Designing a Public Transport Network: Coverage, Frequency, Connectivity and Efficiency

2. A network is not simply the sum of its routes

Many public transport networks were not designed from scratch according to a particular architecture, but are instead the result of decades of successive changes. One route was extended to serve a new neighbourhood, another was modified to reach a hospital, another incorporated a variation to serve an industrial area, and later a new railway station, interchange or higher-capacity transport infrastructure appeared. At the same time, the city continued to grow, the main employment centres changed and new mobility patterns emerged.

Each modification may have been perfectly justified when it was introduced, but the accumulation of local decisions can ultimately result in a network that is inefficient as a whole. Partially duplicated routes appear, some routes become very long, different services perform similar functions, some areas have multiple alternatives while others remain poorly served, and peripheral journeys may still require travelling through the city centre even when the centre is not part of the trip.

A major network reorganisation should therefore not begin by asking what modification each existing route requires. We should first ask what functions the network as a whole needs to perform and only then determine which routes are capable of fulfilling them.



This change in perspective is important because a route ceases to be considered an independent unit and instead becomes one component of a larger system.

3. Before designing routes, we need to define what we want to achieve

There is no optimal public transport network independently of the objectives that have been established.

A solution designed to maximise territorial coverage will probably be different from one whose main objective is to increase ridership, improve commercial speed or attract users away from private cars. Similarly, a network primarily intended to guarantee basic accessibility throughout the territory will have different characteristics from one that concentrates resources on the highest-demand corridors.

Before modifying routes, we should therefore define which variables we want to improve and what level of priority we assign to each of them.

Population and employment coverage, accessibility to hospitals and educational facilities, passenger demand, frequency, journey times, number of interchanges, commercial speed, productivity per vehicle-kilometre, integration with rail or tram services, modal shift from private vehicles and the resources required are some of the indicators that may form part of this evaluation.

The problem is that many of these objectives compete with one another.

One alternative may carry more passengers while providing less territorial coverage; another may considerably improve coverage but require greater vehicle-kilometre production; another may increase frequencies but require some passengers to interchange.



The design process therefore consists precisely of understanding these relationships and deciding which combination best responds to the role we want public transport to play.

4. First, we need to understand the function currently performed by each route

Before reorganising a network, it is useful to understand not only how many passengers use each route, but also what function that route performs within the system.

Two routes with similar demand may perform completely different roles. One may structure a major high-demand corridor, while another may provide the only available connection between a peripheral area and a hospital. From a purely quantitative perspective, they may appear equivalent, but functionally they are not.

This analysis should consider the route alignment, the main origins and destinations served, passenger loads by section, occupancy, frequency, commercial speed, connections with other routes and modes, and the availability of alternative routes.

A route with low productivity may nevertheless perform an essential territorial function, while another with reasonable demand may significantly overlap with other services and consume resources that could be redistributed differently.

Rationalisation should therefore not simply consist of ranking routes according to passengers per kilometre and eliminating those at the bottom of the list.



We need to understand what each route contributes to the system as a whole, which journeys it enables and whether there is a more efficient way of maintaining that function.

5. Territorial coverage and service quality are not the same thing

One of the most commonly used indicators in network planning is coverage.

Using GIS tools, we can calculate how much population, employment or how many specific facilities are located within a radius of 300, 400 or 500 metres of a stop. This analysis is extremely useful because it allows us to identify poorly served areas and compare alternatives using a consistent territorial basis.

However, being close to a stop does not necessarily mean having access to good public transport.

Two homes located 300 metres from a stop may have completely different levels of accessibility. One may have several high-frequency routes, extensive operating hours and direct connections to the main destinations, while the other may have only one route operating every 40 minutes with an uncompetitive alignment.

It is therefore useful to distinguish between physical coverage and effective coverage.

The former tells us whether infrastructure providing access to the system exists; the latter should also incorporate frequency, operating hours, connectivity and journey time.



An area should not be considered adequately served simply because it appears coloured within the catchment area of a stop.

6. The fundamental conflict between coverage and frequency

Suppose we have a fixed number of vehicles and operating hours.

We can use these resources to extend the network across a very large territory through many routes operating at moderate frequencies, or we can concentrate them on a smaller number of corridors and provide much more frequent services.

In the first case, we have more kilometres of network; in the second, we have a greater intensity of service.

Both strategies may be reasonable depending on the characteristics of the territory, but they create very different mobility experiences.

This conflict is particularly important because frequency directly affects waiting time and, consequently, the competitiveness of public transport.

A route operating every 10 minutes can be used relatively spontaneously; a route operating every 45 minutes will normally require passengers to consult the timetable beforehand and will significantly penalise any connection with other services.

Extending a route in order to achieve a marginal increase in coverage may therefore, if additional resources are required, lead to a reduction in frequency that ultimately disadvantages many more passengers than it benefits.

The decision should not simply be formulated as “more coverage is better”, but rather by asking how much service we can actually provide across that coverage.



In many cases, it may be preferable to serve an area through a reasonably accessible connection point with a high-frequency service rather than attempting to bring the route physically closer to every potential passenger through increasingly long and less frequent alignments.

7. More kilometres of network do not necessarily mean greater accessibility

The total length of routes is a simple indicator to calculate and can help characterise a network, but on its own it does not measure its usefulness.

A 500-kilometre network is not necessarily better than a 350-kilometre network if a significant proportion of those kilometres correspond to redundant alignments, very low-frequency routes or services that inefficiently connect the main origins and destinations.

Accessibility depends on the combination of coverage, frequency, speed and connectivity.

A relatively compact network may provide access to a greater number of destinations within 30 or 45 minutes if it concentrates services on well-connected corridors, while a much more extensive network may provide lower effective mobility if waiting and travel times are high.



When evaluating alternatives, we should therefore move away from considering exclusively how much territory public transport covers and begin asking how many opportunities — employment, education, healthcare, retail or leisure — citizens can actually reach using the system within reasonable travel times.

8. Direct connections versus concentration of resources

From the passenger's perspective, a direct route is generally attractive.

Boarding close to the origin and alighting close to the destination without changing vehicles removes uncertainty and simplifies the journey.

The problem arises when we attempt to provide direct connections between a very large number of origin-destination pairs.

A city may contain thousands of significant travel relationships, and it is practically impossible to serve all of them through direct routes with good frequencies.

The more routes we create, the more widely resources must be distributed.

A network with twenty routes using a given fleet will, in general terms, be able to provide higher frequencies than another network with forty routes using the same number of vehicles.

This does not mean that we should minimise the number of routes, but rather that we need to understand the direct relationship between route diversity and the intensity of service that can be provided on each one.

A fundamental question therefore arises:

Is a direct connection every 30 minutes preferable to two services every 10 minutes connected through a reasonably efficient interchange?



There is no universal answer, but the example demonstrates that the number of interchanges cannot be analysed independently of frequency and total journey time.

9. An interchange is not necessarily a network failure

For many years, interchange has primarily been regarded as a penalty that should be avoided.

And it certainly is a penalty when passengers have to walk a considerable distance, wait for a long time, pay again or deal with an unreliable connection.

However, a structured network can use interchange between services to concentrate resources and increase the number of possible connections without having to provide direct routes between every possible origin and destination.

The objective should therefore not be to eliminate all interchanges, but rather to design properly those that are necessary.

This requires considering the physical distance between stops, waiting time, timetable coordination where frequencies are low, the information available to passengers, accessibility of the interchange point and fare integration.

A connection between two high-frequency routes may be almost spontaneous; a connection between two services operating every 60 minutes requires much more precise coordination.

This also changes the way a route should be evaluated.

Its usefulness is not limited to the destinations it reaches directly, but also includes those that can reasonably be reached using the rest of the network.



The real unit of analysis therefore begins to shift from the individual route towards the complete journey.

10. A hierarchical network can use resources more efficiently

One possible strategy is to assign different functions to different types of service.

High-demand corridors may have high-frequency and high-capacity routes, while secondary or feeder services connect lower-density areas to them. Cross-city routes may serve relationships between peripheral sectors, while certain services may address specific needs or particular time periods.

This hierarchy prevents every route from attempting to perform every function simultaneously.

A trunk route may prioritise speed, frequency and capacity; a local route may prioritise coverage; a feeder service may prioritise a reliable connection with an interchange.

Evaluating all of them using exactly the same indicators and thresholds would not necessarily be reasonable because their functions within the system are different.

The value of this architecture is not limited to operational efficiency.



A network in which each level has a recognisable function may also be easier for passengers to understand and more adaptable to future changes.

11. Main corridors should concentrate more than just demand

Identifying the main mobility corridors is normally one of the first steps in network design.

However, a trunk corridor should not be defined solely because it currently carries a large number of passengers.

Its ability to connect major trip generators, integrate with other routes, attract demand from other modes and maintain sufficiently stable operating conditions should also be analysed.

Along these axes, it makes sense to concentrate frequency, capacity and measures aimed at improving commercial speed.

Bus lanes, traffic signal priority, the reduction of bottlenecks or improvements to stops may allow a trunk route to provide a clearly higher level of service than the rest of the network.

When several routes use the same corridor, their combined frequencies can also generate a very high overall service level.



The objective should not simply be to identify where the greatest number of buses currently operate, but rather where it is strategically appropriate to create a public transport backbone around which the rest of the network can be organised.

12. Feeder routes only work if what they feed is attractive

Feeder services are frequently proposed as a means of connecting peripheral areas with high-capacity routes, railways, tramways or major bus corridors.

In theory, they make it possible to avoid numerous low-demand services travelling long distances towards the city centre and allow supply to be concentrated along the main axes.

However, the system will only work if the trunk service provides sufficiently attractive conditions and the interchange is properly designed.

Forcing passengers to abandon a direct connection in order to interchange onto another route with a similar frequency, little journey-time saving or a poorly accessible station is unlikely to represent an improvement.

The benefit appears when the main section provides a significant advantage in terms of frequency, capacity, speed or connectivity.



Feeder services should therefore not be treated as an objective in themselves, but rather as the consequence of a network structure in which the combination of services provides better performance than a collection of independent routes.

13. Peripheral connections can avoid unnecessary journeys through the city centre

Many urban networks retain a radial structure inherited from a city in which the centre once concentrated a large proportion of activities.

Today, however, urban areas are frequently more polycentric.

Hospitals, universities, technology parks, industrial areas, shopping centres and stations may be located far from the traditional centre and generate significant mobility relationships between peripheral sectors.

When the network remains strongly radial, a journey between two peripheral neighbourhoods may require travelling first towards the centre and then returning towards the periphery.

This increases journey time, introduces unnecessary kilometres and may overload central corridors with trips whose origins and destinations are both located outside them.

Cross-city or orbital routes may address some of these relationships, but they should respond to actual demand rather than simply to the desire to create a geometrically complete network.



The relevant question is which journeys they make possible, how much they reduce generalised cost and what frequency can be maintained with the available resources.

14. Route overlaps are not necessarily a problem

Two or more routes may share part of their alignment, and this does not automatically mean that there is an inefficiency.

On a high-demand corridor, overlapping services may provide a very high combined frequency before the routes branch towards different destinations.

In fact, this structure may be entirely consistent with a hierarchical network.

The problem arises when the overlap does not perform a clear function, particularly when several routes operate in parallel over long distances while other areas receive insufficient service.

Overlap may also occur between urban and interurban services when both serve similar movements without adequate functional, fare or timetable coordination.

Rather than simply counting overlapping kilometres, we should therefore analyse what purpose that overlap serves.



If it increases capacity where demand exists or provides a useful combined frequency, it may be positive; if it merely duplicates supply without adding meaningful connectivity or capacity, there is probably scope to reorganise resources.

15. Stop location and spacing are part of network design

A stop is not merely a physical element of service operation.

Its location affects coverage, pedestrian access time, the ability to interchange and the commercial speed of the route.

Increasing the number of stops generally reduces the distance passengers need to walk, but it also increases the time lost through deceleration, boarding, alighting and re-entering traffic.

There is therefore another trade-off between accessibility and speed.

For higher-performance routes, increasing stop spacing may be reasonable in order to achieve more competitive journey times, while local services may require a denser stop pattern.

Applying exactly the same criterion to every route may be just as inappropriate as requiring all routes to achieve the same frequency or productivity.

Coverage should also be analysed using the actual pedestrian network rather than simply drawing perfect circles around stops.



Urban barriers, gradients, major infrastructure or the absence of suitable crossings may turn an apparently short geometric distance into a much longer actual access route.

16. Commercial speed should be considered from the beginning

Frequently, the route alignment is designed first and only afterwards is the time required for the bus to complete it estimated.

However, commercial speed should form part of the initial design because it directly determines the competitiveness of the service.

A route that successively crosses congested corridors, makes numerous conflicting turns and has a stop every few hundred metres may provide excellent coverage while simultaneously offering unattractive journey times.

Slightly modifying an alignment to use a more fluid corridor, reducing the number of stops, introducing traffic signal priority or providing a reserved lane may be more effective than adding new kilometres of route.

Even comparisons between two alignments should not be based solely on distance: a slightly longer route may be faster and more reliable if it operates under better traffic conditions.

The network should therefore be designed not only in space, but also in time.



The map shows where a route goes; commercial speed gives us a much better indication of how passengers will actually experience that journey.

17. Frequency, capacity and regularity should be analysed together

A route is not defined solely by its alignment.

A complete proposal must also establish frequency, capacity and operating conditions.

Transporting 600 passengers per hour using six buses with a capacity of 100 passengers does not necessarily provide the same quality of service as using three vehicles with a capacity of 200 passengers, even though the theoretical hourly capacity is identical.

In the first case, average waiting time will be lower and the network will provide greater temporal flexibility.

Frequency also strongly influences the structure of connections.

For very frequent services, precise timetable coordination may be unnecessary because the expected waiting time for the next vehicle is short.

For low-frequency services, the opposite applies, and a missed connection may add 30, 45 or even 60 minutes to the journey.

Furthermore, the planned frequency must be capable of being maintained with reasonable regularity.

Designing a high frequency on an extremely long route that is highly vulnerable to congestion may eventually lead to vehicle bunching and irregular headways.



Network design and operational design should therefore not be completely separated.

18. The network does not need to be identical throughout the day

Demand varies significantly between peak periods, off-peak hours, nights, Saturdays, Sundays or particular seasons.

A network may therefore maintain a relatively stable physical structure while offering different operating configurations according to the time of day.

Certain routes may be reinforced during commuting or school peaks, others may have extensions operating at specific times, and some services may operate only at night or during weekends.

The important point is to avoid unnecessary complexity that makes the system difficult to understand.

A network containing too many variants may respond very accurately to each demand segment on paper while becoming difficult to understand in practice.

Temporal planning should therefore seek a balance similar to spatial planning: adapting resources to demand without losing service legibility or creating an operating structure that is difficult to maintain.

19. Low-density areas require a different approach

One of the most difficult problems arises in dispersed or very low-demand areas.

Maintaining a conventional route with a good frequency may require a very large amount of vehicle-kilometres while carrying very few passengers.

Drastically reducing frequency improves economic efficiency, but it may make the service of limited practical use and create a downward spiral in which poorer service produces even lower demand.

In these environments, other solutions may need to be considered: coordinated timetabled services, feeder routes towards major nodes, smaller-capacity vehicles or demand-responsive transport.

The appropriate option will depend on the spatial and temporal distribution of trips, the distance to the main urban centre and the social and territorial needs that the service is expected to guarantee.

The important conclusion is that universal coverage does not necessarily mean uniform service provision.



Guaranteeing basic accessibility throughout the territory may require different transport products according to density and demand characteristics.

20. The bus network should be designed together with the rest of the transport system

Passengers do not necessarily distinguish between administrative responsibilities or operating contracts.

They may use an urban bus, then a railway service and finally walk to their destination.

From their perspective, there is a single journey, and the success of the system depends on how its different components interact.

Reorganising an urban network without analysing interurban services, rail, metro, tram, major interchanges, park-and-ride facilities or cycling connections may therefore create duplication or miss important opportunities.

A new high-capacity transport infrastructure can completely alter the function of particular bus routes, which may no longer need to travel all the way to the city centre and may instead adopt a feeder role.

However, this reorganisation will only be positive if intermodality actually works.



The physical existence of a station does not guarantee a good connection; interchange times, frequency, coordination, accessibility and fare conditions must all be analysed.

21. Attracting car users requires designing for journeys that are currently outside the network

If we analyse only those who already use public transport, we risk designing a better network for existing passengers but one that is incapable of significantly changing the modal split.

To attract journeys currently made by car, we need to study the origin-destination relationships of those users and understand why public transport is currently uncompetitive.

In some cases, a direct connection may be missing; in others, the problem may be insufficient frequency, excessive journey time, poor railway connectivity or difficulties associated with the first or last mile.

The solution does not always need to involve creating a new route.

It may be more effective to improve the speed of an existing corridor, increase frequency, introduce a cross-city route, coordinate services or develop a park-and-ride facility connected to a high-capacity route.

This changes the usual question.



It is no longer simply a matter of determining how to improve what we currently have, but rather what the network would need to provide in order to become a reasonable alternative for journeys currently made using other modes.

22. The complete journey should be the real unit of design

From an administrative perspective, it is natural to think in terms of routes, scheduled trips and stops.

However, passengers do not want to use a route; they want to travel from an origin to a destination.

The route is only one of the elements required to achieve this.

The complete journey includes walking to the stop, waiting, in-vehicle time, a possible interchange, another wait, another in-vehicle section and finally access to the destination.

Two alternatives with similar in-vehicle travel times may provide very different levels of service if one requires considerably more walking or introduces an interchange involving a low-frequency service.



Indicators such as generalised cost, accessible destinations, and the number and quality of interchanges are therefore much more representative of the actual usefulness of a network.

23. Alternatives should be genuinely different before they are compared

A sound design process should not produce a single proposal and subsequently use indicators to demonstrate that it works.

It is much more useful to develop deliberately different alternatives that allow us to understand the consequences of adopting different priorities.

For example, we could develop one alternative aimed at maximising coverage, another that concentrates resources in order to achieve greater frequency and efficiency, and a third, more hierarchical alternative that strengthens complementarity with other modes and interchange between services.

All three should be defined in sufficient detail to determine not only their alignments, but also their frequencies, journey times, vehicle-kilometres, fleet requirements, estimated demand and key accessibility indicators.

Alternatives therefore perform a fundamental role: they make the trade-offs visible.

If the network providing the greatest coverage requires 20% more production to carry only 5% more passengers, that information should be made explicit.



Similarly, if the most efficient alternative significantly reduces accessibility for part of the territory, that consequence should also be clearly identified before a decision is made.

24. Multi-criteria analysis should help explain the decision, not hide it

Comparing alternatives requires indicators, but it is difficult for a single value to represent all the objectives of public transport.

Coverage, demand, frequency, commercial speed, accessibility, intermodality, modal shift from private cars, vehicle-kilometres, fleet requirements, costs and environmental impacts describe different dimensions, all of which may be relevant.

Multi-criteria analysis makes it possible to structure this information and assign weights to different objectives, but it should not be used as a black box that automatically produces a winning solution.

The weights represent priorities and should therefore be clearly justified.

One alternative may perform better if efficiency is prioritised, while another may be preferable if greater importance is assigned to territorial coverage; understanding this sensitivity is probably more useful than obtaining an apparently precise final score.

The real value of the analysis lies in making the consequences of each decision explicit.

A public authority may consciously decide to retain lower-productivity services because they perform an essential social or territorial function.



What matters is understanding the cost of that decision, the alternatives available and the benefits it is intended to provide.

25. Conclusion: from drawing routes to designing a network

Designing public transport does not mean placing the largest possible number of routes on a map.

A good network must find a balance between objectives that frequently compete with one another: coverage and frequency, proximity and speed, direct connections and concentration of resources, efficiency and territorial cohesion, simplicity and the ability to serve a complex urban structure.

Planning should begin by understanding the function that each part of the system needs to perform and then translating those functions into corridors, routes, stops, frequencies and connections.

A high-demand route does not necessarily need to perform the same function as a peripheral service; a low-density area does not necessarily require the same transport product as a major urban corridor; and an interchange-based connection may provide greater accessibility than a direct route if it allows resources to be concentrated and sufficiently high frequencies to be provided.

The quality of a network should therefore not be measured solely by the number of routes, kilometres operated or percentage of the population located close to a stop.

We should simultaneously analyse how many destinations passengers can reach, within what travel time, at what frequency, with how many interchanges and using what level of resources.

The criteria currently used by public authorities themselves to assess alternatives — coverage and accessibility, demand, efficiency per vehicle-kilometre, commercial speed, frequency, intermodality, modal shift from private cars, required resources and costs — illustrate precisely why the system needs to be considered from multiple perspectives.

Perhaps the fundamental change is to move progressively away from planning centred on individual routes and begin thinking in terms of the network as a whole.

A collection of routes attempts to make each service independently solve as many journeys as possible.

A genuine network attempts to ensure that the system as a whole provides the greatest possible number of mobility opportunities.

And that difference, although it may appear small on a map, changes practically every network design decision.



Lluis Sanvicens, 2026

Images taken in Athens in 2026. Source: author's own archive.