Prospect uses Amazon QuickSight self-service business intelligence tools to empower clients

This is a guest post by Dr. Mark Einhorn from Prospect.

Prospect is a data analytics company helping the sports industry use artificial intelligence (AI) to power decisions—from the field to the boardroom. Our customers include teams, leagues, governing bodies, broadcasters, ticketing agencies, and sponsors. In this post, we share how we partnered with Amazon QuickSight to shift our business model from a mainly consultative approach to offering our powerful data analysis and business intelligence (BI) tools directly to customers.

At Prospect’s core is a team of data scientists, data engineers, and DevOps engineers who develop bespoke analytics products for our clients. This allows them to gain insight and value from their data as well as use our proprietary performance models and data entities.

As we grew, we needed to scale our processes for making data intelligence accessible for clients. And that meant considering changing our business model—and technology. We have grown from 4 people in March 2021 to 45 people as of writing. Our client base and offerings have also grown. Where we started out with just one performance client in rugby union, we now have multiple clients across rugby union and league, cricket, field hockey, e-sports, and football.

Putting sports analysis technology in our clients’ hands with QuickSight

Our consulting process created timeline and resourcing challenges. Our teams faced a heavy lift before we could deliver impact for our clients. We had to ingest data, build models for each client’s needs, convert the model output into clear and tangible deliverables, and then present the results.

Our aim was to put our AI models into our customers’ hands, allowing them to interact with them, monitor the outputs, and assess data on an ongoing basis to inform the decisions they were making, without relying on the Prospect team.

But we first had to address some serious challenges to move to a self-service mode. We sought a technology solution that could do the following:

• Deliver a robust and intuitive client-facing end product
• Minimize barriers to entry for end-users in understanding and interpreting data
• Enable a real-time view of data as it is updated and processed through our AI pipelines
• Integrate with our existing data stores, tech stacks and AI model pipelines
• Work with our continuous integration and continuous delivery (CI/CD) pipeline to make sure development best practices were followed

Priorities: Integration and clarity of insight

We assessed a number of different self-service BI tools, but spotted issues with excessive licensing costs and the ability to integrate into our existing infrastructure model.

After we engaged the QuickSight team, it quickly became clear that both the cost to scale and the ease of tech stack integration made it the best possible choice for the project. QuickSight provided clarity of insight. We want users to be able to take the exact information that they need away from the dashboard visualizations, whether they have 1 minute or 30 minutes to look at it.

QuickSight impressed us in several key areas:

• Strong and simple integration with our data AWS data stack and CI/CD pipeline
• Ease of prototyping and creating visualizations for dashboards
• Ability for dashboards to display data in real time as it was ingested
• Built-in user management functionality that made it straightforward to onboard clients and manage their data access cost-effectively

Creating bespoke insights quickly with dashboard templates and baked-in security

Our models and associated dashboards vary in complexity depending on the end-users. We can create a more template-based approach for ticketing clients, which are largely similar in appearance and functionality, where the dashboards for two different football teams could look very different, depending on what the coaching staff want.

In our legacy process, we started by building a CI/CD architecture for deployment in Terraform. Then we engineered pipelines for data to be ingested, cleaned, aggregated, and stored, both for modeling and surfacing in our dashboards. Finally, we trained, tested, and validated our AI data models before moving on to hosting and monitoring them.

Working with QuickSight streamlined this process and sped up deployment. The configurable dashboard lets us build templates that are configurable for new clients.

Additionally, the parameters and row-level security in QuickSight allowed us to restrict data to a specific client, even if it queries the same data lake hosted in Amazon Simple Storage Service (Amazon S3), and restrict data access for our customers so they only view their own. This means not having to create datasets from scratch and a faster deployment time, resulting in lower maintenance and fewer datasets.

Solution overview

Our data is stored in Amazon S3, crawled by AWS Glue, tabled in Amazon Athena, and pulled by QuickSight through to our dashboards. It’s a seamless AWS ecosystem that removes a lot of work from our developers.

The following diagram illustrates the solution architecture.

A new business model

QuickSight has transformed the way Prospect operates. At the heart of this transformation is a significant shift from consultancy with a great deal of internal work to a more hands-on, real-time system clients can use themselves.

QuickSight has drastically reduced our build time. Its ability to pull from a shared data lake while preserving client data privacy and templated, configurable dashboards enabled us to scale faster and more cost-effectively.

We’re still measuring the financial impact, but initial signs point to significant gains. Clients across the sports industry are closer to our product, understanding best practices and accessing insights in a new way.

The ability for organizations to access insights in real time using direct queries is also powerful. Data fed into QuickSight is instantly available for analysis, which means our clients can make better decisions, faster.

The following screenshots showcase a few types of sports BI dashboards we offer.

This first dashboard quantifies the monetary value of brand exposure for brands associated with certain football teams.

The following dashboard shows predicted TV viewership numbers for various football leagues, filterable by different national markets.

The following dashboard is a visualization of our proprietary Prospect Interest Index; this is a proxy for fan interest and sentiment in sporting properties (teams, leagues, athletes) in different geographic markets. It blends multiple different data sources, such as website searches, website visits, and social media activity, associated with the entities.

Looking ahead

We’re refining how we use QuickSight in our initial templating, workflows, and analytical power to onboard more clients.

As we explore the capabilities of QuickSight, we have added exciting features to our roadmap. We’re planning to incorporate embedded QuickSight visuals into our technology and are exploring Amazon Q in QuickSight, a natural language model that allows anyone in our clients’ teams to dig into the data we provide and find the insight they need, without any data science or engineering expertise.

Get started with QuickSight

To explore QuickSight and how its cutting-edge features can deliver new applications, reduce in-house development time, and make data insights faster, smarter, and more accessible, visit Amazon QuickSight.

About the Authors

Dr. Mark Einhorn is the Head of Analytics at Prospect. He is a distinguished analytics professional with a PhD in Operations Research from Stellenbosch University. As Head of Analytics at Prospect, he leverages extensive experience gained from his previous role as a specialist data scientist at McKinsey and Company. Mark is known for his expertise in descriptive, predictive, and prescriptive analytics, having successfully led numerous data-driven projects. He also played a crucial role in establishing the Analytics Academy at McKinsey, where he facilitated training programs for client personnel, enhancing their skills as data scientists and data engineers. Passionate about nurturing talent and driving innovation, Mark is committed to the development of his junior team members while incorporating new technologies into Prospect’s client offerings.