Behind the scenes with MBTA data.

In this post, we investigate the occurrence of people traveling in groups using a single CharlieCard or CharlieTicket.

Occasionally, when people navigate transit in larger groups made up of family, friends, colleagues, etc., one rider will use their individual fare card to tap in and pay for the other people that they are traveling with. Both CharlieCards and CharlieTickets can be used by multiple different people to pay for fares at the same faregate or farebox.

Imagine a CharlieCard to be a wallet with two compartments: the first compartment holds a pass and the second compartment holds stored cash value. When a card with both a pass and stored value is tapped against a fare box or a fare gate, the first tap is processed by the pass compartment and the subsequent taps from other group members are processed by the stored value compartment. While both a CharlieCard and CharlieTicket can be used for pass-back, one advantage of using a CharlieCard is that up to four pass-back taps are tracked to provide transfer discounts. This feature is not available with CharlieTickets.

Pass-back Overview

Group travel occurs for various different reasons: e.g., leisure trips, school field trips, during sports events, or general tourist travel. By studying pass-back situations, we can identify group travel patterns and bus routes and Rapid Transit stations at which group travel occurs. We performed this analysis in order to identify where people travel in groups and how many people relied on this functionality. The results will inform how we can best serve these trips when this functionality is eliminated as part of our new fare collection system (AFC 2.0). As part of AFC 2.0 we plan to allow all-door boarding and move to a proof of payment fare verification system. This will require everyone to have their own fare card or other method of payment (smartphone, etc.). 


In the following sections, we will provide some context and data about pass-back use on first a yearly, and then monthly, basis. We first took a broad look at 2017 pass-back use before diving more specifically into October 2017. We’ll explain how the findings of our analysis allowed us to ascertain answers to the following questions:

  • When and where is pass-back used the most?
  • Who uses pass-back?
  • How many people relied on the one fare card when traveling in groups?

To conduct this analysis, we queried our ODX database to select all transactions during October 2017, for all users except employees and contractors. Multiple taps at the same location on the same card or ticket within 10 minutes of each other (excluding the initial tap) were identified and tagged as pass-back taps. We then calculated pass-back rate separately for bus routes and Rapid Transit stations. This analysis helped us identify the share of all taps used for pass-back and the share of all CharlieCards/Tickets used for pass-back.

But first, let’s look at 2017…

Pass-back in the Year 2017

In the year 2017, 1.48% of all taps were pass-back taps. Ninety-three percent of these taps were on Cards/Tickets with stored value and 7% were on Cards/Tickets with a pass. The second and the third quarter accounted for 59% of the total pass-back rate for the year. The below chart shows the month-to-month use of pass-back across bus and Rapid Transit.

We observed that July was the month with the highest pass-back rate of 2.12% and that pass-back was more likely to occur during the warmer months compared to colder ones. This is probably due to higher numbers of tourists using the MBTA during the spring and summer for events like baseball games and festivals. 

The above chart shows the month-to-month use of stored value and pass-back on stored value across bus and Rapid Transit. You can see that while more stored value is used during the summer, there is also a higher rate of pass-back use.

This observation helped us identify the months during which group travel occurs most frequently. With the confirmation that the spring and summer have higher likelihoods of people navigating our system together in groups, we can improve our operations and customer service geared towards group travel.

In the following sections, we’ll perform a deeper dive into the data gleaned from our October 2017 pass-back analysis. 

When and Where is Pass-back Used the Most?


When looking at specific days of the week, we divided the system’s service time into five categories, as listed below. 

Category Time Time Period
1 03:00 - 06:59 Sunrise and Early AM
2 07:00 - 08:59 AM Peak
3 09:00 - 15:59 Midday Base and Midday School
4 16:00 - 18:39 PM Peak
5 18:30 - 02:59 Evening, Late Evening, and Night

By splitting service hours into these five categories, we were able to identify which specific times of day had the highest number of pass-backs. The bar graph below shows the pass-back rate per day and per time category. 

From the above chart, we can see that on weekdays the pass-back rate was higher between 9 AM and 03:59 PM (time period 3) in comparison to other times of day. Additionally, the pass-back rate was higher during weekends than on weekdays. 

It is apparent that people are more likely to travel in groups on the weekends than weekdays, perhaps for social and/or leisure outings. 


In order to identify specific locations where higher rates of pass-back is occurring in the system, we examined buses at the route level and Rapid Transit at the station level. The following two charts show the top routes and stations ranked by the pass-back rate. 

Route 435 (Liberty Tree Mall - Central Square, Lynn or Neptune Towers via Peabody Square) had the highest pass-back rate of 2.78%.

Riverside station had the highest pass-back rate of 4.87%. The top three Rapid Transit stations with the highest pass-back rates are Riverside, Museum of Fine Arts, and Science Park. 

From the above analysis, we noticed that for bus routes, the highest pass-back rates were in the North Shore (Lynn and Salem) and at shopping centers. For rail stations, the highest pass-back rates were at locations known for attracting tourists, short-term visitors, and schools. 

The high usage on bus routes in Lynn and Salem might indicate a need for more CharlieCard distribution in these areas. We are currently working to establish community partnerships to give people more access to blank CharlieCards. 

Who Used Pass-back?

There were about 0.7 million Charlie cards used by Adult users in the month of October 2017 and 12.57% of them were used for pass-back. This was high compared to other fare media (for Adult users). 

We then looked at the number of rides the adult users took in the month of October 2017 and the percentage of trips that used pass-back. The below chart suggests that infrequent adult riders or people who take one to five rides in a month are the ones who most often used pass-back.

People who traveled on the rapid transit or bus network two times in a month, used pass-back on 10.75% of their trips. This led us to identify the percentage of cards using pass-back for each adult rider category. 

Type of Riders Adult User Type Category (Rides) Percentage of Cards Using Pass-back
Infrequent 1-5 rides per month 12.91%
Occasional 6-20 rides per month 16.32%
Frequent 20+ rides per month 9.31%

We categorized adult user ridership based on the number of rides they took during the month of October 2017 and found that 12.91% of infrequent rider CharlieCards were used for pass-back while more than 9% of frequent rider CharlieCards were used for pass-back. 

The above chart suggests that the percentage of trips taken using pass-back was high for infrequent riders, but the above table suggests the occasional and frequent rider cards were also used for pass-back.

The analysis in the above section will help us prepare for AFC 2.0. Infrequent-use cards are likely using pass-back more often because the people they are traveling with do not have cards. In AFC 2.0, we will make cards more available by having them available at vending machines, but also will provide the capability to pay with a device such as a smartphone, which should lessen the need for multiple people to share one type of payment media.

When Someone Uses Pass-back, How Many Other People Do They Tap In?

(How many people relied on the one fare card when traveling in groups?)

We analyzed the number of times people performed a pass-back and how many other people were tapped in with one card. We found that in the 87% of trips where pass-back occurred, one other person was tapped in.

We concluded that people are more likely to travel with one other person than with a whole group. This pattern will inform the way we think about possible group-specific fare products in the future; we should keep in mind that “groups” are often no larger than two riders. 


Our comprehensive analysis of the pass-back data had many interesting takeaways. The study revealed that the pass-back rate is generally higher during the warmer months and at rapid transit stations used by tourists and people attending sporting events. The high usage of pass-back on bus routes on the North Shore indicates that we need to improve CharlieCard distribution in these areas now. 

Approximately 13% of Adult riders with CharlieCards performed a pass-back. And pass-back is used more often by infrequent riders i.e., the percentage of trips taken by infrequent riders using pass-back is higher compared to frequent riders. In advance of AFC 2.0, this analysis is helping us prioritize locations for fare vending machines and consider fare products for infrequent riders.  

This type of analysis of existing AFC system data is very useful as we consider the changes that will occur with AFC 2.0. It is informing the development of the policies that will support the new fare collection system and ensure that it serves the needs of all types of riders.    


Read the below for an explanation of the methodology the MBTA’s Service Planning Department used to estimate the impacts on riders of the 47 Better Bus Project proposals affecting 63 bus routes.

This post has three parts; this is part 2.

Part 1

  • Methodology Overview
  • Estimating Passenger Travel Time Impacts
  • Estimating Change in Frequency and Passenger Wait Times

Part 2

  • Estimating Passenger Walk Time Impacts
  • Estimating Passenger Transfer Time Impacts
  • Estimating Stranded Passengers
  • Estimating Impacts Across the Day

Part 3

  • Estimating Ridership Impacts
  • Combining All Time and Ridership Impacts to Calculate Net Impacts
  • Conclusion

For more on the Better Bus Project and to learn how you can give feedback, see our project page.

Walk time impacts are estimated for riders at the stops that are no longer served because of the proposal. We used Google to calculate walk times from the eliminated stop to the nearest remaining stop. For example, in the Route 1 proposal to eliminate service around Harvard Square, we obtained a walk time of 4 minutes from the stop at Quincy Street at Broadway opposite Fogg Museum to Massachusetts Avenue at Holyoke Street. Walk times are calculated for distances up to a half-mile where sidewalks or other pedestrian facilities are available. 

Walk directions from potentially eliminated stop to replacement stop

We consider a transfer impact when a stop is eliminated from one route’s service but is served by other routes. We assume that riders could take another route to make the same trip, but only if the remaining route structure and riders’ destinations are compatible.

Sometimes the remaining service does not share stops or connect to other routes that serve the same areas as the current service might.  In those cases, we do not assume that riders will be diverted.  For example, the proposal to operate the portion of Route 411 between Kennedy Drive and the Jack Satter House only in the midday leaves existing Route 411 riders in eastern Revere without even a two-seat ride to get to Salem Street or Malden Station. This portion of Route 411 is walkable from several other routes, but riders are not expected to divert to make the same trips. The only portion of Route 411 that is considered for rider diversion is at stops shared with or near Route 119 or Route 426, since these routes travel to Linden Square where a rider could transfer to Route 108 to Salem Street or Malden Station. 

Map showing proposed service changes on route 411

In the proposal for Route 89, which eliminates the variation to Clarendon Hill, we assume that the riders along this section travel to/from Broadway. While some riders along the eliminated section can walk to Broadway and pick up the remaining service, riders beyond a half-mile of Broadway are assumed to take Route 87 or Route 88 to Davis Station and then transfer to Route 89.

Map showing proposed changes on route 89

In the cases where a destination can be assumed, the transfer time comprises several factors: wait time for the remaining route or routes, travel time to the transfer point, and wait time for the revised route. In the Route 89 proposal, the transfer time for riders along the eliminated section of Broadway beyond a half-mile from the revised route is one-half the combined headway for Routes 87 and 88 plus the travel time to Davis Station plus one-half the headway for Route 89. The net transfer time impact is the difference between the transfer time and one-half the former Route 89 headway at Clarendon Hill.

"Stranded" passengers are defined as any passengers who board or exit at a stop today, but will be beyond a half-mile from service if a proposal is put in place. For example, for Route 411, this counts riders who board or exit the bus between Jack Satter House and Bell Circle (on the East side of the map above).  For Route 89, there are no stranded riders because all remaining riders can walk or use Routes 87 or 88 to access Route 89.

We used the methodologies described above to estimate the per-trip change in travel, wait, walk and transfer times. The next step is to estimate these changes across the day. The level of time aggregation varies by the metric. For travel time, our APC query summarizes data by hour. Walk times are assumed to be consistent across the entire day. For wait and transfer times, we summarize into the following time periods:

  • Early AM: before 7:00 AM
  • AM Peak: 7:00 AM to 9:00 AM
  • Midday: 9:00 AM to 4:00 PM
  • PM Peak: 4:00 PM to 6:30 PM
  • Evening: after 6:30 PM

Using the changes in components of passenger travel time, we apply each portion to the count of riders who would experience that change, as identified from our APC ridership counts by stop.  Below is an example of the data presented for an inbound 7:16 AM trip on Route 1 from our 2017 data.

Estimating total ridership changes along route from all changes

The proposal for Route 1 omits the loop around Harvard Square and instead turns left on Dunster Street. The proposal also consolidates Routes 1 and CT1 into a single Route 1.

So, for the example 7:16 AM inbound Route 1 trip, the count of riders with a travel time impact is the sum of ons at Massachusetts Avenue at Holyoke Street and Massachusetts Avenue at Johnston Gate, or 7.1 passengers, minus the sum of offs along Quincy St, or 0.1 passengers, for a total 7.0 passengers.  We assume that passengers boarding at Johnston Gate receive the travel time benefit but passengers boarding along Quincy St. do not receive the travel time benefit because they are at the end of the loop.

Regarding wait time, there are two groups of riders with different impacts—those served by both Route 1/Route CT1 and those served only by Route 1. For riders served by both Route 1/Route CT1, the wait time change is one half the difference of the combined effective Route 1/CT1 headway and the new Route 1 headway. For the riders at stops served only by Route 1, their wait time change is one half the difference between the Route 1 headway based on 90th-percentile cycle times and the new Route 1 headway with the added reinvestment from Route CT1. Note that we only consider boardings so that we count the number of riders being impacted once.

Walk time is calculated for the three stops around Harvard Square – 3 minutes, 6 minutes, and 2 minutes, at Johnston Gate and the Quincy St stops, respectively – and applied to the ridership at these stops. Walk time is calculated for both boardings and alightings because it reflects the additional walk time passengers would incur on either end.

Transfer time impact does not apply for the Route 1/CT1 proposal, so we will use a different example.  First, we identify stops where riders would divert onto other service. Then the impact is applied to either boardings, alightings or both depending on the direction of travel. For example, in the Route 89 proposal to eliminate service to Clarendon Hill, the transfer impact is only calculated for inbound boardings and outbound alightings along the eliminated section. The impact for inbound alightings or outbound boardings along this section is already captured by the walk time impact calculation.

The following table summarizes the count of riders, their time savings/cost, and the total number of minutes saved/cost for all affected passengers by impact type for the Route 1 example trip above.  Negative numbers represent a savings, and positive numbers represent a cost. With travel time savings of 35.7 minutes, wait time savings of 31.4 minutes, and walk time cost of 19.5 minutes, the net passenger-time impact is a savings of 47.6 minutes for this trip.

Impact Type Description Passenger Count Savings / Cost per Trip (minutes) Total Minutes Savings / Cost
Travel Time Riders boarding at Harvard minus those alighting around Harvard Square loop 7.0 -5.1 -35.7
Wait Time Route 1 segments not shared with Route CT1 20.9 -1.5 -31.4
Wait Time Route 1 segments shared with Route CT1 45.1 0.0 0.0
Walk Time Riders boarding and alighting around Harvard Square loop 5.3 Varies by Stop 19.5
Sum       -47.6

We estimate the aggregate daily passenger-time change by summing the impacts and riders affected by these impacts for each trip. In our proposal summaries, we also estimate the time impacts by impact type. Since the time impacts vary by trip depending on time of day, we typically provide the median value in the summary document. 

Next: the final post in this series, discussing how we put everything together to estimate the impacts on ridership.