Live Data Collection

NCAA Football
Problem/Opportunity
Our partner, a betting syndicate and odds provider, is looking for live, low latency and highly accurate play-by-play data for NCAA Football. The existing play-by-play data providers are often several minutes behind the in-stadium action and often riddled with inaccuracies.
Proposed Solution
Leverage SIS’ existing data collection expertise to set up an in-stadium operation consisting of 1 or 2 scouts who will enter play-by-play data as it unfolds using their personal mobile devices and software we would design and develop.
My Contributions
Sole designer & researcher for all related products (Data Collection, Auditing, and Viewing).
Business Result
Multi-year, seven figure contract for the generated NCAA Football data as well as additional subsequent opportunities for NCAA Basketball and NFL Football.

Timeline

Research & Product Reqs
Jan 3, 2023
Wireframes
Jan 17, 2023
Wireflows & Prototype 1
Jan 24, 2023
Hi-Res Screens
Jan 31, 2023
Prototype 2 & Testing
Mar 9, 2023
MVP Handoff
Apr 4, 2023
Week 0 Iterations
Aug 26, 2023

Research & Product Requirements

Who are the users?
This was a unique project in that the ultimate product was the data being sent to the client. That data however was being generated by in-stadium scouts using the proposed collection software. So while the client did not care about the collection software per se, it did need to be designed to allow the scouts to deliver the data quickly and accurately. The project would require both feedback and testing of the collection software with scouts as well as feedback and testing of the data feed with the client.
What are the constraints?
  • Unsanctioned operation
  • Mobile web only
  • Mobile network connectivity
  • Limited testing conditions due to live nature
  • Inexperienced scouts
  • Single scout situations
  • Unsynced & independent data feeds between 2 scouts (discovered later on)
What are the requirements?
The client provided an initial outline of necessary data points which we sorted into three types (Game State, Result, and Feed) and also prioritized from Tier A to D. More important data points would be prioritized in the UX/UI to ensure they were faster and more accurate. For context there were roughly 35 Game State, 20 Result, and 4 Feed data points in our initial list.

Data points were then divided between the two scouts with one responsible for the score, clock, and timeouts and the other responsible for the play-by-play action. These were referred to as the Clock Scout and the Game Scout from this point onwards.
How will the user move through the collection software?
Football unfolds in a very step-by-step fashion in terms of relevant data points (ie we are not concerned with Blown Blocks or non-targeted receivers). My thinking was to map out every possible path a game could take in the style of a choose-your-own-adventure book. Using FigJam, I mapped out these User Flows to determine which options would be necessary for the user to have at their disposal in a given situation. Showing only relevant options maximized the screen space to allow for better usability and reduced errors.

Wireframes, Wireflows & Initial Prototype

Wireframes
The design goal was to provide users with multiple feedback mechanisms (eg clock, field, game log) as well as to maximize speed and accuracy by placing the most frequently used and important inputs in the most accessible positions on the screen (ie within the natural "thumb-zone").

Another consideration was whether both scouts would share a common interface or if each would have a unique one tailored to their role. The following factors were taken into consideration here:
It was decided to have one shared interface however the Clock Scout's UI would be darkened for the Game Scout and vice versa to help reduce the distraction drawback. Once the general layout was established, initial wireframes were designed for every screen permutation that had been identified in both scout's user flows.
Single Shared Interface
  • ✅ Additional Feedback Mechanisms
  • ✅ Scout Familiarity with Both Roles
  • ✅ Possibility for Single Scout Mode
  • ❌ Distraction of Disabled Functionalities
Two Unique Interfaces
  • ✅ Tailored UI to Specific Role
  • ✅ Additional Screen Real Estate
  • ❌ Additional Engineering Complexity & Time
Wireflows & Initial Prototype
The User Flows provided direction as to which screen permutations would be necessary and the plan was that once those were complete, connecting them would be easier. With some small enhancements to the wireframes and a play-by-play of an existing game, I prototyped the flow of what a scout would follow to create that data. This validated the User Flows and provided feedback to make adjustments on certain screens.

You can see the wireflow prototype here.

Hi-Res Screens, Prototyping, Testing & Handoff

Hi-Res Screens
The goal while bringing the wireframes up to hi-res was to keep things minimal with anything more than simple text and boxes having a clear UX purpose. Icons and team logos were used where it made sense to increase user comprehension. A fair amount of time was also spent on the field graphic UI which would act as the users’ primary feedback mechanism for the Game Scout. Buttons were color coded depending on their purpose and sized depending on frequency of use/importance of data point. Animations were eschewed for MVP.
Prototyping & Testing
Using a similar methodology to the initial wireflow prototype, “happy path” prototypes were put together for three full drives of an existing NCAAF game. These were then tested using game tape with members of our Ops team to determine potential pain points. Any time a participant deviated from the happy-path, we discussed the situation at length to understand what had caused them to make an unanticipated decision. There was also ample discussion about the general UI and layout.

You can see the prototype here.

After making necessary adjustments to button layouts and flows, the MVP was determined to be ready for MVP handoff. Handoff screens were grouped into sections and paired with a cleaned up version of the FigJam user flows to ensure our engineers had sufficient context for every possible situation that would be encountered during a game.

Early Iterations & Subsequent Projects

Early Iterations
We had the benefit of not needing to deliver a fully formed, perfect product for Week 0 of the NCAAF season. The client provided us with actionable feedback relating to data type, accuracy, and speed during the first few weeks of the season in hopes of achieving the desired results later in the season.

Several unique situations and new data points arose during this time including:
  • Non-QB1 Passers (eg Wildcat)
  • Lateral Passes
  • Additional indication of updated yard line
  • Last minute roster changes
  • Workarounds for scout disconnectivity
  • Additional auditing functionality
  • Auditing Tool (Oct 2023) - A dedicated tool to allow our Ops team to audit data feeds as they come in, allowing scouts to focus more on in-the-moment action and less on previous events
  • Live Data Viewer (Nov 2023) - A simple way to see the incoming data from a game in real time in order to showcase the data to other perspective customers
  • NCAAB Live Data Collection Tool (Nov 2023) - A similar but in many ways unique tool to collect play-by-play data for NCAA basketball
Subsequent Projects

Conclusion

This was the most successful project I worked on since starting at SIS.

The initial opportunity was presented without any firm evidence that we would be able to execute what the client required. The data collection software proved to be the game changing product that was necessary in order to not only deliver on time but to exceed expectations in short order. The framework of the product was flexible enough to allow for continuous iteration to adapt to new data requirements and unforeseen situations both on and off the field.

The client was so pleased with the results that we secured a multi-year extension on our initial contract before the 2023 football season was over. In addition, we were given the opportunity to set up a similar operation for NCAA Basketball and there are discussions of something similar for the NFL in the future. These datasets are also not exclusive to the original client and opportunities to repackage it for others is being explored.