Network Redesign

Bogota: The Price of Service Complexity

Posted on March 11, 2026 in Bogotá, General, Network Redesign 4

If you look for a map of Bogotá’s TransMilenio network, you’ll find something like this:

Diagram of TransMilenio from the City of Bogotá website, published by Jose Luis Martinez.

Note that north is to the left on most standard Bogotá maps.

This is not, of course, a map of the whole network. It barely even begins to explain what the buses do. In fact, what the buses do — even the biggest buses providing high-capacity transit — is so complicated that it may impossible to map clearly.

This map invokes a familiar stylistic language from metro maps worldwide: A solid line of a certain color means direct service between all the dots on that line. Where two lines cross at the same station dot, that means you can connect between those services there. The style is so pervasive and effective that some cities without much rail service draw their bus services this way. (Here’s the bus map in Wellington, New Zealand, for example). Just down the hill from Bogotá, the Medellín Metro system uses this familiar style to show how their rail, bus rapid transit, and gondolas all interact.

But Bogotá’s map doesn’t work that way. No logical metro network would have lines like B, A, and H as they appear on this map, meeting end to end. In the usual metro map language this would mean you have to transfer twice to keep going in the same direction, which would be silly. But that’s not what it means here.

This is a map of branded corridors of buses, a concept that takes some explaining. The green line B on the map means that along this corridor, a bunch of buses whose line numbers start with B all operate. But those buses are coming from points all across the network, bearing their B number.

Here’s the explanatory sign of a single station, Virrey on the B segment. (Click any photo to enlarge and sharpen.)

This sign lists every bus route that stops at this station, with a list of the other stations that each route serves. The colors match the map. We’re on the green “B” segment so sure enough, there a lot of green B bus routes. But those buses are nowhere near the majority of all the buses that stop here. In fact, if you’re traveling between two B stations, the next bus that does that may not be a B bus at all. What’s more, the next B bus may not take you to the B station you want, because of their complicated skip-stop patterns. So you really have to check that the bus you get on is stopping exactly where you want to stop.

Why so many different patterns? This is the compounded effect of two kinds of complexity:

They are trying to run direct service from this corridor to every other corridor in the network, so people don’t have to transfer.
Not all buses stop at all stations, so stopping patterns generate separate routes even among buses that are all following the same path.

Notice that both of these effects are multiplicative rather than additive. If you wanted direct service between any two or the 12 corridors, that in itself would generate 72 possible routes (12 squared is 144 one-way pairings, which is 72 two-way routes). Then when you start varying stopping patterns you continue to multiply routes. Remember, more routes always means less frequency on each one. It means more time standing in a crowded station before you can start moving toward your destination.

In the lower left of the sign above, there are also six bus routes numbered 1-8, with headings in black. These were added later. These buses are distinctive because they stop at every station, so they provide a simpler service pattern. (It would have been nice to have a map of at least these eight routes, which are clearly simple enough that such a map would be legible.) But these routes are a small part of the network. Most of the great masses of red buses that you see in the TransMilenio corridors are doing one of the more complicated patterns.

Most of these patterns run with a frequency of 5-10 minutes, but of course there are disruptions, especially in routes that run partly outside the infrastructure, so I often waited longer. Ten minutes is a reasonable wait in the transit-starved United States, but on TransMilenio, with many buses per minute flowing past each station, and given the discomfort of most of the stations, it feels like a very long time, especially if the complexity is also heightening your anxiety.

So let’s review all the ways that this complexity affects the rider and the experience of TransMilenio as a whole:

It requires longer waits. In simpler systems, there are fewer service patterns but they run more frequently, so people are more likely to get on the next bus that comes and make progress sooner on their journey, even if that bus doesn’t take them all the way to their destination.
This means that stations are more crowded, as more people wait for a particular bus instead of getting on the next one.
It makes buses more crowded because they are more unevenly loaded, for the same reason. This is likely the cause of the difference between TransMilenio’s “theoretical maximum” capacity (56,000 people per peak hour per direction) and the highest actually observed volume (43,000).
It makes the station wayfinding challenge very difficult, because people have to be navigated to a particular platform within the station, and the huge number of routes means that the user must wade through a lot of information (see sign above).
Most importantly, it trains people to either (a) rely on an app to navigate them or (b) use only services that they know. Both options also tend to steer people away from other services that might be more useful to them for the trip they’re making.

The complexity is staggering enough if you travel within the TransMilenio busways, but once you’re outside of them, it’s truly overwhelming. Here’s a bus stop sign in a very dense inner city neighborhood.

That’s a lot of flavors of buses, all going down the same street in the same general direction. You already know that route numbers beginning with A, B, and C mean that those buses will eventually end up running on those corridors of the busway, though it’s anyone’s guess whether that segment is in the future or the past relative to this sign. And there are so many other routes on top of each other. This is what the network looks like at street level, stripped of the relative clarity of the big busway stations. There is information here — Bogotá’s numbered grid system means that a practiced rider can decode much of it — but the cognitive load is immense, and for a newcomer it is simply a wall of text.

To TransMilenio’s credit, they’ve done all they can on a static sign to explain these bus routes with text. Bogotá is blessed with a grid system of numbered streets, called carreras when they run north-south and calles when they run east west. This allows routes to be described very briefly, because locals learn the code: K for carrera and C for calle. (A for Avenida just means that the calle or carrera is a major street.) So there is a great deal of information here, which I quickly learned to make the most of. But it’s still a lot of complexity.

Later, I attempted a trip from K11 & C92 to K69 & C24. The journey required a transfer, which was fine with me, but Google told me to transfer at a point where there was only one bus route continuing to my destination. As I waited 20 minute for that bus, countless other buses flowed past me, each one maybe taking me partway to my destination, and to a place where there might be more buses to where I was going. But there was no way to make sense of this information.

As I understand, the local app TransMiApp will give you directions based on where the buses actually are, so that for example if my direct bus were late it would serve up a faster alternative involving a transfer. But the right answer would continue to change by the minute during my journey as various buses ran on time or not. This would have helped, but not nearly as much as having a simpler system of fewer routes, all running much more frequently. There are enough buses in Bogotá so that on almost every street, the next bus could always be coming within a minute or two, ready to take people on the first stage of their journey. But that would require a different approach to design, one that is less fearful of transfers.

Let’s acknowledge that Bogotá can be critiqued on this score only because:

They have unified planning control of the whole network, so they theoretically have the power to make things simpler.
They have at least attempted to explain the service, which makes it possible to see what’s wrong.

In many middle-wealth countries, and certainly in poorer ones, you won’t see any effort to make the service legible, so the complexity is overwhelming that it’s hard to even talk about. In many countries, too, private operating companies still control service planning and do not feel motivated to participate in creating a logical total network. If I seem to be picking on Bogotá here, it’s only because unlike so many of its peers, its public transport authority is at least trying to present a network, and it has enough planning authority that it can be held responsible for that network’s design.

Still, if I had been working on this network, I would have questioned how much of this complexity was truly necessary. I would especially have asked:

How much of this complexity is needed only in the peak? One of the most common mistakes in transit planning is to design the all-day network around the rush-hour pattern of demand. Instead, the all-day network should always be designed around the all-day pattern of demand, with rush-hour patterns added on top only as demand warrants. That can help you create a more legible all-the-time network — maybe even one simple enough to draw a diagram of — even if rush hour service is adding a lot of complexity. In our firm‘s maps, we draw rush hour services very faintly, if at all, so that they don’t distract from the simplicity of the all-the-time pattern.
Where in the network is the infrastructure for changing buses so good that there is no reason to insist on providing a single-seat ride? There are already several convenient interchange stations in the network. Asking people to change buses is always the key to a simpler and more frequent network.
Can we make simplifying assumptions about stopping patterns to reduce the number of service patterns? This massively complex system has millions of details that can be fine-tuned to demand, but what is the cost/benefit ratio of keeping track of all those details. Would a simpler system that’s easier to understand produce benefits that outweigh that?

There’s one easy response to these questions: “People hate to transfer!” In Bogotá the decision to avoid transfers is based on surveys in which people are asked if they want to transfer. But these surveys asked a hypothetical question without explaining consequences. It’s always important to ask: “Which matters more to you? Avoiding a transfer or getting to your destination sooner?” Because that is often the real choice.

Toronto: What Eglinton Means

Posted on February 9, 2026 in General, Network Redesign, Toronto 25

It’s been a long struggle, but Toronto’s Eglinton rail line is finally open. Now, the rapid transit network consists of the wide lines on this map, numbered 1-6.

New rail map published by the Toronto Transit Commission, showing the new Eglinton line as Line 5. Please ignore the narrow lines numbered in the 500s. They are streetcars, and many of them are stuck in traffic and slower than buses.

The internet is full of reviews and recriminations about how long Eglinton took and the pros and cons of its design, but I want to make a different point: Eglinton shows a city slowly awakening to its true nature as a grid. This awakening is ultimately more important for the mobility and freedom of people in Toronto than details of what’s on rails and what’s on tires.

Toronto, like Chicago, is a grid city with a very uneven distribution of density. The city has a large grid of arterials that are more or less c0ntinuous across the city, but density is historically concentrated downtown and to some degree along the oldest rail transit corridors, especially Yonge (the east leg of Line 1).

So for a long time it wasn’t crazy for Toronto, like Chicago, to have a radial rail network (designed mostly for trips downtown) fed by buses that followed the arterial grid.

But it’s crazy now. Working from home has reduced the rush of commuters into the downtown towers. Meanwhile, many more towers are being built in the suburbs, often just over the line outside Toronto. And as in every city, the preponderance of travel demand is now much more all-day, all-week and all-direction than it used to be. Finally, lower-income people have always been less likely to be going downtown, and more likely to be accessing lower-wage jobs that are scattered all over the city. So if we care about them, we have to stop building and planning as though downtown is all that really matters.

So Toronto needs a network that people can use to go anywhere at any time.

The downtown-oriented network is not good at this, because by feeding toward downtown in a hierarchical pattern, it makes it hard to travel in non-downtown directions. In short, Toronto needs a grid transit network to match its grid geography, so that people can go from anywhere to anywhere in a simple L-shaped trip, usually with a single transfer.

Eglinton is the first major investment that shows a stirring of that consciousness. It lies well north of downtown, and runs east-west most of the way across the city. For the first time, it’s possible to travel a long distance east-west without transferring and without going downtown, which means that trips to many other places become possible with one transfer instead of two or three.

But the old consciousness is still on display further north on the map above, where the recent Finch West line (Line 6) has opened. It’s very much the old model of a hierarchical feeder, running from the west edge of the city only to the first subway connection. It’s now a mirror image of the equally problematic Line 4, which serves only a fragment of Sheppard on the east side. Both of these lines are largely useless for the kinds of trips that will increasingly prevail in a multi-dimensional city, because they interrupt the logic of the grid instead of working with it. My advice to Toronto years ago was not to build Finch West until they were ready to continue it all the way east at least to Yonge, the east leg of Line 1, so that it could also be used to travel to points on Yonge including northward into the dense suburbs. It will be interesting to see how useful the newly opened fragment turns out to be.

The buses, too, still display the old consciousness. Almost without exception, long east-west lines terminate at Yonge, requiring the passenger to transfer to continue in the same direction on the same street. Again, this means they will often transfer twice to complete an L-shaped trip to an actual non-downtown destination:

A segment of the Toronto bus map along Yonge Avenue from the north edge of downtown to the north edge of the city. Note that every east-west line is interrupted here. Eglinton rail (orange) is the only exception.

The situation is even more remarkable along the Bloor-Danforth subway Line 2, where again many bus routes are interrupted. Toronto is about twice as big east-west as north-south, so these corridors are shorter, which means there’s even less reason to interrupt them:

The west end of Line 2. Note that every north-south bus line is interrupted as it crosses the rail line, adding many unnecessary transfers to L-shaped trips that do not use the rail.

These designs reflect an obsolete notion of hierarchy in which rail is king and buses are servants, existing only to “feed” the rail. In an era when downtown commuters mattered more than anyone, that made a bit of sense, but only a bit. There were always people trying to go everywhere and struggling against these patterns. Now, buses need to be rethought to make sense on their own terms, as parts of complete grid networks. (Our firm brought the same awareness to the upcoming redesign of Atlanta’s MARTA bus network, where for the first time, many bus routes will continue past rail stations instead of ending there)

So amid all the other debates about Toronto, let’s celebrate the most important thing about Eglinton: For almost the first time, the city has made a major investment that recognizes that in a big, dense city, surrounded by dense suburbs, people need to go all directions, all the time.

Portland: Dire Transit Service Cuts Planned

Posted on January 28, 2026 in Cuts, Network Redesign, Portland 30

Very bad transit service cuts are coming to Portland.

Just a couple of years ago, we worked with Portland’s TriMet to develop an ambitious service expansion plan called Forward Together. Now, the agency is saying instead that they are facing a dire financial shortfall and need to make service cuts. I’m not sure why this message has changed so suddenly, apart from the failure of the state legislature to provide a new funding source for operations. In any case, the agency’s current position is that they have to cut service now to avoid worse cuts later, although worse cuts may be coming later anyway.

You can peruse the cuts here. If you live in the region, you should comment before Saturday, January 31.

I have several thoughts, which are further below, but it’s best to start by looking closely at the single most shocking cut they propose.

Abandoning a Major Hospital?

This may well be the first time that a US transit agency has proposed to abandon all service to a major medical center: Providence in inner northeast Portland. It happens to be a landscape I know well. I live nearby, grew up even closer, and go to the Providence complex for most of my healthcare. Also, I first moved to Portland in 1969 (I was 7, but already a transit geek) so I know some useful history.

This part of northeast Portland is a fairly dense area with a good street grid where a lot of housing is being added along frequent transit corridors (red in my sketch, which is based on a Remix plot):

In this inner-city context, most of the blue lines, which currently run every 20-30 minutes, should be every 15 minutes, because they can perform well as part of a frequent grid. But in fact only the red lines are frequent. There’s also the sinuous light rail line, following I-84, but it only has two stations in this area, at 42nd and 60th. Do not ask me why there is no station at Providence, whose campus lies between 47th and 52nd right next to the rail line.

Now, TriMet proposes to entirely delete Line 19 on Glisan. This would remove all service within 1/4 mile walk of the Providence complex.

I was an intern at TriMet around the time of the 1982 redesign that created the current grid. At the time, all the lines on this map were made frequent (every 15 minutes or better) except for Glisan. The network is designed mostly around the principle of 1/2 mile spacing between lines, because many people will walk up to 1/4 mile to useful service. Glisan is only 1/4 mi from Burnside, too close for both to be frequent (at least in the context of the scarcity of service that is typical in the US).

Glisan is a mixed bag as a transit street. West of Providence, the neighborhood of curving streets is Laurelhurst, an area of low density and hence low demand. But Providence itself is a massive complex, a major hospital and a large building of medical clinics where many people come for appointments. Further east, beyond 60th, Glisan is a better transit street than Burnside for a while: it has a grocery store and apartments in this segment, while Burnside is climbing over the north shoulder of Mount Tabor, which limits development potential there. But long ago the decision was made that Burnside, not Glisan, would be the Frequent Service street, where TriMet would protect 15 minute frequencies at most times of day.

Now, if TriMet has to remove the 19, all the options are truly hideous. Abandon Providence entirely, along with the moderate income area and moderately dense area along Glisan east of 60th? Deviate the Halsey line (77) down to Glisan, just between 47th and 60th, to touch Providence, making it longer and thus less useful for through travel? Deviate the Burnside line up to Glisan, violating the principle that Frequent Service lines should aim for permanence since they’ve been used as the basis of dense housing development (including some on the segment that we would miss if we deviated in this area)?

I don’t know what TriMet will do. I don’t know what I’d recommend, except to say that a city with Portland’s pretensions to sustainability should not be in this position.

The Overall Design of the Proposed Cuts

The fact that service is being cut is a financial decision out of the control of TriMet’s service planners. Given the direction to make harmful cuts, I think they’ve done a good job in minimizing the harm. Some things I especially respect are:

Sharing the pain with the light rail network. Until 10-20 years ago, many agencies would have started this process from the assumption that the rail service is special and must be protected, leading to even more destructive cuts to bus service. Instead, TriMet proposes to cut back the Green Line to just its unique segment south of Gateway, where it would operate as a feeder to the Red and Blue lines. This is a frequency gut cut all along the east-west segment now served by Red, Blue and Green (from 5 min to 7.5 min) but it’s also a cut to north-south frequency along the transit mall in the heart of downtown, from 7.5 to 15 min. A 15 minute frequency is really not relevant to internal circulation in a downtown, serving trips of under 2 miles, and the whole design of the transit mall (as redone in 2008) presumes that rail, not buses, serves this circulator function. Now that won’t work at all. So yes, a terrible cut at the heart of the system. But light rail operating costs are high, and if they didn’t cut light rail they’d have to utterly devastate the bus network.

Some service designs that are improvements. Our Forward Together project included many ideas that have been carried forward here, though without enough frequency. (Continuous service the whole length of Woodstock Blvd, for example.)
Balanced removals of coverage. The principle of the Forward Together project, as endorsed by the agency’s Board, was that service needed to be justified by either ridership or equity. That means that low-ridership service can be offered only where it responds to a demonstrated social or economic need. As part of the Forward Together plan, TriMet has already deleted low-ridership “coverage” services in relatively affluent parts of the region, and they continue to do so in this proposal.

But overall, the plan’s impacts are dire.

What’s more, there’s a serious risk that in the public outreach process happening now, more people will defend the deleted routes than defend the Frequent Service network. This could pressure TriMet to cut frequencies on this backbone of the region. We have already done this experiment: In the 2009 financial crisis TriMet cut Frequent Network frequencies from 15 minutes to 16-17 and triggered a dramatic loss of ridership. Frequency is never visible enough on the map, which makes it hard to defend when people are complaining about losing all of their service, yet frequency is the key to ridership. This is the eternal ridership-coverage tradeoff.)

Do We Really Want to Do This?

Oregon’s legislature recently went through a spectacular failed effort to pass a statewide transportation funding measure, where rural legislators demanded maintenance for their roads but were eager to strip out transit operating funds for cities. A measure passed that funds transit only through 2028, but that has been referred to the voters, with an election scheduled for May.

It seems likely that the best we can hope for from the state is a short-term rescue. Leaders in the region — probably working through Metro or the City of Portland — are going to have to step up if they want to save what was once one of America’s most admirable transit agencies.

How do Network Redesigns Increase Ridership?

Posted on November 10, 2025 in Network Redesign

The last few posts, starting here, reviewed some recent US network redesigns at our firm, all of which increased ridership above the rate at which they increased service. In other words, they increased productivity, the ratio of ridership divided by the quantity of service provided. They also increased productivity faster than it was rising overall in peer systems.

What can these examples tell us? The same thing that we know from our many other ridership-increasing redesigns:

Think about the ridership-coverage tradeoff. Virtually any transit agency could increase ridership by abandoning areas whose development pattern is unfavorable to transit and focusing more service where the pattern is more favorable. But as government agencies accountable to elected officials, they usually aren’t being told to do this. Coverage service — which means service run in low demand areas knowing that ridership will be low — is very popular and fiercely defended. That’s why an effective plan involves clear decisions about the unavoidable tradeoff between ridership goals and coverage goals. (No, microtransit does not get you around this problem.) This requires a public conversation and a clear decision by the Board, and to get to that, you must …
Include everyone in a reality-based conversation. What is critical to the success of these networks and projects is that in each case we worked closely and carefully with the agency, stakeholders, and the community to have a clear conversation about why and how to change the system. Everyone affected could understand why each agency was changing and what the goal of the major change was. But these conversations were reality-based. We created tools to help people see what was mathematically possible, so that they spent less time advocating fantasies and more time thinking about the actual problem facing the agency.
Get the fare rules out of the way. Monterey-Salinas and Santa Cruz both made changing buses free at the same time they implemented the redesign. Charging people to use two buses instead of one is insane. It would be more convenient for our analysis if the fare change and service change had happened at different times, because it makes it hard to sort out the causes of the ridership jump. But in fact, they had to eliminate transfer charges for the redesign to work, because efficient networks tend to require a bit more transfering. Many authorities still have fare rules that penalize transferring, and that therefore require the operation of less efficient and liberating services.
Talk about freedom. We never talk about network redesigns in terms of ridership predictions or modeling, because many people sensibly don’t trust these models, and they’re not really necessary. What matters is that we’re expanding freedom, and that’s what we measure. We’re making it possible for more people to get to more places sooner, so they can do more things in their lives. This turns out to be a great way to increase ridership, but it’s also something that everyone cares about.

At our firm we’ve lost count of the number of network redesigns we’ve led, both in the US and elsewhere. But they have almost all increased ridership, to the extent that was a goal. We’re proud of our work in agencies of all sizes, from small cities with a few buses to giant cities like Houston and Dublin. We have a lot of well-honed instincts and methodologies that make us good to hire for your community, but the real source of what we do is simple and we’re happy to explain it to the world: We have clear conversations about mathematical reality, and help people see all the possibilities that are before them.

Long Island, NY: Ridership Grows after Suffolk County Redesign

Posted on November 5, 2025 in General, Network Redesign

(This series of posts on ridership-increasing bus network redesigns starts here, and its wrap-up is here.)

Suffolk County covers the eastern two-third portion of Long Island. Its vast expanse hosts historic towns built up around Long Island Railroad stations and swathes of suburban development. Around 90% of Suffolk’s population and activity is concentrated in its western third, while the rural eastern end includes the popular resort area known as “the Hamptons“.

Starting in 2020, Scudder Wagg led our firm’s work on an initiative to redesign the bus network for Suffolk County Transit (SCT). The new network we designed was launched in late 2023. JWA also supported SCT in implementing the network: we developed a new detailed system map and a schematic map, revamped their route timetable brochures and bus stop signs, and advised on many other technical elements like website design.

The plan consolidated services to offer much better frequency (at the cost of a reduction in overall coverage). In the old SCT network, only 3 routes had a frequency of every 30 minutes or better, but now 12 corridors (11 full routes and one offset corridor) have service every 30 minutes on weekdays. The plan also invested in reliability improvements and greatly expanded service in the evenings, on Saturdays and on Sundays. With the new network, nearly all routes now run 7 days per week and much later into the evening. All these investments in more useful frequencies and spans (keys to useful, ridership-oriented service) and improving reliability by fixing outdated timetables added up to a 30% increase service.

Here is the SCT network before the redesign (click all images to enlarge and sharpen):

Colors still represent frequency but note that the levels are different in this map: red means service every 30 minutes while deep blue is service up to every hour. A lot of service was concentrated in the much denser western third portion, but it was a jumble of infrequent, uncoordinated routes with many deviations and irregular timetables.

Compare this to today’s SCT network, in the official network map that we designed:

You can see many more red corridors, particularly in the County’s denser western portion. We were able to increase frequencies and provide much more legible service pattern by consolidating service. There are now only two service types: 30-minute service and 60-minute service. This frequency is proclaimed clearly in the new maps, brochures, bus stops, and the website.

However, achieving this much frequency required streamlining many complexities. In the western part of the county, the old network had many infrequent routes close together. The new one has more frequent routes, but further apart. Walking distances are greater but waiting time is less, which produces an overall increase in access to opportunity but many complaints from individuals who have to walk further. We were on the front lines of much of this conversation. In fact, our planners were there helping on the street when the plan was implemented.

Another key feature of the new SCT network is timed transfers (or pulses). At seven locations across the County, buses on several routes arrive together every 30 minutes or every hour. Passengers can transfer between routes with a short wait and be on their way. Ensuring reliable timed transfers is critical to making the new SCT network as useful as possible, given that routes only run every 30 or 60 minutes. This network design strategy is useful where you can’t afford better frequency and you have multiple but dispersed sets of moderately dense places and no dominate central core. However, it requires thinking about the connection points first, and designing the network around them.

With more frequent service on many corridors, and timed connections in the places where the most people would need to transfer, we estimated that the average resident would be able to get to 50% more jobs in 60 minutes and the average low-income resident would see a 60% increase in access to jobs. These outcomes meant not only that transit users would have more choices in their lives; they were also a solid indication of the higher ridership potential of this new network.

The County implemented the new network on October 29, 2023 and the results since have been impressive. Based on data reported to the FTA National Transit Database, ridership increased by about 15% in just the first three months (November 2023 to January 2024) and after 12 months, ridership showed an average of a 25% gain compared to the same month in the prior year (November 2022-October 2023 compared to November 2023-October 2024).

Ridership is now about 39% higher as of July 2025 (comparing the 12-month rolling average ridership in July 2025 versus October 2023) and County staff recently touted the ridership success at a Car Free Day event. Our team at JWA is excited to be continuing to support SCT with various efforts to continue to improve this network including developing updated schedules based on the most recent real-time tracking data and some route adjustments now that we have more data on reliability, ridership, and customer concerns.

Akron: Major Ridership Growth Through Network Redesign

Posted on November 2, 2025 in General, Network Redesign

(This series of posts on ridership-increasing bus network redesigns starts here, and its wrap-up is here.)

During the pandemic, we did a network redesign for Akron’s transit agency METRO. The agency covers all of Summit County, Ohio, the next county south of the Cleveland area. We already knew the neighborhood, having worked on Greater Cleveland RTA’s redesign a few years earlier. Evan Landman was our project manager.

Historically an industrial town, Akron has experienced the familiar rust-belt decline. There is still manufacturing, but it relies on a smaller and more diverse economy, including a university downtown, medical centers, and so on.

The agency had already adopted a Strategic Plan that called for a focus on the busiest corridors, while maintaining connectivity for people with the greatest level of need.

Here is the old network on the left, and the new network that emerged from the project on the right (click on all images to enlarge and sharpen.) Note the legend! As always on our maps, colors mean frequency, and that’s critical.

The old Akron network on the left, the new one (“Stable Scenario” on the right. Click to enlarge and sharpen.

A few things are especially worth noting here:

The old network’s most frequent routes were every 20 minutes, just below the 15 minute threshold where service starts to really become useful. Because most other routes were every 30 minutes, there was no way to coordinate timed connections downtown.
The old network had a downtown circulator that was just too short, and that could be replaced by a portion of a longer, more frequent line.
The south central part of the city, a relatively low-income and diverse area, had a complex tangle of four overlapping routes that all came only once an hour. We replaced these with fewer, more frequent services that overlap each other less, a common way to increase access.
We’re especially proud of what was possible in the vast rural area north of Akron, which borders the Cleveland urban area just off the map to the north. This area had an ineffective tangle of “express” services that ran just a few trips a day, and that were highly specialized around certain employers even though there wasn’t much ridership. The place replaced them with a simple pattern of a half-hourly service all the way from Akron to Southgate (north off the map) which is a major hub for Cleveland RTA services. The half-hourly service splits into two hourly strands in the rural area, to cover more territory, but converges at both ends to provide 30 minute frequency where demand is higher, and to provide 30-minute frequency all the way from Akron to Southgate. Our analysis of outcomes at the time showed that the Reimagined network would be immensely more useful in connecting people in Akron and Summit County to the places they need to go. In 45 minutes or less:
The average county resident would be able to reach 60% more jobs using transit.
The average person of color would be able to reach 88% more jobs using transit.
The average low-income person would be able to reach 103% more jobs (more than double!) using transit.

Major activity centers and destinations, such as Summa Hospital, saw even greater increases in access to people and jobs in the region. The isochrone example below shows how the new network increased the number of people who could reach the hospital in 45 minutes by 77%. These kinds of access gains are the reason we knew that this new network would improve ridership relative to the previous network.

The Metro Board approved the Transit Development Plan in March 2022 and Akron METRO implemented the new network in June 2023. Ridership increased 24% in the first year after the network launched. As of July 2025, ridership was up around 40% compared to the year before the new network. (In our introductory post we cite a different figure because those figures are standardized to the period December 2022 to July 2025, for comparison to other redesigns.)

The Reimagined network was a significant expansion, adding about 27% more service, but it was also a huge step in the direction of being useful to more people. It is much easier to achieve major gains with more service, yet the access outcomes improved far more than the additional investment would suggest, indicating that the new network didn’t just expand service, it drastically improved ridership potential.

On net, the total network productivity (ridership relative to service hours) is up about 9% compared to the year before the new network launched. This is an extraordinary improvement in the context of overall trends in the industry.

The enormous improvements in ridership contributed to Akron METRO winning the 2025 APTA Outstanding Public Transportation System award for systems in the 3 million to 15 million annual riders category and the immense success was recently featured on the CBS Evening News: https://youtu.be/DNZnZJPvytU

We really enjoyed working with CEO Dawn Distler and the whole team at METRO. This is a remarkable story about how much network redesign can improve people’s lives.

Our Bus Network Redesigns Are Increasing Ridership!

Posted on November 2, 2025 in Network Redesign

Amid all the gloomy headlines about transit funding crises, some transit agencies are reporting great news. Our firm has been behind many bus network redesigns that are bringing more people onto public transit, especially in the US. These plans have not just dramatically increased ridership. They’ve even increased the productivity (ridership divided by service provided, or roughly, “bang for buck”). That means more people using each bus, and more people benefiting from expanded access to opportunity. Here are three of our most recent examples. This table compares the average for the 12 months ending June 2025 to the average for the 12 months ending Dec. 2022.

The last row shows how productivity grew nationwide in similar sized systems (all US agencies serving populations of 250,000 to 1.5 million). So clearly, some of the ridership growth might have happened anyway. But our redesigns all came in above that trendline, growing productivity more than the national average for peer systems.

This series of posts will provide deep-dives into some of these recent redesigns. In each case, I don’t want to just brag about our firm’s achievements. I want you to see why these were ridership-increasing projects, and why they contain insights that other agencies can use, especially if they are trying to achieve more with less.

For a long time, I’ve resisted providing high level summaries of ridership changes resulting from our work, for a few reasons:

Ridership is not the only goal of most transit agencies. The competing goal of coverage is important to almost all transit agency boards in the US, so we have never unequivocally pursued ridership as the only goal. We encourage agencies to give us clear direction about what percentage of their service they want to devote to a ridership goal, but that’s never 100%.
Ridership is affected by the network design but also by many other things: fares, service quality, the economy, pandemics, etc. In many projects we’ve done, other things happened at the same time that make it hard to sort out which thing caused what ridership change. For example, Alexandria, Virginia started free fares on the same day they started the new network. Our San Jose, California redesign was implemented a month before the pandemic, so we only got one (promising) month of data.
We are most sure of the impact of a network redesign right after it’s implemented, before too many other things have happened, but we never see the full benefit then. It takes time for people to discover a new network and change their routines. Benefits that arise from better serving existing travel patterns mostly show up within a year, but there are two other benefits of network design that take years to appear: (1). People relocate in response to better service, and (2) actual development may occur in response to the redesign, especially because of policies that permit more density around service of a certain frequency. Those longer-term benefits are even harder to attribute to the network design as opposed to other things, because after more time, more other things have happened that also affect ridership.

So attributing the causes of ridership is always messy. We like to talk instead about expanding access to opportunity, because that explains why a network redesign improves ridership in most cases. So in the series of posts to come, we won’t just brag about our ridership numbers; we’ll look at exactly what each redesign did to produce that outcome, and what difficult tradeoffs were required.

Links to the individual articles are below. This post will be updated as we get more data.

Conclusion: How do Network Designs Increase Ridership?

(Thanks to Alex Boccon-Gibod of JWA for the peer analysis.)

Atlanta: A Draft Redesign of the Bus Network

Posted on December 12, 2024 in Atlanta, General, Network Redesign

After much hard work, intense discussions, extensive public outreach, and a pandemic, we’re finally ready for public comment on the draft MARTA “NextGen” bus network redesign, whjch covers Atlanta and the neighboring cities in Fulton and DeKalb Counties. (Clayton County is also part of MARTA but being addressed by a different process.)

The project website has all kinds of maps and useful information. But if you’re curious, I hope you will delve into our readable full report (download high resolution pdf, or view low resolution pdf). There you’ll find a complete explanation of the thinking that led to the proposal as it stands.

This is the first complete rethinking of the bus network since the MARTA rail system opened 45 years ago. Back then, the core idea of the network was that the purposes of buses was to feed the rail system, producing a network overwhelmingly suited to bringing people into downtown Atlanta. But since then, much has changed about the region and it transit demand:

Many suburban employment and activity centers have developed, some of which are well suited to transit service.
The rise of working from home after Covid-19 has reduced the downtown rush hour commute market.
The need is greater than ever for all-day, all-week, all-direction trips that matter to lower income people. Many of these trips are not going downtown, but to activity and job destinations all over the region.

We have done our best to redesign the network reflecting these changes. Big ideas include:

More lines that run frequently (every 15 minutes or better) all day and on weekends. These are in red on our maps below.
More lines that continue past rail stations instead of ending at them, to connect more destinations with fewer transfers.

Is this all of the service that the service area needs? No, it is what MARTA can afford, given its other commitments and the decisions that have been made about priorities. MARTA directed us to plan for a total service budget that is slightly lower than 2019 though higher than 2023. I wish we could have proposed far more service than this.

Still, within these constraints, the plan achieves some dramatic improvements. Here is the Fall 2023 network on the left, which we used as a baseline, and the proposed “NextGen” network on the right. (These are just diagrams. Much more detailed maps are on the project website and in our report.)

Some key facts:

The average number of jobs reachable in an hour goes up by 21% for all residents, 23% for racial minority residents, and 23% for low income residents. (Why does this matter?)
The number of residents within 1/4 mile of service goes up by 2%.
The number of residents within 1/4 mile of frequent service (every 15 minutes or better all day, shown in red above) goes up by 245%.

I hope you’ll take the time to peruse our friendly report, which talks through the whole thought process and explains how we got to this recommendation. Then, if you live or travel in the MARTA service area, please comment! MARTA is taking comments through February.

Waterford, Ireland: A Plan to Double Public Transport Service

Posted on July 10, 2024 in General, Network Redesign

Waterford’s central bus interchange point. Source: Google Street View

One of the highlights of my career, and of our firm‘s recent history, has been a series of contracts for the National Transport Authority in the Republic of Ireland to redesign the bus networks of all the main cities, part of a larger national strategy called BusConnects. We started in Dublin in 2017, and have been progressing through the other four Irish cities: first Cork, then Galway and Limerick, and now finally Waterford, which has a metro area population of about 83,000. Dublin is now half-implemented, while the rest will be implemented over the next several years.

Our Draft New Network for Waterford was released Monday for public comment. It proposes to improve frequencies; replace one-way loops with two-way lines; add weekend service; and cover new areas. It would double the quantity of bus service in Waterford. I’m not sure we have ever before been asked to plan such a large expansion of service for relatively short-term implementation.

The level of policy clarity behind this expansion is remarkable. Ireland has “done the maths” about what must be done to achieve national goals for emissions and for social welfare, and has concluded that a big investment in urban buses is key to those goals. Ireland has a cost-effective rail improvement program, but it is still mostly a nation of roads. Its development pattern is dense compared to US cities but low-density compared to Continental Europe. (For example, most urban Irish housing is either rows of townhouses or semi-detached houses, what North Americans would call duplexes.) So the Irish government has concluded that only road-based public transport, developed at scale, can reach enough of the population to be relevant. They’ve also followed the logic through to land use planning, hiring us to write a guidebook, now published, on how to plan for useful, efficient bus services when laying out new suburbs and towns.

The planned service increases are massive. Dublin already had the most extensive network of the Irish cities, but even in Dublin NTA is increasing service by over 30%. The increases will be around 45% in Cork, Galway and Limerick, and now 100%, a doubling of service, in Waterford.

Why so much more? Waterford is starting with a fairly minimal network by Irish standards. Here’s the existing network. It’s five routes, with frequencies of 20-30 minutes. (Note that in a departure from our usual mapping style, we’re using linewidth rather than colour to indicate frequency, mostly because if we drew all these entangled loops in the same two colours you’d never be able to follow it.)

Considering that it is made up of only five routes, the Waterford network is quite complex. Much of the complexity is caused by one-way loops: two routes are entirely one-way loops, and the other three routes have sections of one-way loop or one-way split. The benefit of all these one-way services is that a large area can be covered without using much service, but it makes people’s trips time-consuming, and it makes the network harder to understand.

And here is the proposed network. Every route will offer two-way service. The wide lines stand for service every 15 minutes, 7-days-a-week. The narrow lines stand for service every 30 minutes, 7-days-a-week.

This design achieves several things:

It’s radically simpler, because all routes are two way (with the exception of some short segments on one-way streets in the centre).
It fits frequency better to demand. It offers 15-minute “turn up and go” frequency linking the biggest destinations, including the University (SETU), the city Centre, and the main hospital on the east side. Most of the really dense parts of Waterford are on these segments.
It creates secondary focal points at the University (SETU) and hospital, offering direct service to these points from most of the areas surrounding them.
It covers several recent new developments.
Fewer routes end in the city centre. Instead, proposed Routes 1, 2, and 4 run through the centre and onward to the other side of the city. This reduces the need to interchange for many cross-city trips, and also makes better use of the limited space for terminating buses in the city centre.

As always, this is a draft! We know it will be improved by public feedback, which is already coming in. We look forward to the public conversation that starts now, and runs through 16 August.

Click here to explore the plan and express your view. You can find our entire report, with all of the details, here.

Miami: The Better Bus Network Is Here!

Posted on November 9, 2023 in General, Miami, Network Redesign

A slice of Miami-Dade’s Better Bus Network. See the link below for the full map.

On November 13, the greater Miami area will see the biggest transformation in where you can go on public transit since Metrorail opened almost 40 years ago. Not a new rail line, but a huge redesign of the bus network that will make it useful to more people for more trips, all over the County. A complete map of the new network is here.

Miami-Dade County’s buses run in patterns that have often been the same for decades, even as the county has grown and many new destinations have appeared. It’s hard to change bus service, because even the most inefficient bus route has people who depend on it and will object to any changes. So, to redesign a bus network, we have to show big benefits that make the change worth the trouble, and that’s what this redesign does.

The plan is the result of a planning project that begin in 2019. In an unusual partnership, Transit Alliance Miami funded much of the work and hired us (Jarrett Walker + Associates) to lead the planning process in partnership with the County. We led a public conversation around key trade-offs, by sharing contrasting network design concepts that showed the consequences of different possible goals. Based on the response to that process, we developed a Draft Plan in early 2020. We were then rudely interrupted by Covid-19.

Since 2020, the County has worked to finish the plan and as published a revised plan, with 30% more service, in 2021. Unfortunately, the transit industry was hit by the labor shortages of the post-Covid era, and the plan had to be reworked to manage with a smaller workforce than previously imagined. Even with the changes, the final plan now being implemented still delivers great results for huge swaths of the county, its residents, workers, and visitors.

With the new design, a frequent grid will cover large parts of the county with huge benefits. The simplest measure of that improvement is how many people or jobs are near service. The chart below shows the change in residents or jobs near service by the frequency of service at midday on weekdays. The number of residents who live near frequent transit will increase from 380,000 (14 of the County’s residents%) to 814,000 (30%) during weekday service. The new network will bring frequent service near almost 60% of households without cars; that’s 20,000 additional households without cars near more frequent service. And jobs near frequent service will increase from 29% to 43% on weekdays.

A key measure of a network’s usefulness is the access it provides, or how much stuff you can reach in a reasonable travel time. The animated map below shows an example of this change from Little Haiti.

In pink is the area you can reach in the Existing Network in 45 minutes. In blue is the area you can reach with the Better Bus Network. The blue area is larger, but more importantly it has a lot more stuff in it: 30% more jobs and 60% more residents. So if you lived near this place, you would effectively be 30% more free. And if you owned a business here, you’d now have access to 60% more customers, or 60% more workers.

We can measure this exact thing over and over again across the whole county and when we summarize the results we find that the average resident will be able to get to 28% more jobs (or other useful destinations) in 45 minutes. The benefit is even greater for lower-income residents and people of color. That means more people, when they look up a trip they might make, will find that the travel time is reasonable. For more on why we use this measure, see here.

The other big improvement in this new network is a major increase in frequency of service on weekends. Across the country, we’ve worked on network plans that have increased service on evenings and weekends and they’ve often shown huge ridership gains. People value flexibility and spontaneity. Everyone wants the ability to get home outside of the traditional 8-to-5 workday. Critically, though, people working in retail or restaurant jobs often need to work on weekends. A route that runs infrequently on the weekends is missing the peak time for people in these industries, and there are many, many people in these industries in Miami-Dade.

Implementing Big Change

Our team has been working closely with Miami-Dade staff to assist with a range of implementation needs. How does a huge change like this happen overnight? Months, sometimes years, of planning leads up to a big day like this. These efforts included:

Work by staff across County government to help people find out about the change, and can learn about it easily.
Briefings of elected officials including partners in the city governments.
An analysis of compliance with Title VI, the US Civil Rights law that ensures racial equity in transit planning.
Siting of new bus stops and removing old stops.
Writing new schedules for customers and bus operators.
A big effort by operations and safety teams working on testing turns, reviewing stop locations, working with operators to learn new routes, and much more.
Finally, an infusion of temporary staff near the change date, to be out on the street helping people find their way.

It takes an enormous team effort and County staff have worked hard for a long-time to make this day happen. Transit Alliance has continued to partner with the county to help with communicating the network changes to the public and we’ve been please to assist in that process. Our team has contributed in a few key ways:

A new system map with routes color-coded by frequency. These maps will start showing up in shelters around the county soon.
An interactive trip comparison tool to help folks find out how they can make their trips on the new network.
Developing bus stop signage to inform riders at each stop about which routes are changing.

We’re excited to see how people respond to this new network and we hope it helps set the stage for many transit improvements to come as the County implements its long-term SMART plan.

Older posts