You could call it animated FAQs. You could think of it as the precursor to Data, the famous starTrek android. Or, you could see it for what it is: a new development in the quest to gather information about customers and thus provide them with relevant products and services.
"It" is Eve, the eGain Assistant that was developed by Big Science, a company recently purchased by eGain. Eve calls herself a "virtual service agent." But Tom Rearick--now the director of products for eGain and previously the founder and president of Big Science and one of the main brains behind Eve--likes to say of assistant technology that, "In the right hands, it's the ultimate permission-based marketing/sales tool."
When customers log on to an assistant-equipped Web site to ask questions or seek information, Eve's face appears. Her expression changes each time she answers a question--or asks you a question. eGain defines her as a conversational software robot: an application server that understands natural language and can access back-office processes. Eve can handle hundreds of simultaneous conversations on a regular PC. As she responds to customer questions, a log is created that helps companies track activity and make appropriate adjustments, such as adding a new question to their FAQs.
It is possible to stump her. Thinking as a future customer, and trying to learn what might happen to me after I bought eGain, I typed in "eGain implementation." I got her clueless expression. Her answer: "Click here if you want to use a search engine. Otherwise, wrap some grammar around 'eGain implementation.'"
I figured "eGain installation" would be more understandable. It was. She correctly referred me to the installation and consulting area of the site. But typing, "I have a problem with an eGain product" caused her to go into selling mode and suggest I "click here to check out an overview of our innovative solutions."
So Eve has more to learn, something that even her developers freely admit. However, the concept behind this product, and the lessons the developers have learned, would be useful to anyone attempting to humanize a Web site, reduce costs and split the difference between the frustration of search engines and the high cost of live, human interaction. We can certainly picture the Eves of the world getting to the point where they answer questions relatively well and always remember who a customer is, engendering and increasing customer loyalty.
stands to Reason
Assistants like Eve are based on case-based reasoning. They store what they learn, along with the knowledge you've already supplied them, in "cases." When given new questions or information from an end-user, Eve then selects the nearest matching case from its programmed and learned repository, and then uses that knowledge to respond appropriately to the new input.
I see this as the science of appropriate responses, which is still in its infancy (with bots and humans). Rearick is right; it does share characteristics with permission marketing, where you always make the next move after you've gotten permission--or earned the right--to do so. It's a give-and-take dialog. A conversation between two people, Rearick points out, is "really a very complex process of correction, validation, pacing, turn taking and acknowledgement.
"If you ask a potential customer to answer several questions online, early in an interaction, they're most likely to refuse," Rearick asserts. "But if the customer asks the bot a question, and she either answers the question or escalates it, and then the customer thanks her, she can respond with, 'By the way, do you mind if I ask you a few questions?' This follows a very simple emotional logic that comes from kindergarten: 'If you're nice to me, I'll be nice to you.' In this case, an expression of gratitude triggers the survey." People who have just been helped or entertained are more willing to take a survey, fill out a form or provide their e-mail address.
If you're interested in adding a bot to your site, first you need to decide what you want your bot to do (marketing, sales or service, for example), then compile the demographics of your end-users (so her responses will be appropriate), identify the sources for the assistant's knowledge base, create task flowcharts, define how the bot will operate within your Web site and determine the review process.
You will need to create a character that is appropriate for your company, including the story behind the character, which should be integrated into her responses to make her (or him) more believable. (If you ask eGain's Eve when she was born, she replies, "I was born September 24, 1998. I'm a baby... even in Internet years. I am constantly learning, though.")
You'll want to photograph a real person, rather than use an animated character. eGain provides a set of instructions for the photographer. And use an actor, not a model. Rearick discovered that models are typically expressionless, whereas actors are experts at communicating with their faces.
What the user will see on your Web site will be comprised of three browser-accessible items: the image of the Assistant, a speech bubble for the Assistant and an input text box for the user.
Interestingly, the more helpful the bot appears, the more intelligent she seems, according to Rearick's research. A catalog of emotional responses associated with the eGain Intelligent Assistant includes neutral expressions, joy, distress, satisfaction, relief, disappointment, fear, pride, shame, admiration, gratitude, mistrust, perplexity, remorse, and even love and hate.
You'll need to assign a programmer to do the coding and a writer to write the bot's responses. Programmers can't write appropriate response language, any more than marketers can write clean code.
Then you'll want to gather up all the expertise that you'll be transferring to the bot. "Usually the best information comes from a help desk, where they know which questions are always in the top 20 percent of questions asked," Rearick says.
The Assistant is an application server that acts as an intermediary between customers and back-office processes such as messaging systems, remote Web servers, file systems and databases. eGain has tools that automate the data-gathering to build the knowledge base. Since the bot will be interacting directly with your databases, you don't have to worry about her being up-to-date.
A new version of eGain Assistant has build-in defaults that will get you started quickly. The default Assistant can answer questions, guide a Web site tour, help a user take a survey and escalate the user to e-mail or live interaction. The Assistant's knowledge is categorized, including self knowledge, company knowledge, Web-site knowledge, product line knowledge and eRetail pre-sales support. Customization beyond these areas will require consulting help from eGain. However, the eGain "Quick start" guide claims that an experienced developer (or sales engineer) could build a prototype in two to three hours, starting with the "fully-functional but non-customized bot."
Rearick is an interesting guy who believes that the Web is truly in its infancy as a medium. He laments the fact that search engines can "keep you searching for days," and "frequently asked question sections only prove that your questions are not asked very frequently." He also thinks that publishing customer support phone numbers on a Web site is "another way of saying, 'You must leave this place because there is no help here.'"
"What we need to do is make computers emotionally intelligent so they can communicate better with emotional humans," Rearick postulates. "The ridiculous alternative--which is the situation we have today--is to expect humans to communicate like computers." Rearick is on a quest to "rehumanize the human race," by introducing intelligent and engaging automated characters. Eve is just the beginning.