MVP: Web Integration
2023
AwesomeQA is an early-stage startup, that offers an AI-powered solution delivering instant responses to web3 communities across Discord, Telegram, and websites. The bot taps into FAQs, documentation, and chat logs for precise answers with verifiable references. Its web app includes a Support Portal for query resolution and knowledge management.
Opportunity
From Brainstormings sessions, we identify that 40% of our paying costumers where not offering any support besides Discord and Telegram, and ~13% where using another tool for web support, so by venturing into this new territory, we aimed to solidify our position as a strong player in the support solutions market while maintaining our commitment to innovation and user satisfaction. We commited to build a full support solution for our customers.
Building on the success of AwesomeQA, we recognized an opportunity to expand our product's reach beyond platforms like Discord and Telegram to include Websites. This strategic move would enable us to target a broader audience and meet the diverse needs of our customers seeking reliable and fast solutions to address support gaps.
Process
01 Research and Discovery
One of the founders of the company and I joined forces to better understand this potential new product. We embarked on an exhaustive journey in which we conducted user interviews with a wide range of current customers to understand, firstly, whether what we thought was a pain really was, their preferences and expectations.
​
We identified two major pain points:
​
01 Our customers expressed challenges in providing round-the-clock customer support on their websites due to limited staffing and resources. Users highlighted the need for a solution that could offer immediate assistance and answers to customer inquiries even outside of regular business hours.
​
02 Delays in receiving assistance led to customer dissatisfaction and increased the likelihood of losing potential conversions.
02 Benchmarking
In parallel, we conducted benchmarking exercises to evaluate existing chatbot solutions in the market, assessing their features, user experience, and overall effectiveness in providing customer support. This benchmarking process helped us identify industry best practices, innovative features, and potential gaps that we could address with our solution, looking at Intercom for instance, helped us a lot to get a better understanding of this field.
03 User Flow
Based on our research findings and insights, as I always like to do, I started a detailed user flow to visualize the interaction paths and user journeys within our web chatbot interface. The user flow served as a foundational framework for designing an intuitive experience as well to get the engineers, founders and myself on the same page during before designing anything.
Initial chatbot flows.
Design Iterations
After completing the initial high-fidelity version, I conducted design critiques with both the team and a few users. This process enabled me to iterate and refine the design based on feedback received.
Prototypes.
Feedback and
Further Iterations
We launched the product as a private beta for some of our clients. Through MixPanel, I began tracking performance, and with user interviews, I gathered interesting qualitative data. Despite initial promise, we realized through this feedback that key features were missing to drive the desired adoption. Among the most crucial issues:
01 Customization - Users highly valued having control over the look and feel.
02 Dissatisfaction with the bot's response time - unfortunately, this wasn't solely within our control. However, to improve user sentiment about wait times, we decided to alter the bot's interaction when generating responses.
03 Human interaction - Despite the bot offering 24-hour support, users wanted support agents intervention for scenarios such as lack of information in the knowledge base, resulting in the bot's inability to provide a response.
We sat down, analyzed, and decided that it made sense to address these points, which were feasible in terms of time and effort in order to achieve our objectives
01. Chatbot Custom UI
We enabled users to customize and adjust the chatbot to their brand. They could switch between dark mode and light mode, utilize their colors for interactive states, incorporate their logo, as well as personalize the bot's name and its message.
02. AI Thought Bubble
We enhanced the waiting time experience by transitioning from a simple three-dots loading interaction to a thought bubble when the AI is processing and generating a response. This adjustment provided users with immediate feedback on the AI's activity, fostering a more positive sentiment and improving the overall perception of wait times during interactions with the bot.
03. Support Portal
We implemented a web support portal exclusively for support agents. This portal enabled agents to promptly respond to user queries when the bot encountered limitations, ensuring a seamless support experience.
In addition, we introduced the capability for agents to directly add questions and answers to their knowledge base. This enhancement streamlined manual work processes, empowering agents to continuously improve the knowledge base's content and accuracy.
Flow.
Design critiques and user reviews.
A few final interfaces.
Results & Impact
Response Time Reduction
The implementation of our chatbot and support solutions on websites led to a 40% reduction in average response time for customer inquiries, from an average of 24 hours to just 14 hours.
Customer Satisfaction
Following the deployment of the solution, the CSAT score increased by 20 points, reaching an impressive 90% satisfaction rate among users who interacted with the support system.
Response Accuracy
Through continuous refinement and optimization, the bot achieved an impressive response accuracy rate of over 90% within the first month of deployment. This high accuracy rate demonstrates the bot's ability to provide precise and relevant answers to user inquiries, leading to enhanced customer satisfaction and confidence in our solution.
Adoption Rate
Within the first three months of launching the web version of the AI bot, the adoption rate among our customers exceeded expectations, with over 80% of eligible customers implementing the bot on their websites. This high adoption rate reflected the confidence and trust our customers have in the effectiveness and value of the AI-powered support solution.