Streaming the Data

By Brendan B. Read, Senior Contributing Editor  |  September 01, 2010

This article originally appeared in the Sept. 2010 issue of Customer Interaction Solutions

Contact centers act as bidirectional gates on the vast streams of data between customers and enterprises. Handling and managing the flow is becoming more challenging as the volume is increasing while the contents that must be sifted and analyzed become more complex.

Customers and organizations are now reaching out to each other across many different channels that must be carefully managed to provide a consistent and seamless experience. The data contained in these interactions provide invaluable service, problems, performance and product knowledge and insights.

There are also new and richer data sources to enable effective marketing campaigns via contact centers both inbound and outbound. For example Harte-Hanks’s Ci Technology Database (CiTDB) has been expanded to provides insight into technology installations, buying potential, and business demographics for more than four million locations (up from 500,000) and identifies over five million (up from 1.2 million) U.S. and Canadian IT decisionmakers. Harte-Hanks plans to further expand the CiTDB to include cloud computing, mobility and green IT in the near future and to include more businesses in Europe, Latin America and Asia Pacific in 2011.

“Contact centers are being asked to drive more revenue for the business,” says Jay Bourland, group technology officer of Pitney Bowes Business Insight (PBBI). “Instead of simply working from ‘one-size-fits-all’ scripts, they are being asked to execute ‘best next action’ steps. These enhanced interactions require a deeper knowledge of the customer in order to provide a personalized experienced tailored to that individual.”

Analytics are KeySuppliers are successfully meeting increasing, and increasingly sophisticated data processing needs with faster generations of hardware and software.

For example, Teradata (News - Alert) has boosted data warehousing performance by 65 percent in its Teradata 13 database on an Active Enterprise Data Warehouse 5600 series over the previous hardware generation shipping a year earlier. Data throughput has improved by 30 percent. Backup, archive and recovery are four times faster. Teradata has also come out with the Data Warehouse Appliance 2580, which delivers up to two times the performance of the previous version, the 2555.

Instead the core matter is determining what does this information mean and that requires analytics tools. “So you collect all the bits, so what?” Dave Schrader, director of strategy and marketing, Active Enterprise Intelligence, Teradata points out. “You need to know what it means so you know what you need to do, including nothing.” 

Schrader says analytics is used to better predict why customers are calling and what information is needed to service their requests.  If companies looked at their data, they would find that 85 percent of the time, companies can predict why a customer is calling. And with 95 percent accuracy they can guess the top three reasons.

Analytics is also being deployed to effectively handle customer complaints by applying complete context information across all channels on the screen, says Schrader. For example if a high-value on time-paying customer bought a product that was damaged upon inspection and that person is calling because the credit did not make it on this month’s bill, that customer would be given a discount on the next purchase of 35 percent. All of this needs to be on one screen so it’s “one and done.”

One critical data issue is pulling together and analyzing information from across channels to deliver this single customer view in a usable and efficient manner to the contact center agent. This is a significant hurdle to many organizations, reports Tim Shaw, chief technology officer, Portrait Software (now owned by PBBI parent Pitney Bowes). He says this matter is a top 5 challenge for 60 percent facing B2C according to a Forrester (News - Alert) report, “Craft Your CRM Investment Plans In Light Of Technology Adoption Patterns”.

The solution is applying cross-channel analytics that Shaw says “allows an organization to understand the full customer journey, behavior predictors, improve its segmentation and targeting, and identify process bottlenecks and customer drop-out points.”

Social Channel Data

The third and newest use of analytics is on data representing social media interactions. In this case, companies need to learn to “listen” proactively. Says Teradata’s Schrader: “this is often done by marketing department. New analytics can help measure sentiment by analyzing texts like Tweets or mentions of the company – good or bad – on blog sites.”

There are a lot of comments to analyze, the Teradata official points out. Twenty percent of Tweets are about brands or products, 34 percent of bloggers post opinions about products on the 200 million blog sites out there. Nor should they be ignored; according to a Forrester report, “Tapping the Entire Online Peer Influence Pyramid” 50 percent of people trust consumer reviews, 42 percent trust blog writeups on products from people, they know, and 39 percent get their recommendations from social media sites. In contrast, 32 percent that believe TV ads and only 11 percent believe information in a blog written by a company.

One challenge is how to extract intelligence out of this information because social media data is unstructured. Emerging in response says Schrader are tools including buzz analytics, emotion detection and sentiment analysis tools enable organizations to determine the significance and priority of social media comments. For example CrimsonHexagon monitors buzz and Attensity and SAS (News - Alert) that perform tasks like natural language processing to extract keywords and tone, resulting in sentiment analytics.

Meanwhile there are companies like Touchgraph provide solutions to visually see the connectivity graphs of who is interacting with whom, leading to who is influencing who. 

With these solutions, as well as access to data providers like BlueKai, Exelate, or TARGUSinfo (News - Alert) who aggregate information from websites, “companies can build databases to monitor customer behavior in the aggregate, or where identification information is available, down to individual customers,” says Schrader.

With social media some activities are noteworthy while others are not. For those that are database triggers can be used to alert either the contact centers or the marketing departments, depending where the feeds are landing.  

“Social media data is on a par with web clicks because there is a steady stream of data coming from multiple servers that must be integrated so that contact centers and other users can see and understand what is being said and what is happening, so they can formulate the right company reactions,” explains Schrader. 

Taking Data to the Cloud?

Cloud/hosting is becoming a popular option to premises-based servers for applications. Is it practical though for handling, storing and warehousing data used by them?

“The cloud was engineered for highly concurrent, simultaneous transactions of the form that are typical in a contact center,” says Bourland. “The cloud may be an issue for large batch processes that occur in the back office because the cost of moving large volumes of data is high.”

Clouds are ideal, reports Schrader for low-volume experimental applications such as social media market research and offer testing. Teradata recently partnered with (News - Alert) to launch Teradata Express, which provides for free a 1 terabyte system via the public cloud.“When you have low latency and bandwidth requirements i.e. you are not pushing a lot of data and you do not require it in real time – such as social media analytics – the public cloud is good for that purpose as it saves money and is highly scalable,” says Schrader.

Cleaning, Enriching the Data

For data to be of value it must be accurate: Up-to-date names, phone numbers, e-mail and other contact information. This can be very challenging as customers use an increasingly widening array of channels.

One of the big hurdles is the shift by consumers and even some businesses from landlines to wireless devices, whose numbers are not publicly available. That can make verifying mobile number accuracy from Web forms, agent record entries, lists and in databases difficult.

In response TARGUSinfo has significantly improved its On-Demand Lead Verification and On-Demand Lead Scoring solutions by increasing the authoritative linkage to “right party” contactable leads by 30 percent. It has re-architected and improved linkages at the identity layer (i.e., name, address, and phone) combined with the addition of mobile phone coverage that permits firms to check against TARGUSinfo’s data repository of more than 219 million mobile numbers. The information housed is extremely current; it is refreshed ten times a day.

“These enhanced interactions require a deeper knowledge of the customer in order to provide a personalized experienced tailored to the individual,” says PBBI’s Bourland. “This uses sophisticated analytics based on a rich supply of data. The data must be clean – the old adage of “Garbage In, Garbage Out” is really multiplied in these settings. So users are responding to larger quantities of data with higher requirements on the fitness for purpose.”


The following companies participated in the preparation of this article:

Harte Hanks

Pitney Bowes Business Insight



Brendan B. Read is TMCnet’s Senior Contributing Editor. To read more of Brendan’s articles, please visit his columnist page.

Edited by Stefania Viscusi