Recently, our Chief Customer Officer Donna Peeples discussed the importance of building bots with ‘purpose’ in an article for Direct Marketing News. Peeples, along with other industry thought-leaders and executives, shared insights into the process of implementing chatbots and automation.
As Peeples explains, the challenge with bots is keeping expectations in check. The technology, for the most part, is still new to the mainstream. Consumers and businesses alike are unaware of their limitations or true potential, and this is causing a disconnect between what people expect of bots and what they can actually deliver.
“One of the biggest misconceptions companies have about chatbots is they can answer anything and everything,” says Peeples. “The belief that automating conversations in an open-ended way will in itself add value for customers. The reality is the most effective bots are purpose-built to solve very specific problems for customers –making common customer service requests and commerce easier while ensuring customer privacy.”
This topic has raised a lot of questions, particularly for those new to the world of chatbots and intelligent automation.
Naturally, with artificial intelligence such a hot topic right now, misconceptions have led many to assume bots are so advanced that they can engage with people in a very open and fluid manner. But the reality is quite different. Instead of an open-ended conversation, bots use decision-tree modeling to answer specific questions and commands. The conversations run through a set of parameters that keep the user on a specific course until they reach a point of resolution.
So for firms to take advantage of this technology they should determine the ideal use case and purpose for the bot. As Peeples explains:
“Before designing a bot, businesses need to think deeply about their goals. How will a bot relate to customers? What specific problems will it solve? How will it improve existing processes for customer service, communication, and commerce?”
The reality is that the idea of a free-flowing, open-ended experience, is still a long way away. The amount of data needed for such an exchange is staggering, and huge leaps need to be made to get us to that point.
Consider the amount of data that the likes Google has at their disposal. There are over 100 billion Google searches every month. That’s 2.3 million per second. And yet, with all those user inquiries their Allo messaging bot still doesn’t function like a ‘human’. Instead, it operates like most other chatbots, (albeit with some impressive functionality and processing). The point is, if Google can’t successfully program a human-like chatbot due to lack of data, then there’s little point expecting any other business to do so. It’s simply not realistic.
Instead, it’s more beneficial for firms to explore how strategic intelligent automation can drive customer engagement, loyalty, and revenue. Strategic automation is about honing in on specific areas of the customer engagement cycle and building programs that can streamline repetitive processes. Easily-answered, frequently-asked questions are particularly great examples of use-cases for successful chatbot implementation.
In fact, in an article on VentureBeat titled, 7 Tips for Making a Successful Chatbot, Mariya Yao from TopBots.com outlines how beneficial a simplistic approach has been for some of the best bots available today. Yao draws on examples from well-known bots such as 1-800 Flowers, Swelly, and TheScore, to illustrate how fewer user options result in a better the overall experience. These bots don’t need to provide users with open-ended conversations. They serve one purpose, and they do it exceedingly well.
“1-800-Flowers initially offered customers three delivery date options: Today, Tomorrow, or Choose A Date,” says Yao. “The third option was removed after date input errors were found to be causing the bulk of failed transactions.”
This is an apt example of how providing clear parameters for the conversation streamlines the process and increases the likelihood of a successful interaction. To do this, Yao suggests using buttons where possible, replace user actions with automatic transitions and “limiting the options to core use cases.”
5 Steps to Intelligent Automation
Not sure where to start with chatbots? Enterprise bot implementation involves five key stages. The process ensures automation is strategic and addresses the most business-critical parts of the customer service experience to ensure a positive impact. The steps are outlined below:
- Opportunity analysis: Focus on assessing the current situation and identifying where there are high volumes of customer inquiries that require repetitive answers from agents. Through this phase, qualitative interviews with customer service agents reveal common pain-points and thorough analysis of IVR scripts reveals areas ripe for chatbot automation.
- Bot design: Once the initial focus has been set and the specific customer issues have been identified to be automated, bots are designed to solve only for those problems. This stage also includes assessing the integrations needed to ensure the bots work in conjunction with other back-end systems.
- Engineering: Bots are built to the specifications as outlined in phase 2 with close collaboration with client to ensure all needs are met.
- User-acceptance testing: Bots are used rigorously in a test environment and adjusted as necessary after input from the client and developers.
- Activation & optimization: Once deployed, the bots need to be tracked and analyzed to identify areas of optimization. Bot performance is measured against a set of clear KPIs and executed against specific goals.