June 4, 2019

The Way We Build Better Transit

Haomin Xu
No items found.
At Swiftly, we’re relentless about setting up processes that impact the most people while reinforcing our mission to make cities move more efficiently.
June 4, 2019

The Way We Build Better Transit

Haomin Xu
June 4, 2019

The Way We Build Better Transit

Haomin Xu

The way we build

Great design at its core is the ability to understand people’s needs in order to create the best solution for them.

At Swiftly, our goal is to create the best solutions for transit agencies and cities to improve urban mobility. We know that what we build impacts not only dozens of cities that use Swiftly, but also millions of riders.

We want to build it all, but we have to prioritize. That’s why we’re relentless about setting up the right process to balance what’s helpful to the most people, and reinforces our mission to make cities move more efficiently.

To outline how we approach product development, here’s how we went about building our latest product, Headways.

Developing the Headways Module

Step 1: Understand the problem

The first step of our process is all about understanding the problems that our customers face. That way, we can base the rest of the process on solving the right problems.

For decades, the way transit agencies measured service reliability was through on-time performance: Did the bus arrive when the schedule said it would?

Where this method fails riders is on high-frequency routes. Here, riders care less about a schedule and more about having a short wait time between bus arrivals (the wait time in between buses is what we refer to as headway). For people who take public transit to go to work or to pick up their children from school, having to budget in extra time for unreliable service impacts their everyday.

To ensure that our product team understood the problem, we read literature to familiarize ourselves with headway performance and best practices around measuring reliability. We also learned how existing methodologies were far from perfect when it came to painting the full picture of the rider experience.

Step 2: Explore the current landscape

After understanding the problem on a broader scale, we draw our focus back to our users and the current landscape in which they operate. Learning our users’ current work flows and tools are key to understanding if and where we can provide value.

At Swiftly, we have a close relationship with our customers. Our product team regularly visits our agencies, observing everything from their day-to-day tasks to their overall goals. We want to keep a pulse on how our product fits into the larger ecosystem that allows our users to succeed in their jobs.

For headways, our product manager, data scientist, and product designer interviewed agencies to start narrowing in on our focus areas. We found that we could provide value in the following ways:

Surface problematic routes and impact on riders
  • Only a handful of large agencies have in-house reports on headway performance. All others have very limited insight into this information.
Answer the “why” behind the data
  • Once agencies find their problematic routes, understanding potential causes can help determine next steps.
Support custom headway definitions
  • Most agencies have different methodologies for measuring headway, and some have none at all. If we expect to our solution to scale, we have to support multiple methodologies.
Step 3: Be okay with killing the project here

After the initial discovery, we sift through our findings to determine if there is a real problem we can solve here.

We are ok with killing a project if we learn that the problem is not that severe and current workarounds suffice. There is no shortage of work, so we have to be rigorous about the problems we take on.

For headways, we learned that agencies had a real desire to improve reliability on high-frequency routes and current tooling were not measuring up. We decided there was enough meat to the problem to aim to find a solution.

Step 4: Scope the project

Once we determine that there is a real problem with high customer impact and business opportunity, the product manager facilitates conversations with the product team to discuss learnings, understand constraints, and align on the project scope.

Our product team scoped the first version of Headway Insights by focusing on the three value propositions, while working within the available data and time constraints. Ultimately, our goal with the MVP was to build a valuable product, but not spend a ton of time finding the perfect way to do it. We built the first version knowing that the product will evolve as we learn more.

With an iterative process in mind, we landed on building the following views:

  1. Bunched and gapped departures for all routes
  2. Bunched and gapped departures for all stops by route
  3. Bunched and gapped departures for all stops by time
  4. Custom headway rule settings
A perfect headway is when all buses on the same route heading in the same direction are evenly spaced. An imperfect headway is when buses are bunched and gapped, leading to potentially long wait times for riders.
Step 5: Put ideas to paper

Once the product team scopes what will be included, it’s time to start visualizing solutions and thinking through how users can interact with it. This helps us know whether it’s feasible, and allows developers to weigh on the technical lift.

For headways, our data scientist wrote Python scripts to see what story the data told. Using real data was critical in understanding the actual potential usefulness of the visualizations that we were considering.

Our Data Scientist generated visualizations in Python to test the most effective visualizations to surface insights.

Our designer started tackling the complexity of supporting custom formulas. We needed the tool to support multiple formats, while being incredibly easy to read and understand.

Our designer created different iterations of a filter that would support dynamic headway parameters.

We worked with front-end engineers to understand what was technically feasible and back-end engineers to understand if our data infrastructure supported the types of metrics we wanted to show.

All in all, this step demands cross-team collaboration.

Step 6: Validate solutions are heading in the right direction

Testing early and often ensures that we head in the right direction with little cost.

We are lucky that Swiftly is filled with employees who come from a transit background. Our Growth and Customer Success teams are also on the front line, tethered to our customers needs. Designers can quickly get valuable feedback quickly just by tapping a coworker on the shoulder.

After designs go through the wringer internally, we show updated mockups and prototypes to agencies. We keep our insights organized in a spreadsheet to ensure thoroughness and easy collaboration. After rounds of testing for Headway Insights, we found that some data visualizations easily surfaced key issues, while others were difficult to understand.

We keep findings organized in a spreadsheet, similar to a Rainbow Chart format.

At Swiftly, we’re pushing beyond what exists today. We’re re-envisioning ways agencies can use their data to uncover insights. This requires many rounds of testing and iterating to ensure that we are getting it right.

Step 7: Design to development

Once we have collected internal and external feedback and iterated to get to the final solution, it’s time to move to high-fidelity designs and prototypes for development.

Since engineering is involved early in the process and weighing in on technical lift during initial sketches, nothing in this stage should come as a surprise.

Headway interactions were documented in InVision and Principle, final designs were imported into Zeplin, and design and developers worked asynchronously in a Pivotal board. We also met regularly in person for design reviews, discussing edge cases and expected interactions.

This is our Zeplin board, which contains final mocks that are development ready.
Step 8: Get our product into our users’ hands

After our product is developed and thoroughly tested, marketing and sales step in as a crucial part in highlighting its value and bringing it to our users.

In preparation for the external launch of Headway Insights, our marketing team collected powerful user stories and shared thought pieces on how headway analytics would empower agencies to focus more on the rider experience. Our sales teams dug into agency data and showcased new insights on their high-frequency routes.

A healthy amount of push and pull

Our process ensures that we have healthy tension throughout the development process. Each team — data, design, development — pushes boundaries and organically constrains one another to ensure that we are building innovative, stable, and accurate solutions.

Because in the end, building a solid product doesn’t just simplify a transit planner’s job. It makes sure that millions of people around the world have access to convenient and reliable transit.

You might also like

May 6, 2024

Miami-Dade’s network redesign excels with iterative approach and proactive communication

Miami-Dade Transit recently completed the launch of their Better Bus Network, the biggest network overhaul in county history.
April 25, 2024

Swiftly Run-Times data now available right in Optibus

Our new integration with Optibus puts Swiftly’s best-in-class historical time data right in the Optibus Performance Suite.
April 17, 2024

X-trafik in Sweden increases on-time performance by 5% with Swiftly

Since 2020, X-trafik has partnered with Swiftly to improve service reliability—working towards increasing overall on-time performance to 73%.
No items found.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.