How to Engineer Anomalies on a Seating Map – Or, Building the StubHub-76ers Ticketing Platform

By: Camila Franco, Director, Product Management, StubHub

In July, for the first time ever, the Copa America Centenario was held outside of Latin America.

“COPA” – as it’s known – is one of the largest futebol tournaments in the world – and for a native Brazilian like me, it was a chance to welcome my sports culture to the United States.

I volunteered with StubHub’s last-minute sales team during the Mexico-Chile match, and helped distribute tickets to fans who were picking them up. The excitement was incredible as fans came out to support their respective countries.

That’s what people usually think when they hear “StubHub” – a way to access their favorite teams, artists, performers and events. But there’s a lot that goes on behind the scenes with our platform, and as director of product management, I get as excited about the technology and innovation our teams are driving as the chance to connect our customers with must-see events.

One of the most innovative new features on StubHub launches today, in partnership with the Philadelphia 76ers.

StubHub is now the Official Ticketing Partner for the team, and we have built a first-of-its-kind ticketing experience that blends all types of ticket inventory.

What does that mean? In the past, people with 76ers tickets used StubHub to resell those tickets if they couldn’t go to, or weren’t interested in, a particular game. Now, all 76ers fans can use StubHub to buy all types of 76ers tickets – tickets sold directly from the team, and tickets resold by other fans. We’re making this happen through a new platform and purchase experience on StubHub – which is also a complex technology problem when you consider the landscape of tickets and the need to make it possible to purchase a combination of tickets, potentially from multiple sellers, in one transaction – and, to bundle other game-day services like parking.

For example: Think about a fan who wants to buy eight tickets in a single row to a 76ers game. In the past, that fan could only search for sellers offering a set of eight tickets in a single row, or do multiple searches for smaller sets of tickets that, when combined, would provide eight tickets in a row. StubHub’s new platform allows this same fan to engage in one search to locate all possible seat options that meet her needs – including tickets from an individual seller offering eight tickets in a single row, as well as tickets from multiple sellers that, when combined, total eight tickets in a row.

In short, the platform puts the most optimal sets of tickets together according to a fan’s needs, in one transaction.

Going behind the scenes

Much like os jogadores de futebol assess an individual team when they prepare for game-day, at StubHub, we had to evaluate the 76ers’ ticket ecosystem separately from any other team, venue and even seating configuration to understand how to best build this new platform.

Not only did this allow us to solve the complex challenge of blending different types of ticket inventory within a venue-specific configuration, but by applying data science and manual data cleaning, we made sure the technology underpinning the new platform learns as it goes to provide the best option for customers’ needs.

At the highest level, we decided to optimize the ticket platform for mobile and search, which not only required us to design for simplicity and screen size, but also allowed us to capitalize on fans’ increasing demand to browse and buy anytime, anywhere, and to give them a great experience whether they’re familiar with where they want to sit or not.

As a result, the platform uses a new, proprietary algorithm to facilitate ticket selling, browsing and buying at the best value that meets fans’ specific needs.

To make this easier to visualize:

sixers-sh-iphone-1sixers-sh-platform-2

World-class engineering

There were three prevailing engineering considerations we took into account when building this technology:

  1. We needed to ingest ticket inventory from multiple sources and build an algorithm to blend all types of tickets together, both primary and resale, as well as layer on additional goods or services such as parking.

Blending this data together was unique in that the inventory is perishable — the value of tickets and parking disappear after the event ends. This means the algorithm needs to work in real time to consistently ingest and refresh new inventory.

  1. Data science and machine learning were critical to building the platform. We needed the system to learn the nuances of various seating configurations in order to configure what inventory could and could not be packaged together.

Not all venues are created equal. While one might assume that all seats in an arena are configured in a uniform manner, with seats spaced and numbered consistently (seat AA equidistant to seat BB), in reality, there are often empty spaces behind seats, gaps between corridors, suite configurations, and other facts that create anomalies for our mapping technology.

Our algorithm had to understand these anomalies in seating so that we didn’t split groups of fans, or put them in a seat that doesn’t exist. To solve this problem, we worked together with the 76ers to map the entire venue manually, noting various configurations for basketball games and other events. We then built an algorithm using machine learning and data science to decipher which seats/tickets should and should not be blended together, building different formations to ensure ticket-buyers are sitting together when they get to the event and ensuring that the platform will improve with every event and new configuration.

  1. In building the platform, no stone was left unturned. We evaluated and updated the entire StubHub backend to ensure our new platform functions properly and that additional StubHub features like ticket fulfillment, customer service, and fraud protection work seamlessly. We overhauled much of our existing infrastructure when creating this technology.

Throughout this process, we evaluated a number of technologies to determine which would provide the most flexibility and stability on a global scale.

We selected Solr because it provides fast computation at the massive scale required by StubHub, as well as excellent stability. While we initially prototyped Hbase, it ultimately did not perform as well, and neither Spark nor neo4J had the stability we required.

Creating a customer-friendly experience

The majority of StubHub customers – 97% – purchase the best-available listing. Understanding this, the new 76ers platform surfaces the best-available seats first to make it easier for customers to make their purchase decision.

Because of the quantity of tickets available on StubHub, as we were building the platform, we realized that a customer searching for a group of four seats could be presented with more than 1,000 different purchase options. That can be an overwhelming number of results, so we used machine learning to scale back the display results and recommend the best combination of those four tickets based on value. Once those tickets are purchased, our technology updates the listings with the next best available tickets for that set of parameters.

Most websites allow customers to make all of their purchases in a single transaction, and our technology now provides a simple shopping cart experience that allows customers to layer on additional items such as parking passes. To do this, we built a platform that not only blends primary and secondary tickets, but also bundles other inventory into a single purchase.

Finally, we know that StubHub customers and 76ers fans, like any of us, can have a change in plans, so our barcode integration makes it just as easy to resell tickets on the platform as it is to buy them.

Looking ahead

Building the technology to support the first ticket marketplace that offers all types of available tickets in partnership with the 76ers is a significant accomplishment. Understanding the nuances of one stadium allowed us to address many of the complexities we expect to face at other venues as we make this type of platform available to more partners.

For today’s fans, a ticket is a ticket is a ticket. They want the best seats for the best value. The new StubHub.com/Sixers delivers in an industry-changing way, and we’re excited to scale and improve upon it through interactions with more users and fans.

 

5 thoughts on “How to Engineer Anomalies on a Seating Map – Or, Building the StubHub-76ers Ticketing Platform

Add yours

  1. That is so interesting, Camila! These are formidable UX improvements that totally make the difference. I wish everyone else were that mindful of their customers. I’m pretty sure this 76ers experiment will improve your KPIs and give you the green light for expanding further to other big teams and even other sports!

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Up ↑

%d bloggers like this: