only satisfies more than 20 million users' individualized movie interests and strengthens Moviepilot's position as a resource -- it also allows the company to give moviemakers the type of marketing data they crave. Moviepilot works with 21st Century Fox, Fox Searchlight, Hallmark Channel and other studios, helping them devise targeted social and marketing strategies. Armed with thousands of keywords used by its members -- that have been separated by gender, age and other factors -- Moviepilot determines which search terms work best to identify where and how to place a studio's advertisement on the Web.
"It's information the studios don't have, so we can say, 'Your audience for this movie is older males,'" Walden said. "Before, everything was done on a gut feeling, and movies that were made for young men missed their mark. The data proves and disproves those ideas."
Patient care gets a boost
But analytics can do more than determine people's movie preferences; it can help save lives. Analyzing real-time patient information such as vital signs and medication use lets healthcare providers extend the best treatment and enhance, inarguably, the most important customer experience of all.
The international auditing company KPMG uses data analytics to measure treatment patterns, costs and medical outcomes to answer the eternal question of how to best serve a patient. KPMG pulls data from Medicare and Blue Cross Blue Shield claims to reach conclusions on how much a particular treatment should cost and how providers can improve patient care.
“Especially now, with the regulations coming up with the Affordable Care Act, you can use data analytics and actually follow through with quality of care,” said Bharat Rao, an advisory principal at KPMG. “Providers can ask, ‘Am I reducing the number of avoidable errors? Am I following guidelines?’ It’s harnessing the data of day-to-day treatment to improve healthcare.”
By analyzing patient data, KPMG aims to help the healthcare industry reduce variability. In other words, the data will reveal that medication worked better for a group of patients on the West Coast than surgery did for another group on the East Coast, according to Ashraf Shehata, principal of KPMG’s Global Healthcare Center of Excellence.
“When you see that variation, the key is to address it,” Shehata said. “So you cannot lock the data in the analysts’ hands, but you have to put it in the hands of the providers. That’s why you're seeing an improved relationship between the payers and the providers.”
Clear, personalized data, in other words, has started to knock down silos in the healthcare industry and has led to new procedures and practices that improve patient care.
“There’s a whole movement toward the transparency of data,” Shehata said. “The data shows what’s behind out-of-pocket costs or the time it took to get an appointment. The comments and ratings that make Yelp successful are making their way to healthcare products and services. Patients are demanding it.”
While KPMG reviews personal data on a large scale, many hospitals struggle with analyzing smaller data sets. “A lot of hospitals are in a position where they collect data across multiple IT systems and multiple archives, but they can’t reposition it,” said Christopher Rosin, chief scientist at Amara Health Analytics in San Diego, which crunches patient data through a platform from DataStax Enterprise to catch the early stages of medical conditions such as sepsis.
“The data we get is in a very rare form. Hospitals have only recently reached a point where they have a critical mass of patient data,” Rosin said. “Data sources from the same hospital may steer to the patient in different ways, so we need to standardize it. A big part of it is natural language processing and the ability to extract things like diagnosis of a disease or treatment.”
Although healthcare providers can independently crunch all the data they acquire from patients, privacy expectations — as defined by the Health Insurance Portability and Accountability Act — limit the type of patient information that providers can share with third-party sources, including those that perform analytical services. Hospitals, clinics and insurance companies can’t release patients’ conditions, what medical services they received, what medication they took or other identifying information.
That doesn’t go for aggregated data, which can be shared because it doesn’t trace to a particular person’s medical condition.
“The data is not personally identifiable, but it is the foundation for academic studies and population health management,” Shehata said. “The aggregated data is used for developing medical protocols and resource planning. Employers can use this data to see what sort of healthcare spending they have per employee and benchmark it against other organizations. Much of the work on big data in healthcare is centered on aggregated data.”
Still, relying largely on aggregated data limits the healthcare industry from seeing the full picture of patient care. “The protection of private patient data is a very significant, and perhaps inherent, barrier to the free flow of data for healthcare research and the development of new insights that could improve patient care,” Rosin said. “To avoid disclosing private protected patient health information, patient data cannot flow freely unless it is totally de-identified. But in the case of full rich data, such as narrative notes by physicians and nurses, de-identification risks corrupting the underlying meaning; and full de-identification is very hard and it is difficult to prevent re-identification.”
In other words, academic researchers are precluded from using large-scale comprehensive data to develop new insights on patient care, Rosin said. “So we’re a long way from fully realizing the kinds of healthcare insights we might get with deep analysis of the full richness of data from hundreds of millions of patients,” he said.
The sea of data
Of course, approaches to harvesting data vary among industries and businesses. For example, healthcare data analysts, according to Shehata and Rosin, have to mine information from a multitude of sources: hospital records, insurance information and retail records of healthcare purchases.
Trupanion, the medical insurance provider for cats and dogs, employs actuaries and veterinarians to help interpret crunched industry data that considers customer price preferences to determine cost, BI director Houk said. Using a business analytics platform from Sisense, the company also analyzes customer data from its website and call center recordings.
While it’s easy to assume that data can’t be wrong, it can. “It’s important to ensure you’re dealing with accurate data before making use of analysis,” said Amara Health Analytics’ Rosin. “It’s easy to run a query and think you see a trend, but 90% of a row is missing. So you don’t actually know anything yet. You have to fully understand your data, so make sure you have clean data.”
Once customer data has been analyzed, businesses should be mindful of acting too soon or too late on the information. “Some people get the first five data points and go pull a switch, while others wait until they’re comfortable and miss an opportunity,” Houk said. “Take small, valuable steps to find the right course. If you see a customer tendency on a website, but you’re not sure the data is right … make small tweaks that are correctable instead of waiting to make a big dramatic change that may also be wrong.”
Data sources that seamlessly connect will improve the odds of understanding what customer information is saying. That’s why it’s crucial for businesses to always ensure that they’re analyzing data that is uniform and not disparate, Shehata said.
“Rather than focusing on the implementation of a particular system, think about how it’s implemented in a department and how the data there will integrate with other data sources. … For customer systems, you can use CRM tools such as Salesforce.com or Siebel, but it’s still not about the tool. It’s about how you predict the issue and how the data will be used,” Shehata said.
Jennifer Folsom is the senior manager of branding for Summit Consulting. The company has a “tremendous amount” of data gathering and analytical tools to review customer data — including a HubSpot inbound marketing tool that moves many different data sources into a CRM platform “that makes it very easy for non-quantitative types to create a personalized experience,” she said.
But Folsom said all the tools in the world won’t, in turn, provide a personalized customer experience “if you don’t sell a great product and have a great sales team. The data is there to help to advance to a point.”
Of course, it’s easy for a business to get lost in all the information it collects. And all that data would be useless without employees knowing precisely what they’re looking for and which key performance indicators should serve as guideposts in assigning value to customer data.
Houk recommends that businesses employ people who can make informed decisions with customer data and not just present colleagues with “big stacks of numbers.”
“Oftentimes, it’s as simple as just getting the information out there,” he said. “The risk is some people don’t have the same experience with the data as others do, but that’s why you should have data experts to help. You can remind those who aren’t experts that they can get help.”
Sales analytics a sweet spot for BI