From Scheduled to Delivered Service: How to Measure the Real Quality of Public Transport

2. There Are at Least Three Different Services: Designed, Delivered and Experienced

When analysing quality, it is useful to distinguish between concepts that are often treated as interchangeable. The first is the designed service: routes, alignments, frequencies, timetables, journey times, vehicles, stops and all the other conditions established during the planning process. The second is the service actually delivered by the operator. Finally, there is the service experienced by passengers, which may differ from both.

Suppose a route has been designed to provide six trips per hour. The operator may indeed operate all six trips and, from a service delivery perspective, appear to meet the planned requirements. However, if the intervals between vehicles are 4, 16, 5, 15, 8 and 12 minutes, passengers are not actually receiving a regular ten-minute service. Service delivery may be quantitatively correct while the passenger experience remains poor.



Performance assessment should focus precisely on these differences. It is not enough to check how much service has been scheduled, or even how much has been delivered. We need to understand how that service has been distributed across space and time, and what conditions passengers have ultimately experienced.

3. Public Transport Quality Cannot Be Reduced to a Single Indicator

There is a tendency to summarise service performance through a limited number of figures: passenger numbers, punctuality, kilometres operated or overall satisfaction. All provide useful information, but none can describe the quality of the system on its own.

The quality plans included in the documents reviewed cover a much broader range of dimensions: time, comfort, information, accessibility, safety, service availability, customer care and environmental impact, together with aspects related to vehicles, operating procedures, staff behaviour, complaints and passenger satisfaction. This broad approach is logical because a single journey may perform well in one dimension while presenting significant shortcomings in another.



A bus may run on time but be overcrowded. It may offer a good frequency but provide incorrect information. It may achieve an excellent operating speed but serve an inaccessible stop. It may complete all scheduled kilometres while distributing them irregularly. Quality must therefore be understood as a system of attributes rather than a single numerical value.

4. Before Measuring Quality, We Must Define What Level of Service We Consider Acceptable

We cannot monitor what we have not previously defined. Stating that we want a service to be "punctual", "regular", "accessible" or "high quality" is insufficient unless we establish what each of these concepts means technically and what minimum standard we expect to achieve.

This is precisely one of the functions assigned to Quality Plans in the documents reviewed: establishing service characteristics, assigning weights to different attributes, defining performance standards and determining the consequences associated with their level of compliance. This approach is particularly valuable because it transforms general principles into verifiable commitments.



For each indicator, we should know, at a minimum, what is being measured, which data source is used, the period covered, the level of aggregation and the threshold that determines whether the service meets the required standard. An incomplete definition may produce apparently precise indicators that subsequently allow very different interpretations.Nuevo párrafo

5. The First Check Should Be Whether the Scheduled Service Was Actually Operated

Before discussing punctuality, regularity or satisfaction, there is a much more basic question: did the scheduled trips actually operate?

A route may have 120 trips scheduled on a particular day and operate 116. The overall completion rate will be high, but we still need to know which four trips were lost. Cancelling four services spread throughout the day does not have the same effect as losing four consecutive trips during the peak period.



The same applies to kilometres operated. Vehicle-kilometres are a fundamental unit for describing public transport supply and appear repeatedly in technical transport documentation. However, aggregate compliance with scheduled kilometres should be complemented by information about when, where and through which trips those kilometres were delivered. Quality assessment begins by verifying that the service exists, but it does not end there.

6. Averages May Hide Precisely the Problems We Want to Detect

Aggregation is necessary when managing large volumes of information, but it can also conceal important patterns. A route may show reasonable average punctuality while experiencing systematic problems between 7:30 and 9:00 a.m. Another route may maintain an acceptable operating speed over most of its length but consistently lose considerable time along a particular section.

Indicators should therefore be capable of being disaggregated at least by route, direction and time period, and, where necessary, by section or stop. The objective is not merely to identify that a problem exists, but to locate it.

This capability is particularly relevant with today's monitoring tools. The documents reviewed explicitly propose systematically processing information from operators, automatic vehicle location and management systems, and operational management platforms to generate service quality reports. The availability of detailed data makes it possible to move beyond exclusively monthly or annual indicators and obtain a much closer understanding of how services actually operate over time.

7. Punctuality and Regularity Are Not Exactly the Same

For services operating according to specific timetables, particularly those with low frequencies, it makes sense to measure the difference between the scheduled and actual arrival or departure times. A passenger waiting for a bus scheduled at 10:05 needs that service to arrive reasonably close to the published time.

However, when service frequencies are high, the regularity of intervals between vehicles may be just as important, or even more important, than strict adherence to individual scheduled times. If a route operates every ten minutes, many passengers will be more concerned about vehicles maintaining approximately those intervals than about whether a particular bus was scheduled to arrive at 8:10 or 8:12.



Public authorities' diagnostic plans also consider frequency, regularity, operating speed and transfer times together when assessing public transport quality. These variables are related but distinct, and confusing them may lead to a situation in which punctuality is assessed correctly while the actual waiting experience remains poor.

8. Meeting the Average Frequency Does Not Necessarily Mean Providing a Good Service

Consider again a route that should operate every ten minutes. During one hour, all six scheduled vehicles complete their journeys, meaning that the planned service volume is delivered exactly. But if two buses travel almost together and are followed by a twenty-minute gap, the hourly capacity may remain unchanged while the passenger experience deteriorates significantly.

This phenomenon can also reinforce itself. The first vehicle following a long interval encounters more passengers waiting, requires additional time for boarding and alighting, and may become even more delayed. The following vehicle encounters fewer passengers, completes stops more quickly and eventually catches up with the vehicle ahead. Initial irregularity can therefore generate further imbalances.



It is consequently important to analyse the distribution of headways rather than focusing exclusively on average frequency. Public transport quality is delivered minute by minute, even if we subsequently need to summarise it through indicators.

9. Departing Early Can Be Just as Problematic as Arriving Late

When discussing punctuality, we usually think primarily about delays. However, for a passenger arriving at a stop according to the published timetable, a vehicle departing several minutes early can have an especially serious consequence: missing the service entirely.

The impact clearly depends on frequency. Missing a bus on a route operating every five minutes has a relatively limited effect; missing a service that operates once an hour may add almost sixty minutes to the journey.



This illustrates once again why indicators should be designed from the passenger's perspective. A measure focused exclusively on positive delays may provide an incomplete picture. Monitoring should assess deviations from the expected service and the consequences those deviations have for passengers' actual ability to use it.

10. Waiting Time Is One of the Most Direct Ways Passengers Experience Operations

Passengers do not directly observe the operator's headway regularity indicators, but they experience their consequences while waiting at stops. Two networks with identical scheduled frequencies may generate very different waiting times if one maintains regular intervals while the other experiences substantial variations.

Moreover, waiting time tends to be perceived particularly negatively because it occurs outside the vehicle and involves uncertainty. If passengers do not know when the next service will actually arrive, their perception may deteriorate even when the expected waiting time is not exceptionally long.



Monitoring regularity is therefore not exclusively an operational matter. It is a way of managing passengers' total journey time. In the article dedicated to transfers, we also saw how this uncertainty can spread throughout an entire journey when a subsequent connection is involved.

11. Journey Times Must Also Be Verified After the Service Has Been Introduced

During the planning process, we use journey times to establish timetables, calculate operating cycles, determine fleet requirements and assess the competitiveness of routes. However, these values remain assumptions until they can be compared with actual operating conditions.

Automatic vehicle location and management systems, together with GPS data, make it possible to obtain actual passing times and reconstruct journey patterns. The first article in this series already highlighted how these systems provide vehicle positions, journey times, regularity, operating speed and trip completion information.

Once a network is operating, these data allow us to verify whether the assumptions made during the design process remain valid.



If a route was planned using a 90-minute operating cycle but consistently requires 100 minutes, this is not simply a punctuality problem. Timetables, recovery times, fleet size or infrastructure improvements may need to be reconsidered.

12. Average Operating Speed Does Not Tell the Whole Story Either

Operating speed is particularly useful because it connects passenger service quality with operator productivity. However, a daily average can conceal substantial variations over time.

A route may operate at 18 km/h for much of the day and experience a significant reduction during peak hours because of congestion at particular locations. From an annual perspective, the average may appear acceptable, while the service loses competitiveness and regularity precisely during the periods of highest demand.



It is therefore advisable to analyse operating speed by section and time period. If we identify where time is consistently lost, we can propose targeted measures: traffic signal priority, stop modifications, dedicated bus lanes, route changes or other interventions. Quality monitoring then becomes more than a compliance mechanism; it becomes a technical tool for improving the network.

13. Reliability Can Be Just as Important as Average Journey Time

A passenger can adapt reasonably well to a journey that consistently takes 35 minutes. It is much harder to plan a trip that takes 25 minutes on some days and 50 on others.

Knowing only the average journey time may therefore be insufficient. Variability is also a fundamental service characteristic, particularly for commuting, educational journeys or trips involving connections with other transport modes.



This is especially important because reliability determines the additional time passengers must allow when planning their journeys. If someone needs to arrive at a railway station at a particular time and knows that the connecting bus experiences significant variations, they will probably leave much earlier than strictly necessary, or may even choose another mode. Quality assessment should consider not only how long a service takes on average, but also how much confidence passengers can place in that journey time.

14. Occupancy Shows Whether the Service Is Actually Providing the Designed Capacity

A service can operate punctually and still provide insufficient quality if demand regularly exceeds available capacity. Passenger measurement should therefore not be limited to daily demand or total ticket validations along a route.

We need to understand how demand is distributed along the route and throughout the day. Boarding and alighting counts make it possible to reconstruct passenger loads by section and identify where maximum occupancy occurs. Ticketing data provide additional information about usage and temporal distribution, although, depending on the system available, they may not be sufficient on their own to reconstruct the complete passenger load profile. This complementary use of ticketing data, vehicle monitoring systems and passenger counts was already discussed in the first article in the series.



The important distinction is between theoretical capacity and the capacity effectively available to passengers. If certain vehicles consistently operate at or above comfortable capacity while nearby services have spare capacity, there may be opportunities to adjust timetables, frequencies or vehicle allocation.

15. Quality Also Exists Inside the Vehicle

Not every aspect of the passenger experience can be measured in terms of time. The quality plans reviewed explicitly include attributes related to comfort, vehicle characteristics, safety, driver assistance systems, cameras, staff behaviour and other components of the service.

These variables may appear more difficult to quantify than punctuality, but that does not make them less relevant. Vehicle condition, cleanliness, climate control, ease of boarding, equipment functionality and staff conduct all form part of the service passengers actually receive.



This is where automatic data sources need to be combined with inspections and specific checks. An automatic vehicle location system can indicate exactly where a vehicle is, but it cannot, by itself, reliably determine whether the interior meets expected cleanliness standards. Not every quality attribute can be measured using the same instrument.

16. Accessibility Must Be Monitored as an Actual Service, Not Merely as a Vehicle Feature

A technically accessible fleet does not guarantee that every journey is accessible. The operation of ramps, vehicle positioning at stops, physical obstacles, platform conditions and available information can all determine whether passengers can actually use the system.

Accessibility is therefore explicitly included among the quality commitments identified in the documents reviewed. Monitoring should verify that the conditions established during planning are maintained throughout operations and that the necessary equipment functions effectively.



This perspective once again places passengers at the centre. The relevant indicator is not simply how many vehicles have a particular accessibility feature, but how many journeys can actually be made by passengers who need to use it.

17. Passenger Information Must Also Be Audited

A network may be operating correctly and still produce a poor passenger experience if the information provided does not reflect reality. Outdated timetables, poorly communicated route changes or inaccurate arrival predictions can create uncertainty even when the physical service is available.

Recent Intelligent Transport Systems (ITS) clearly demonstrate the importance of this information layer. The documents reviewed describe modules for receiving and processing automatic vehicle location data, estimating vehicle arrival times at stops, publishing transport information and providing open data.



Passenger information should therefore be assessed as another service attribute: availability, updating, consistency across channels and correspondence with actual operations. The objective is not simply to provide an application or electronic displays, but to ensure that the information they present genuinely helps passengers make decisions.

18. Incidents Reveal a System's Real Ability to Manage Quality

All services experience incidents. Vehicle breakdowns, accidents, exceptional congestion, roadworks, severe weather or infrastructure disruptions can temporarily prevent services from operating exactly as designed.

In these situations, quality cannot be assessed solely by asking whether a disruption occurred, because some incidents will be unavoidable. We must also examine how the system responded: how quickly the incident was detected, what measures were implemented, how passengers were informed and how long it took to restore reasonably normal service conditions.



The minimum monitoring requirements included in the documents reviewed explicitly address the effective implementation of contingency procedures. This is particularly relevant because it turns operational resilience into a measurable component of service quality.

19. Complaints Are Also Operational Data

A passenger complaint is usually handled administratively as a customer service matter. From a technical perspective, however, it can also be a valuable source of information.

If different passengers repeatedly complain about delays on a particular route, accessibility problems at a specific stop or incorrect information at a certain location, those complaints may reveal a pattern that deserves investigation. The documentation reviewed explicitly incorporates complaint management, monitoring and response quality into service quality control.



Naturally, complaints do not represent a statistically representative sample of all passengers. They should not be used in isolation to measure overall quality. However, when combined with vehicle monitoring data, inspections and surveys, they can help identify problems that may not be visible through automated indicators.

20. Objective Quality and Perceived Quality Do Not Always Coincide

A service may improve technically while passenger satisfaction remains unchanged. The opposite can also happen. Perception depends not only on objective performance but also on existing expectations and on which attributes matter most to different passengers.

The quality plans reviewed explicitly recognise this dimension through satisfaction indicators and a Subjective Quality Index derived from passenger surveys, expressed on a scale from 0 to 10. This approach is particularly useful because it separates the objective measurement of service performance from the perceptions of those who use it.



Both are necessary. If measured punctuality deteriorates but passengers do not appear to notice, we should not ignore the objective evidence. And if technical indicators improve while satisfaction declines, we should not automatically conclude that passengers are wrong. There is probably some dimension of their experience that our indicators are not adequately capturing.

21. Satisfaction Surveys Should Complement Data, Not Replace Them

Surveys allow us to ask passengers directly about aspects that cannot easily be observed through sensors: comfort, perceived safety, ease of understanding information, staff behaviour and overall satisfaction. The documents reviewed identify satisfaction measured through surveys as one of the minimum quality monitoring requirements.

However, as we saw when discussing demand, a survey is still a sample, and its results must be interpreted accordingly. Overall ratings may also be influenced by recent experiences or by attributes that carry different levels of importance for different passenger groups.



The most robust methodology is once again to combine sources. If a survey identifies a punctuality problem and vehicle monitoring records confirm increased variability, the evidence becomes much stronger. If the two results diverge, that difference deserves investigation rather than simply selecting whichever source is more convenient.

22. Vehicle Monitoring, Ticketing and ITS Systems Enable a Shift from Sample-Based Control to Almost Continuous Monitoring

Technological transformation is fundamentally changing the ability to supervise public transport services. Historically, many checks had to be carried out through specific observation campaigns. Today, automatic vehicle location and management systems continuously generate operational information, while ticketing systems record a substantial proportion of service usage.

The central ITS platforms described in recent documents even include specific modules for service quality and compliance management, KPI generation, performance assessment and visualisation, reporting and dashboard development. The direction is clear: monitoring is moving from an occasional activity towards becoming a permanent management function.



This does not eliminate the need for fieldwork, surveys or inspections. It means that these methods can be reserved for aspects that automated systems cannot observe, while continuous data can be used to identify when and where more detailed investigations are required.

23. Having Millions of Data Points Does Not Guarantee Good Indicators

Digitalisation can create another risk: confusing the volume of information with actual knowledge. A system may store millions of vehicle positions, ticket validations and stop-passing records, yet still produce indicators of limited value if the data are not correctly structured or if there is no clear definition of what needs to be measured.

Quality management therefore also requires traceability, validation and proper organisation of data sources. The documents reviewed specifically emphasise the ability to systematically process information from operators, vehicle monitoring systems and management platforms to generate quality reports and monitor operations.



Before using a KPI, we should understand where the data come from, which records have been excluded, how incidents have been handled and what period the indicator represents. Otherwise, we may end up spending hours discussing differences of a few decimal points in an indicator whose methodological basis is not sufficiently consistent.

24. Indicators Can Be Used Not Only to Observe a Service, but Also to Manage It

Once measurable quality standards have been established, an additional question arises: what happens when those standards are met, exceeded or breached? Some of the models described in the documents reviewed directly link particular quality attributes to financial consequences through bonus and penalty systems. Attributes may be classified according to their importance, with positive, negative or neutral effects, or warnings, depending on the level of compliance achieved.

This approach can be powerful because it makes quality an integral part of the service management framework. However, it requires careful indicator design. If we incentivise what is easy to measure rather than what genuinely matters to passengers, we may end up optimising the KPI instead of improving the service.



Indicators should reflect real objectives and remain closely connected to the passenger experience. Kilometres operated, trips completed, regularity, punctuality, availability, operating speed, satisfaction and complaint handling may all form part of the assessment, but their weighting should reflect what we want the system to achieve. The objective is not to build the dashboard with the largest number of variables, but to identify those that allow us to determine whether the service is genuinely meeting its objectives.

25. Conclusion: Public Transport Does Not End When the Design Is Approved

Planning a network, defining routes, establishing frequencies, determining fleet requirements and developing an economic model are all essential activities. However, none of them guarantees that passengers will receive the service we have designed. Between planning and the actual passenger experience lies the full complexity of daily operations.

We should therefore always distinguish between the scheduled service and the service effectively received. We need to verify whether trips operate, whether they maintain the expected intervals, whether journey times remain stable, whether operating speeds allow timetables to be met, whether sufficient capacity is available, whether vehicles and stops maintain the required conditions, whether passenger information is accurate, whether incidents are managed effectively and whether passengers perceive that the system meets their needs.

Current technology allows us to observe many of these variables with a level of detail that would have been difficult to imagine only a few years ago. Automatic vehicle location systems, ticketing platforms, ITS technologies and dashboards can provide almost continuous information about operations, while inspections, surveys and complaints help us understand dimensions that automated systems cannot capture. Public authorities are also beginning to organise this information through quality plans that establish attributes, minimum standards, objective indicators, satisfaction indices and, in some management models, financial consequences linked to compliance.

But perhaps the most important question is not technological. Measurement should not become an end in itself. A dashboard may contain hundreds of indicators and still fail to answer the fundamental question: are passengers actually receiving the service we decided to provide?

Answering that question requires continuously comparing what was planned with what was delivered, and what was delivered with what passengers experienced. When a difference emerges, we need to identify it, understand its cause, introduce a corrective measure and subsequently verify whether that measure has worked.

This also completes the cycle that began with the first article in this series. First, we sought to understand how people move around the city. Then we designed a network capable of serving that mobility. Next, we made its routes function as an integrated system through connections and transfers. Finally, we observe actual operations to verify whether what we designed really works.



Because a public transport network should not be considered good because of what it promises on paper.

It should be judged by the service it manages to deliver every day.

Lluis Sanvicens, 2026

Photographs taken in Berlin in 2024. Source: author's own archive.