Microsoft Bot Framework
AI powered Support BOT
Company
Microsoft
My Role
UI/UX Designer & PM
Tools
Adobe XD Microsoft Bot Framework LUIS
Timeline
2017 – 2018
Description
AI-powered BOT that provides an improved customer support experience across Microsoft Products.
Context
The Microsoft Bot Framework enables users to build intelligent, enterprise-grade bots with ownership and control of their data. Begin with a simple Q&A bot or build a sophisticated virtual assistant. Use a comprehensive open-source SDK and tools to easily connect your bot to popular channels and devices. Give your bot the ability to speak, listen, and understand your users with native integration to Azure Cognitive Services.
Problem
Microsoft comprises a large number of legacy programs. Over 10 million users pass through every quarter, with ~5-7% needing support assistance.
Phone and email were the preferred methods but came with large disadvantages: high budgeting costs, long SLAs, and difficulty diagnosing root problems.
Process
An internal and external search was conducted for automating this strenuous process. The team included 3 support agents with intimate knowledge of user's pain points. This turned out to be crucial.
We narrowed down the top 7 support categories representing ~80% of the total support volume.
We identified a back-end demo for support bots (Q&A Maker) developed by an MS team and decided to use it as a backbone for a Support BOT able to answer our top 7 support categories.
This idea checked all our initial requirements:
Low budget costs & great scalability
Instant SLAs
Easy to diagnose problems
Demo & What would success look like
How do we measure Success?
Ticket Deflection Rate: between 10% - 20% (reduce # of support cases)
Success Rate >75% for Top 7 FAQs
Customer Satisfaction - nSat >140*
Evolving design:
Key Insights
1.5 years testing, designing, iterating, and using all the latest technologies. Having a team comprised of support engineers working directly with users gave us incredible insights, helping us to design a targeted bot with clear goals in mind.
Microsoft Bot Framework
AI powered Support BOT
Company
Microsoft
My Role
UI/UX Designer & PM
Tools
Adobe XD Microsoft Bot Framework LUIS
Timeline
2017 – 2018
Description
AI-powered BOT that provides an improved customer support experience across Microsoft Products.
Context
The Microsoft Bot Framework enables users to build intelligent, enterprise-grade bots with ownership and control of their data. Begin with a simple Q&A bot or build a sophisticated virtual assistant. Use a comprehensive open-source SDK and tools to easily connect your bot to popular channels and devices. Give your bot the ability to speak, listen, and understand your users with native integration to Azure Cognitive Services.
Problem
Microsoft comprises a large number of legacy programs. Over 10 million users pass through every quarter, with ~5-7% needing support assistance.
Phone and email were the preferred methods but came with large disadvantages: high budgeting costs, long SLAs, and difficulty diagnosing root problems.
Process
An internal and external search was conducted for automating this strenuous process. The team included 3 support agents with intimate knowledge of user's pain points. This turned out to be crucial.
We narrowed down the top 7 support categories representing ~80% of the total support volume.
We identified a back-end demo for support bots (Q&A Maker) developed by an MS team and decided to use it as a backbone for a Support BOT able to answer our top 7 support categories.
This idea checked all our initial requirements:
Low budget costs & great scalability
Instant SLAs
Easy to diagnose problems
Demo & What would success look like
How do we measure Success?
Ticket Deflection Rate: between 10% - 20% (reduce # of support cases)
Success Rate >75% for Top 7 FAQs
Customer Satisfaction - nSat >140*
Evolving design:
Key Insights
1.5 years testing, designing, iterating, and using all the latest technologies. Having a team comprised of support engineers working directly with users gave us incredible insights, helping us to design a targeted bot with clear goals in mind.
Microsoft Bot Framework
AI powered Support BOT
Company
Microsoft
My Role
UI/UX Designer & PM
Tools
Adobe XD Microsoft Bot Framework LUIS
Timeline
2017 – 2018
Description
AI-powered BOT that provides an improved customer support experience across Microsoft Products.
Context
The Microsoft Bot Framework enables users to build intelligent, enterprise-grade bots with ownership and control of their data. Begin with a simple Q&A bot or build a sophisticated virtual assistant. Use a comprehensive open-source SDK and tools to easily connect your bot to popular channels and devices. Give your bot the ability to speak, listen, and understand your users with native integration to Azure Cognitive Services.
Problem
Microsoft comprises a large number of legacy programs. Over 10 million users pass through every quarter, with ~5-7% needing support assistance.
Phone and email were the preferred methods but came with large disadvantages: high budgeting costs, long SLAs, and difficulty diagnosing root problems.
Process
An internal and external search was conducted for automating this strenuous process. The team included 3 support agents with intimate knowledge of user's pain points. This turned out to be crucial.
We narrowed down the top 7 support categories representing ~80% of the total support volume.
We identified a back-end demo for support bots (Q&A Maker) developed by an MS team and decided to use it as a backbone for a Support BOT able to answer our top 7 support categories.
This idea checked all our initial requirements:
Low budget costs & great scalability
Instant SLAs
Easy to diagnose problems
Demo & What would success look like
How do we measure Success?
Ticket Deflection Rate: between 10% - 20% (reduce # of support cases)
Success Rate >75% for Top 7 FAQs
Customer Satisfaction - nSat >140*
Evolving design:
Key Insights
1.5 years testing, designing, iterating, and using all the latest technologies. Having a team comprised of support engineers working directly with users gave us incredible insights, helping us to design a targeted bot with clear goals in mind.
Microsoft Bot Framework
AI powered Support BOT
Company
Microsoft
My Role
UI/UX Designer & PM
Tools
Adobe XD Microsoft Bot Framework LUIS
Timeline
2017 – 2018
Description
AI-powered BOT that provides an improved customer support experience across Microsoft Products.
Context
The Microsoft Bot Framework enables users to build intelligent, enterprise-grade bots with ownership and control of their data. Begin with a simple Q&A bot or build a sophisticated virtual assistant. Use a comprehensive open-source SDK and tools to easily connect your bot to popular channels and devices. Give your bot the ability to speak, listen, and understand your users with native integration to Azure Cognitive Services.
Problem
Microsoft comprises a large number of legacy programs. Over 10 million users pass through every quarter, with ~5-7% needing support assistance.
Phone and email were the preferred methods but came with large disadvantages: high budgeting costs, long SLAs, and difficulty diagnosing root problems.
Process
An internal and external search was conducted for automating this strenuous process. The team included 3 support agents with intimate knowledge of user's pain points. This turned out to be crucial.
We narrowed down the top 7 support categories representing ~80% of the total support volume.
We identified a back-end demo for support bots (Q&A Maker) developed by an MS team and decided to use it as a backbone for a Support BOT able to answer our top 7 support categories.
This idea checked all our initial requirements:
Low budget costs & great scalability
Instant SLAs
Easy to diagnose problems
Demo & What would success look like
How do we measure Success?
Ticket Deflection Rate: between 10% - 20% (reduce # of support cases)
Success Rate >75% for Top 7 FAQs
Customer Satisfaction - nSat >140*
Evolving design:
Key Insights
1.5 years testing, designing, iterating, and using all the latest technologies. Having a team comprised of support engineers working directly with users gave us incredible insights, helping us to design a targeted bot with clear goals in mind.