Software Review: E-Mail Response Management

Share this article:
The speed and convenience of the Internet raises high expectations for service. But Web pages are sent automatically, while answering e-mail takes considerably more work. So companies wishing to provide quick e-mail response need specialized tools.


In many ways, e-mail response resembles telephone-based customer service. There is the same need to route messages to appropriate agents, to ensure accurate and consistent replies and to keep a record of the interaction. But there are some significant differences -- especially in the amount of information available when a message is initially processed.


With the telephone, a caller might have dialed a particular number or selected one of several broad categories. But people will tolerate only so many telephone menus, so the specific issue is not known until an agent speaks with them. In an e-mail, the question is contained in the text itself, so a clever system can not only classify the call but even choose a likely response before an agent gets involved. This allows more precise routing and faster resolution -- vendors claim they double the productivity of agents compared with manual e-mail processing.


Embedded information also opens the potential for fully automated response, delivering still greater efficiencies but adding the risk of inaccurate replies. So e-mail response systems need additional controls to ensure customers are handled properly when no one can hear them scream.


Given the huge growth in e-mail and the obvious benefits from handling it efficiently -- not to mention the bottomless pools of money available for Internet investments -- it is no surprise that many vendors offer solutions in this market. These range from simple message routing systems to comprehensive customer interaction managers. Somewhere in between are systems built specifically for e-mail response centers.


Kana CMS <B>(Kana Communications, 650-325-9850; www.kana.com)<B> provides a suite of tools to manage all aspects of e-mail response. The system is organized like a conventional telephone-based call center: messages are placed in queues and then tracked until they are resolved successfully. But it also relies heavily on categories.


These describe attributes of a message, including its contents, the type of sender, the response sent and the final resolution. Categories are linked to response templates, attachments, distribution lists, routing sequences and rules for whether the same agent should handle subsequent related messages. Administrators set up the categories and arrange them in hierarchies. They also can import existing categories that might have been set up for, say, a telephone customer service system. Several categories can be associated with a single message.


When messages are received, they can be assigned to categories either manually, through rules based on key words in the message or by an automated text analysis system based on Bayesian Networks. New messages also are given a case number that will be included in subsequent messages to indicate they are related. The rules determine whether a response is sent automatically or the message is sent to a queue. The messages are transferred from the queues to agent mailboxes as the agents become available. Agents are assigned to the queues by system administrators, who can reassign them in real time as needed. The system also can route messages directly to specific agents, based on their particular skills or on whether the agent has been assigned to handle a specific case.


The system tracks which agents are active on the system and is capable of rerouting messages for individuals who are not available. Each queue has escalation rules that automatically reroute messages if no one is logged into the queue or if the messages remain unanswered after a specified period. Escalation occurs whether or not a message has been sent to an agent mailbox or even if it was directed outside the system, perhaps to an engineer for additional research. Although Kana generates real-time performance statistics, it does not send alarms if service falls below an acceptable level. Escalation rules also can ensure that a reply to an automated response is sent to an agent, avoiding customer frustration or auto-responders talking to each other.


Agents use a specialized interface. When they open a message, they see the original text, a recommended reply, distribution list, categories and pull-down menus to add standard headers, footers, greetings and closings.


Other options let users assign new categories and attach external documents or Web addresses. Reply templates can retrieve first name, last name and a few other standard fields from the Kana database. Fields outside of Kana's standard data model would have to be accessed by custom links to external databases. The user interface exists as a standard Windows application and in a browser-based form run over the Web. Both provide the same functionality, although the Web-based version creates its replies as Microsoft Active Server Pages, which makes it easier to embed external data or functions. Kana also provides an Applications Programming Interface, which lets other systems access its user-interface functions.


Kana keeps a full history of each message including the text of the original message, subsequent replies, who worked on it, how much time they spent and how long it took to send a reply. It also lets agents attach comments that are not seen by the customer. Information is summarized in about 60 standard reports, providing views of agent productivity, service levels and category usage. Users can write custom reports using a copy of Crystal Reports bundled with the system or by accessing the underlying data tables.


The full Kana system includes an outbound mailer, which lets users import e-mail lists or select segments based on customer history, create personalized messages and send them. There is no scheduler for deferred or repeated transmission. Responses can be tracked by including a response code or links to specific Web addresses in the original message.


Kana runs on a Windows NT application server and an NT or Sun data server with Oracle or SQL Server database. A one-time license of the basic system for five users costs $39,500 while a version including outbound marketing, automated text classification and external system integration costs $99,500 for 20 concurrent users. The system was introduced in early 1998 and currently has about 45 customers. Instead of outright purchase, users can also pay a monthly fee and have Kana run the system for them.


eGain EMS <B>(eGain Communications, 408-737-7400; www.egain.com)<B> follows the general model of receiving e-mails, classifying them and building responses. Like other products, it uses ticket numbers to identify related messages, supports manual or automatic message classification, and can generate responses automatically or send messages and suggested replies to queues for manual processing. Messages can be


assigned to individual agents based on queue membership, agent work load, specific skills, or ownership of a particular case. When messages have gone unanswered for too long, they can be automatically escalated to a higher priority queue. The system can also issue alarms when the number of messages in a queue exceeds a specified level.


The message creation interface can apply standard headers and footers, look up response templates with a search engine and use macros or bookmarks to find related templates. Replies themselves are Active Server Pages, so they can be linked to external sources. The system cannot currently recommend more than one response to a query, although this is scheduled for a release this month.


eGain lets users add custom fields to its standard data model. These can be populated directly by agents or automatically by importing data from structured e-mails. These e-mails are created by specialized Web forms created with an eGain form builder. Custom fields can be carried with messages to use in routing or personalized responses.


Data outside the eGain database can be hyperlinked to eGain screens without programming. However, scripts are needed to use the external data in routing rules. Scripts also may be needed to write complex routing rules even when they only use internal data. An API lets external systems use eGain data and services.


eGain does not have its own automated message interpretation tool, although it recently announced an agreement to integrate the Aptex neural net-based system for classification. The system does have an outbound mailing capability with basic list import and selection capabilities. It can assign a deferred start date and time to a project, but not schedule a mailing for repeated execution. Users can limit the number of messages generated per hour, to avoid overloading the company mail server. Response analysis depends on tracking codes embedded in reply messages. The system provides real-time performance reports and standard report templates along with Crystal Reports.


The entire eGain system runs through a browser interface. It uses Windows NT servers and the SQL Server database, and it's designed to support large volumes through multiprocessor, multiserver operations.


eGain EMS was released in August 1998 and has about 25 customers. About 80 percent of its customers let eGain run the system for them for a monthly fee. A one-time license, without Aptex, costs $50,000 for an unlimited number of users.


<I>David M. Raab is a consultant specializing in marketing systems and analysis. He is based near Philadelphia.<I>
Share this article:
close

Next Article in Database Marketing

Sign up to our newsletters

Follow us on Twitter @dmnews

Latest Jobs:

Featured Listings

More in Database Marketing

What's H-appending? DiscoverOrg Taps Marketo's Webhooks

What's H-appending? DiscoverOrg Taps Marketo's Webhooks

Cloud-based marketing automation behemoth Marketo joins forces with marketing intelligence company DiscoverOrg to improve its data collection capabilities.

A Toast to Marketing Attribution

A Toast to Marketing Attribution

Vino accessories and storage company Wine Enthusiast indentifies top and underperforming affiliates using algorithmic marketing attribution.

Q&A: When (and How) to Bust Down the Data Door

Q&A: When (and How) to Bust Down the ...

Some people run into issues with trying to build the perfect solution when often an 80% solution will do, says MailChimp's chief data scientist.