Customer analytics techniques boost sales and service strategies

adam121 - Fotolia

Companies still struggle to unlock customer data analytics insight

Companies today know they may relinquish their competitive edge without a customer data analytics strategy.

Just a year ago, Miner Corp.'s contact center relied on manual processes and human discretion to send technicians on customer-service visits.

But that setup was rife with problems associated with human-driven processes. Dispatchers would assign a technician to a customer-service issue, but that field service technician might not have the right skills or be too far from a job site to meet the service-level agreement (SLA) requirements to arrive on time. For emergency requests, manual dispatching of technicians was a time sink.

"If you have a two-hour emergency job, you can't wait 20 minutes to find out a certain technician or vendor can't do it," said Mirza Chughtai, CIO at Miner, based in San Antonio. "You need to be able to make that decision and without burning a lot of your SLA."

So, Miner, which provides loading-dock services, including forklifts, waste management products and other services, turned to Service Wave Analytics, an application that embeds customer data analytics directly into the Salesforce Service Cloud. Agents access Service Cloud to manage all customer data and service issues. Salesforce, the leading cloud-based CRM vendor, released Service Cloud Wave in 2015, and it makes native analytics dashboards in other applications, such as Service Cloud. With the data on technicians, routes and service requests readily available in Wave, technicians can now be assigned to customer-service issues through an automated queue that accounts for technician skills, availability, proximity to the site and other variables that enable techs to service customers more efficiently.

Chughtai said, with Service Cloud Wave, Miner has been able to save, on average, about 15 minutes per service request. With about 300 daily service requests, that amounts to significant time savings. For Miner, using software removes the human element from technician assignments, which means it can honor its SLAs by providing efficient customer service and knowledgeable technicians.

"Today, we don't rely on tribal knowledge," Chughtai said. "It's all systematized."

Customer data analytics moves to the must-have realm

This kind of automation and intelligent workflow is critical for companies that want to differentiate themselves from the competition by providing superior customer service and customer experience (CX). For companies like Miner, providing on-time, efficient, knowledgeable service requires intelligence.

As a result, data analytics is no longer a nice-to-have; rather, it's a requirement for understanding how customers use products and services, capitalizing on their known preferences and determining how prospect behavior can be converted into sales opportunities. At the same time, experts said companies are struggling to use data analytics intelligently. Given the massive influx of data associated with serving customers, companies can get easily overwhelmed by these flows of unfiltered data -- especially in real time.

Companies are also struggling to stitch together data inflows from multiple communication channels and devices -- such as customer browsing activities on the web, email and live chat conversations, and social media comments -- and get a 360-degree view of the customer. According to Frost & Sullivan's report, "Moving from Multi-channel to Omni-channel Customer Engagement," while 53% of more than 300 enterprises used traditional channels, such as phone and email, in 2014, newer channels -- mobile, social, web, chat and video -- have come to predominate for 54% of companies in 2016.

But companies are finding it difficult to integrate that omnichannel customer data. A Forrester Research 2014 survey, Customer Desires vs. Retailer Capabilities: Minding the Omnichannel Commerce Gap, revealed 40% of companies struggle to integrate back-office applications that house this data. Companies are wrestling particularly with unstructured data -- data sources generated by customer tweets on Twitter, for example.

"Channels like social media generate data that's really messy, really noisy, and it's difficult to extract meaningful or actionable insights from [it]," said Forrester Research analyst Brandon Purcell. "You need to understand what your business objective is and how the analysis maps back to it; otherwise, it's analysis for analysis sake."

That's why it has become critical to use technology to bring automation and data integration to back-end processes, which can then flow upward to customer-facing processes that enhance a customer's experience. Integrating customer data from CRM and ERP databases with unstructured social media data is an important goal for companies, but it remains difficult.

Miner, for example, has boosted the technician-to-dispatcher ratio, with the dispatcher able to handle far more requests more quickly. "In a traditional world, that ratio was 4:1 or 5:1," Chughtai said. "Now, we have been able to push that to 15:1 or 20:1, because a lot of decision making has been taken out of the dispatcher and [it has been] automated."

Instead of having dispatchers waste time deciding whom to send where, Chughtai said, "We're making him focus on customer experience. We want him to call [the customer] and make sure they understand the requirements, make sure they are accurate, rather than focus on making mechanical assignments." By moving the job of dispatch to automated software, dispatchers can focus on the nuances of customer service by communicating with customers.

The goal, Chughtai said, is to make customer data analytics native to where agents, technicians and managers are working, with no manual decision making necessary. The result is the company has information about its operations, not just data. "Every company has become data-rich, but not information-rich," Chughtai noted. "I'm trying to put the focus on relevant information."

While Chughtai said Service Cloud Wave has helped transform Miner's operations, he acknowledged issues associated with bringing different data sources into the Service Cloud Wave analytics app. Data can't be magically brought in from different data sources without some data cleaning and massaging along the way. Miner needs to do some extract, transform, load processes to make the data consistent and compatible with other applications and possible to view natively in Service Wave.

"You have to be cognizant of how you ingest data and bring it into the platform," he said. "But you can overcome [that] with good enterprise architecture."

That complicated customer journey map

Analytics is bringing sales and marketing above the ground floor of operations and helping companies dig into true customer needs. But if companies are gaining insight from customer data analytics, they're also wrestling with the limitations of their own expertise in gaining insight.

TripAdvisor hosts a travel site that culls predominantly user-generated content: The user community posts reviews of hotels and restaurants, as well as photos from travelers' sojourns. The Needham, Mass., company wants site visitors to more easily get information on destinations and hotspots, so it's constantly A/B testing information displays on the site. But the company saw opportunities to do more with machine learning and analytics. As a result, data analytics has become the core method in providing the site's users with a seamless experience. Ultimately, TripAdvisor wants to enlist customer data analytics to optimize sales.

Since TripAdvisor relies on the user community, it needs technology assistance to vet the content for quality and accuracy, and to categorize that content more efficiently. But relying solely on humans to comb content for accuracy and quality was difficult and prone to error. The job required numerous staffers to sift through the content and summarize lengthy reviews into easy-to-read snippets.

So, about three years ago, the company enlisted machine learning to evaluate its user-generated reviews for fraud and accuracy. With cognitive computing tools, TripAdvisor has reduced worker headcount by 70%, while improving speed and accuracy. "With automated tools, you need fewer people and you make the people you keep more productive," said Eric York, principal analyst at TripAdvisor.

About 18 months ago, TripAdvisor wanted to gain insight into its users' pathways on the website to not only make browsing more efficient and useful, but also to convert more user pathways on the site into sales, such as hotel bookings. But understanding how customers browse through the site and which pathway is optimum in making a sale is easier said than done, York said. To do so, the company needs to enlist data-gathering tools to analyze customer journeys. TripAdvisor wants to track and segment users based on their sequence of events using data on when a user goes to, say, X, then Y, then Z pages. By analyzing user pathways, the company hopes to better optimize sales opportunities and reduce the friction that can get in the way of closing a deal.

While TripAdvisor has myriad business intelligence tools, including Hadoop for data warehousing, Hive for database activities and Spark for machine learning, it needed to efficiently measure the user path information in real time. So, it turned to Apache's Samza technology to analyze user pathways. It also employs Tableau and R to enable business users to gain data insights.

"We want to optimize the path and gauge intent," York noted. "People who enter the site a certain way are more ready to buy than someone who enters a different way."  Even though the company believes certain pathways are more successful in converting users, the feeling still isn't backed up with data. "We know anecdotally that to be true, but we haven't done analysis to quantify it," he said.

York said TripAdvisor has been gathering data through log files for some time, but aggregating it into actionable information is another story. It's easy enough to gather singular events, but to stitch them together into a cohesive story to understand why customers buy more when they choose one path over another is more complex. Further, York said analyzing a few steps isn't problematic, but understanding user behavior over a long period of time with many steps and sequence possibilities is challenging.

"You can get, '[A user] clicked on this link,' [and], 'They went to this page," he explained. "But the challenge is, how do you pull out the relevant steps for a user that is on the website for two hours, bounced around, clicked on an SEM [search engine marketing] link, SEO link, looked at 30 hotels? How do you understand all that data? You get this massive volume of data. When you do path analysis, a lot of it needs to be calculated in real time. Calculating after the fact has too much overhead."

According to Forrester's Purcell, "There are only a few companies that are providing customer journey mapping. It's very difficult for companies to do this on their own. Even those that can pull together analytics are in a trial-and-error phase today."

Forrester Research's 2015 CX Index benchmark indicated 86% of companies want to excel in their customer-experience efforts, but only 27% of these companies are seen as good in the eyes of their customers. In the customer experience today, there's a substantial gulf between aspiration and reality. Industry analysts said it's all about the data: Intelligence hurdles stand in the way of executing a positive customer experience.

"One [hurdle] is linking a customer across multiple data sources and siloed data," Purcell said. "How do you understand it's the same customer on the website, which is one cookie, and that customer on a mobile app, which is another different identifier? That's the Holy Grail in terms of understanding the customer across multiple devices. That overall customer identity resolution is foundational for all sorts of journey analytics."

While identity management is one issue, understanding customer intent is another, which can be done only by analyzing unstructured data, such as that in text, video and other content that can't be easily categorized into the columns and rows that characterize structured data. But companies are struggling to analyze unstructured data. According to Purcell, only 27% are analyzing unstructured data as part of Forrester's voice of the customer analysis.

"Companies need to crawl and walk before they run," he said. "They need to get their structured data in order before they deal with text. But, eventually, the text is absolutely essential, and companies will need to get a handle on it."

TripAdvisor can also target users via its mobile app with geofencing; half of the company's total traffic comes from its mobile site or mobile app. Using location-based services, the company can gauge a user's location using his cellphone, then send a push notification with a restaurant menu recommendation, for example, if a user is in the area.

Combining customer data with location data is just emerging, of course, but it's due to become a major revenue generator. Recently, TripAdvisor developed a new timeline service that syncs with Facebook. The timeline functionality automatically tracks all of a user's destinations, restaurants and hotels, then provides a map that tracks all the places a user has visited. The timeline helps jog a user's memory when submitting a review, even if a user is recounting details of a visit several weeks later. The timeline feature also enables a user to share his trip and photos with the community, creating a digital album of sorts.

Next stop: Intelligent systems

By using customer data analytics, companies have made significant strides in understanding customer preferences and behavior. They've even begun to divine their next moves, largely by bringing to bear cognitive computing and artificial intelligence (AI) in their understanding of customer data. Companies are approaching this vision intelligence system by combining the forces of customer data, AI and cognitive computing.

To gain the full benefits of analytics, the data must be informative, digestible and easy to act upon. For Miner's Chughtai, arming agents, managers and technicians with data enables them to adhere to SLAs and avoid nail-biting moments. "We have taken their pressure away, giving them information, not [just] data," he said.

Forrester's Purcell said he believes the next 12 months will yield a treasure trove of information for companies trying to improve customer experience through customer data, AI and cognitive computing. But, he observed, "the companies that are customer-obsessed are going to differentiate, succeed -- otherwise, they will be disrupted, and we're seeing some movement toward that."

Next Steps

Customer insight at the center of improved customer experience

Trends to watch in customer experience

Social media and customer engagement

Dig Deeper on Customer analytics