Making a Case for a Chat Bot.

A conversation with Kiefer consultants yield some thoughtful insights about chat bot development, cognitive services, artificial intelligence and more.

When our consultants discuss solutions with clients, we focus on the problems that the solution will solve. Artificial intelligence, cognitive services and bot technologies certainly have the “cool” factor, but do they have a business application? And furthermore, say we did find a good business case, would the solution be affordable and could we count on the solution to reduce costs and improve efficiencies.

I had the opportunity to spend time with a few of our consultants to talk about bots and what went into building the Kiefer bot. Gatika, Satya and Harkeerat had some great perspectives.

Brian: Thanks for taking the time to join me today. My first question is an important one. Bot development seems like a pretty significant departure from our core business areas. Is it?

Satya: Actually, it really isn’t. Our practice areas are well defined. Bot development requires the expertise we’ve developed across our Modern Workplace, Data Analytics, and Cloud Services practice areas. Bot development is exciting because it takes everything we do and takes it to another level. Bot development without a strong foundation in cloud services and data analytics could result in a solution that does not meet the expectations of the client.

An image promoting the blog called Making the Case For Chat Bots.

Brian: Tell me a little bit more about the data analytics aspect of bot development.

Gatika: Without data, a bot is useless. The Microsoft Bot Framework will give you a pretty good start when you build a bot, however, for a bot to really be effective, you must create a knowledge base that addresses the questions that people would typically ask your bot. When building a bot, you aggregate all the possible questions and answers and train your bot. The bot will attempt to answer questions based on the data you provide. The bot will also apply a “confidence score” to the answers it provides, which allows you to validate the bots responses and ultimately improve the accuracy and quality of the answers the bot provides. So, it is important to highlight the importance of a good knowledge base when you start developing a bot.

Satya: I agree with Gatika. We had a client that wanted to build a bot that could be used to respond to users that required IT help. The bot was going to be used internally. Unfortunately, the knowledge base (KB) was not built. We suggested that the client start by coming up with the most common questions that come to the help desk. Once you have all the questions, we instructed the client to answer the questions.

What we did learn in our conversations with the client, about 15-20 questions make up 75% of the questions the help desk typically responds to. This indicates that the bot would be very helpful in addressing common questions and could be effective in reducing the number of help desk tickets that require staff intervention.

Banner for a free Kiefer eBook called the Indispensable Tools of the Modern Workplace

Brian: This is interesting. Satya, you seem to make a pretty good business case for a bot. Reducing the amount of help desk tickets by resolving issues before they have to go to a staff member.

Satya: Exactly. You can justify the cost of building a bot by determining how many help desk tickets or human interactions could be reduced by having clients or customers interact with a bot before they are routed to a staff member.

Harkeerat: This is an important part of bot development. Designing an experience that users find helpful is critical. If the user of your bot finds that it can’t provide an acceptable answer or if it feels like a barrier to getting a question answered or an issue resolved, then you risk losing user confidence in your ability to serve the client’s needs. We recommend that the user be offered a way to speak to a real person if the bot interaction is proving to be ineffective.

Gatika: It’s important that the team building your bot understands data. We look at data from the bot to determine what questions are being asked and whether the bot’s responses are helpful. We can measure how successful the bot is at answering questions and gather metrics that tell us how satisfied users are with the interaction. This is an important part of bot development and we rely heavily on what we learn from the data.

Brian: Besides the example you shared on that resulted a reduction of help desk tickets, what other benefits do you see for organizations that start using bots?

Harkeerat: A bot gives you a unique opportunity to learn a lot about your user. A bot is interactive and is “on” 24/7. What you learn through the bot enables you to improve your message and may even result in you redesigning your site to highlight specific information that users looking for.

Satya: As you may have noticed, many websites don’t readily provide visitors with a phone number. The goal is to use tools like bots as “tier 1” support. The technology has improved and there is a real opportunity for organizations to use bots strategically to relieve the burden placed on a call center, help desk or customer service department. I see a huge opportunity for bot technologies in state and local government. It is also very important to point out that this development effort is typically a “NO CODE” effort. To deploy and maintain a bot you don’t need developer expertise.

Harkeerat: Reduction in tickets, reduction in wait times… this is a benefit. And a measurable one.

Gatika: I think it is also important to mention integration of the bot with other systems. By integrating your bot with ticketing systems like Salesforce or ServiceNow, organizations can leverage their investments and extend the capabilities of their systems, while streamlining the user experience.

Brian: This has been fascinating. I can see this type of technology continuing to evolve. While our company has been a critical partner in helping organizations better organize data through logical information architecture. In the future, will people seek out information through natural language?

Satya: You may be right. A webpage with a complex menu and submenus could be challenging to navigate.

Assume you are looking for a form or specific information on a complex public-sector website. A bot on the front page of the website could replace top navigation completely. You ask the bot… “Where can I find Form 19-8?” – The bot can take you right to the form. This could provide a better a user experience and deliver all the benefits mentioned previously – shorter wait times, fast resolution, reduced call volume.

Brian: Thanks for everyone’s time. Let’s do this again!

Gatika: Yes.

Harkeerat: No problem.

Satya: Let’s do it again.

Interested in Learning More?

If you’d like to talk about bots with our team, contact us. We can show you a demo of the bot framework, virtual agent or our bot. We look forward to talking to you about your business case!


Follow Us