This post was co-authored with GoDaddy’s Mayur Patel, Nick Koenig, and Karthik Jetti.
GoDaddy empowers everyday entrepreneurs with all the help and tools they need to succeed online. With 21 million customers worldwide, GoDaddy’s global solutions help entrepreneurs seamlessly connect their identity and presence with commerce for profitable growth. At GoDaddy, we pride ourselves on being a data-driven company. The relentless pursuit of valuable insights from data drives business decisions and strives to achieve customer satisfaction.
This post describes how GoDaddy’s Care & Services team worked closely with the AWS GenAI Labs team to build Lighthouse, a generative AI solution powered by Amazon Bedrock. Amazon Bedrock is a fully managed service that makes leading AI startups and Amazon’s Foundational Models (FM) available through APIs. So you can choose from a wide range of FMs to find the best model for your use case. With the Amazon Bedrock serverless experience, you can get started quickly, privately customize FM with your own data, integrate it into your applications using AWS tools, and deploy it without managing any infrastructure. With Amazon Bedrock, GoDaddy’s Lighthouse uses well-crafted prompts to mine insights from customer care interactions, identify top call drivers, and identify friction points in customers’ product and website experiences. and improve customer experience.
GoDaddy’s business challenges
Data has always been a competitive advantage for GoDaddy, and so is our Care & Services team. We recognize the potential to derive meaningful insights from this data and identify key call drivers and pain points. But in a world before generative AI, technologies for mining insights from unstructured data were computationally expensive and difficult to operate.
Solution overview
That changed with GoDaddy Lighthouse, an interaction analytics solution powered by generative AI. This unlocks a rich mine of insights within customer care record data. By leveraging customer care interaction data, you can now scale deep, actionable analytics to:
- Detect and evaluate customer friction points in your product or website experience, leading to improved customer experience (CX) and customer retention.
- Improve customer care operations such as quality assurance and routing optimization, leading to improved CX and operational costs (OpEx).
- Eliminate dependence on expensive vendor solutions for speech analysis
The following diagram shows Lighthouse’s high-level business workflow.
GoDaddy Lighthouse is a large-scale language model (LLM)-powered insights solution that enables prompt engineers across your company to create, manage, and evaluate prompts using a portal that allows them to interact with the LLM of their choice. By designing the prompts that run against your LLM, you can systematically derive powerful, standardized insights across your text-based data. Product subject matter experts use the Lighthouse platform UI to test and iterate generative AI prompts that generate customized insights about care and service interactions.
The diagram below shows the iterative process of creating and enriching prompts.
After the prompt is tested and verified to work as intended, it is deployed to production and scaled across thousands of interactions. The insights generated for each interaction are then aggregated and visualized in dashboards and other analytical tools. Additionally, Lighthouse allows GoDaddy users to create one-time generative AI prompts to uncover rich insights into very specific customer scenarios.
Let’s take a closer look at how Lighthouse’s architecture and features help users generate insights. The following diagram shows the Lighthouse architecture on AWS.
Lighthouse UI leverages data generated from Amazon Bedrock LLM calls on thousands of transcripts, leveraging a library of prompts from GoDaddy’s internal prompt catalog. The UI facilitates the selection of LLM models based on user selection, making the solution independent of a single model. These LLM calls are processed sequentially using Amazon EMR and Amazon EMR Serverless. Seamless integration of backend data into the UI is facilitated by Amazon API Gateway and Amazon Lambdas capabilities, and UI/UX is supported by AWS Fargate and Elastic Load Balancing to maintain high availability. For data storage and retrieval, Lighthouse uses a combination of Amazon DynamoDB, Amazon Simple Storage Service (Amazon S3), and Amazon Athena. Visual data analysis and representation is achieved through dashboards built on Tableau and Amazon QuickSight.
quick assessment
Lighthouse offers a unique proposition by allowing users to evaluate one-time generated AI prompts using the LLM of their choice. This feature allows users to create new one-time prompts specifically for evaluation purposes. Lighthouse uses the actual transcript and response from the previous LLM call to process this new prompt.
This feature is especially valuable for users looking to refine their prompts through multiple iterations. By iteratively adjusting and evaluating prompts, users can gradually strengthen and solidify the effectiveness of their queries. This iterative improvement process ensures that users achieve the highest quality output tailored to their specific needs.
The flexibility and accuracy provided by this feature makes Lighthouse an essential tool for anyone looking to optimize their interactions with LLMs, facilitating continuous improvement and innovation in rapid engineering.
The following screenshot shows how users can use evaluation prompts in Lighthouse to verify the accuracy of model responses.
After the quality of your prompt has been evaluated and your users are satisfied with the results, you can promote the prompt to the prompt catalog.
Response summary
When a user submits a prompt, Lighthouse processes this prompt against each available transcript and generates a set of responses. Users can view the responses generated for their queries on a dedicated page. This page serves as a valuable resource, allowing users to review the answers in detail or download them to an Excel sheet for further analysis.
However, the sheer volume of responses can make this task overwhelming. To address this, Lighthouse provides functionality that allows users to pass these responses through new prompts for summaries. This feature allows users to get a concise, one-line summary of their responses, greatly simplifying the review process and increasing efficiency.
The following screenshot shows an example of a prompt that uses Lighthouse to allow users to meta-analyze all answers into one, reducing the time required to review each answer individually.
This summarization tool allows users to quickly distill large datasets into easily digestible insights, streamlining workflows and making Lighthouse an essential tool for data analysis and decision-making.
insight
Lighthouse generates valuable insights and provides a deeper understanding of key areas of focus, opportunities for improvement, and strategic direction. With these insights, GoDaddy can make informed strategic decisions that improve operational efficiency and drive revenue growth.
The following screenshot is an example of a dashboard based on insights generated by Lighthouse, showing the distribution of each insight category.
Through Lighthouse, we analyzed the root cause and intent distribution across the vast number of calls handled by GoDaddy agents each day. This analysis identified the most frequent causes of escalation and the factors most likely to lead to customer dissatisfaction.
Business value and impact
To date (as of this writing), Lighthouse has generated 15 new insights. Most notably, teams can now use insights from Lighthouse to quantify the impact and cost of friction within their current processes and prioritize needed improvements across multiple departments. That’s what it means. This strategic approach streamlined the password reset process, resulting in fewer support calls related to the password reset process, faster resolution times, and ultimately significant cost savings.
Other insights that can improve your GoDaddy business include:
- Discovering call routing flows that are suboptimal for profit per interaction
- Understand the root causes of repeated touching behavior
conclusion
GoDaddy’s Lighthouse, powered by Amazon Bedrock, represents a transformative leap forward in using generative AI to unlock the value hidden in unstructured customer interaction data. Lighthouse extends deep analytics and generates actionable insights, enabling GoDaddy to enhance customer experiences, optimize operations, and drive business growth. As a testament to its success, Lighthouse has already delivered financial and operational improvements, solidifying GoDaddy’s position as a data-driven leader in the industry.
About the author
Mayur Patel He is the Director of Software Development on GoDaddy’s Data & Analytics (DnA) team, specializing in data engineering and AI-driven solutions. With nearly 20 years of experience in engineering, architecture, and leadership, he has designed and implemented innovative solutions to improve business processes, reduce costs, and increase revenue. His work enables companies to realize their full potential through data. Passionate about leveraging data and AI, he aims to create solutions that delight customers, increase operational efficiency, and optimize costs. Outside of work, I enjoy reading, hiking, DIY projects, and exploring new technology.
Nick Koenig He is the Senior Director of Data Analytics and has spent the past 10 years building data solutions across GoDaddy. My first job at GoDaddy involved listening to calls and finding trends, so I’m especially proud to be helping build an AI solution for this 10 years later.
Kartik Jetty I’m a senior data engineer in GoDaddy’s data and analytics organization. He has over 12 years of experience in data technology, AI, and cloud platform engineering and architecture, generating data to support advanced analytics and AI initiatives. His work drives strategy and innovation, with a focus on revenue generation and efficiency improvements.
Ranjit Rajan I am a Principal GenAI Lab Solutions Architect at AWS. Ranjit works with AWS customers to help them design and build data and analytics applications in the cloud.
Satvir Krupa He is a Senior Solutions Architect on the GenAI Labs team at Amazon Web Services. In this role, I leverage my expertise in cloud-based architectures to develop innovative generative AI solutions for clients across a variety of industries. Mr. Satveer’s deep understanding of generative AI technologies allows him to design scalable, secure, and responsible applications that unlock new business opportunities and drive tangible value.
richa gupta He is a solutions architect at Amazon Web Services, specializing in generative AI and AI/ML design. She helps clients implement scalable, cloud-based solutions to drive business growth using advanced AI technologies. She also presented generative AI use cases at the AWS Summit. Prior to joining AWS, he was a software engineer and solution architect building solutions for major carriers.
