This content has been archived, and while it was correct at time of publication, it may no longer be accurate or reflect the current situation at Microsoft.
When communicating, people like to have options. Some prefer chatting face-to-face, while others like to gather their thoughts and send messages when they’re ready.
Microsoft has found the same flexibility is needed to help employees who reach out to the company’s Helpdesk.
“With any customer-support interaction, you want people to use the medium they’re most comfortable with,” says Pete Sysum, a principal IT service operations manager for Global Support in Microsoft Digital.
Before the company’s conversational bot made its debut, a Microsoft employee seeking customer support had one choice for remote IT support—they could call a global-support phone number and speak with a live agent. Later, employees could use a chatbot, but they received only hard-coded responses based on keywords, so there was no way to dynamically identify the questioner’s intent.
Fast forward to today, and Microsoft Digital’s Global Support Team is giving employees the option to interact with a conversational chatbot that was built using Microsoft technology. The bot uses natural language processing and keywords to understand what an employee needs. Then, it connects them to relevant support content.
“The chatbot provides a single front door to Microsoft’s Helpdesk,” says Kevin Verdeck, a knowledge manager for Global Support. “Anytime a user has an issue, they only have to go to one location. It makes things easier to find, and resolutions are quicker to obtain.”
[Learn how Microsoft uses AI and chatbots to simplify finance tools at Microsoft and how Microsoft increases sales by using AI for lead qualification.]
Supporting employees around the globe
Microsoft employees access the chatbot through Microsoft Teams or via a web portal. Once the bot identifies an employee’s needs, it walks the employee through a workflow for common global support scenarios, directs them to online customer-support content, and, if needed, connects them to a live agent.
“We’re attempting to solve ease of support, reduce customer effort, and provide different modalities that people prefer to use,” Sysum says.
The employee has the option to connect to a live agent at any point in the interaction.
“That’s one piece of feedback we received from our user community,” Sysum says. “They don’t mind using a bot as long as they have the option to speak with a real person.”
Sysum emphasized that the bot is only as good as the content it draws from. To ensure that the content is accurate and easy to follow, the team rewrote its self-service content using the readability index in Microsoft Word. It evaluates readability based on the average number of syllables per word and words per sentence.
“To make sure we got it right, we tested our revised content with live agents to make sure it was helping them be more successful and faster at resolving queries,” Verdeck says. “Once we make sure they work for our agents, we’ll start releasing them to employees who reach out to us for help.”
Additionally, while most of Microsoft’s support content is written in English, the bot can translate it into more than 60 languages using the Azure Translator Text Cognitive Service. This ensures that employees around the world can get help in their native languages.
“We have a global customer base, so we want to make sure that this experience is optimized for everyone,” Verdeck says.
Getting bot deployment right
Global Support’s chatbot usage has increased from 500 to 4,000 tickets since November 2018. Verdeck is excited about this success, and he hopes to keep iterating on this chatbot.
“There is a lot of opportunity to reach more users, and have a greater impact,” Verdeck says. “We can offer more functionality, capabilities, and accuracy.”
In the future, the team is refining the bot’s ability to ask follow-up questions so it can understand what employee needs and route them to the right answer. If the bot can better identify the source of the problem, it can direct them to the most relevant support content.
“An employee may come in and say, ‘I have an access issue,’” Sysum says. “We want the bot to say, ‘Are you having a problem connecting to the Access app, network, or a specific application or service?’ We want the bot to drill down to the next level.”
What are the biggest lessons he’s learned while developing the chatbot? Verdeck says that it’s imperative to start the work by identifying the bot’s core features and requirements.
“Does your bot need to have a live agent chat? Does it need to be able to do workflows? Does it need to sync up with a knowledge base?” Verdeck says. “Start with your vision and work backward. Otherwise, you will start over every time you add new features.”
Learn how Microsoft uses AI and chatbots to simplify finance tools at Microsoft and how Microsoft increases sales by using AI for lead qualification.
