[Transport] The PAN: centralising data to decarbonise travel

The National Access Point (NAP) for multimodal transport data

Walid: Welcome to all, welcome to the podcast Projets Libres. I’m very happy, we’re going to continue the series on transport. We’re on the fourth episode and today we’re going to pick up on a topic that was discussed in episode 15 of season 3 with Tu-Tho Thai ; we had talked a little about European regulations and in the conversation came the National Access Point (PAN), a very interesting subject and so I took the liberty of contacting the head of the National Access Point who is called Maxime Siret. Maxime, who is the product manager transport.data.gouv.fr is with me, and today we are both going to talk about the national access point, what it is, what it is for, in what environment it operates, who it interacts with. In short, a lot of very interesting questions that I will be able to discuss with him. Maxime, delighted to have you with me, thank you very much for being here. Welcome to the Projets Libres podcast, I hope everything is going well on your side.

Maxime: Hi Walid, thank you very much for the invitation, very happy to be able to present the product to your audience. Especially since after what you did with Tu-Tho, I have a bit of pressure because I think she’s pretty knowledgeable about the subject… But thrilled to be able to make the product today, thanks for the invitation.

Presentation of Maxime Siret

Walid: Cool, let’s start. Before I introduce the National Access Point, I’m going to ask you to introduce yourself quickly. Can you tell us what your background is, where you come from and how did you come to discover open data?

Maxime: Listen, I’m 30 years old, I grew up in Bordeaux, I studied engineering quite traditionally. I liked transport, I loved taking the tram when I was in Bordeaux. I grew up with the tram in Bordeaux, which arrived at the same time as my teenage years. And so, that’s how I naturally headed for the world of public transport. I started with a long-standing operator, Transdev, for those who know. Then afterwards, I had the opportunity to work for a digital giant, Google Maps, for four or four and a half years. And despite what you might think about Google Maps, I had the opportunity to work via Google on open data projects since Google Maps integrates a lot of open data, and it is in this capacity that I then had the opportunity to join the ministry to work directly on open data with the open data product transport.data.gouv.fr that we are going to present today. That’s kind of my background.

What is the National Access Point?

Walid: So, let’s get to the heart of the matter. Can you explain to us what the national access point is, in what environment it evolves? I’ll let you speak.

Maxime: First of all, in a very concrete way, the national access point is a website, it’s something digital, on which we will be able to access data that is initially of a national nature. Obviously, there is some data that will also be cross-border, and so it is an access point that is at the level of a country, and its neighboring countries, on which we will be able to access a large amount of data. There are NAPs on each theme, let’s say. The NAP in question today is the multimodal transport NAP: Transport.data.gouv.fr. He evolves in an environment that is that of the administration. We will come back to this later, but it is historically a product that was created under the Ministry of Digital Affairs, well the Interministerial Digital Directorate (DINUM), and which was then attached to its attached ministry, namely obviously the Ministry of Transport. So it’s in an ecosystem that is basically very digital and which, as we go along, is turning to more business components, namely now the Ministry of Transport.

Walid: Okay, you said it’s one of the national access points. Are there other national access points?

Maxime: There are other national access points on the theme of transport. There are mainly two. The multimodal NAP, mine, is the PAN that brings together multimodal transport data: buses, metros, bicycles, carpooling, car-sharing. And then, there is a more road PAN which is hosted on the Bison Futé site for the moment and which contains road data. There, it’s no longer everything about speed limits, stops, road safety, tolls, etc. And then, of course, there are NAPs of other themes. For example, there is the National Address Database which, you could say that it is also a PAN, but on address data. And I think there are many more. I’m less of an expert on the others, obviously, but a PAN is just, in fact, a site whose purpose — which is normally rather neutral and which is managed at the administrative level — to aggregate data on the same theme on a national scale.

The background before the NAP was created

Walid: We’ll talk about it on the podcast in the coming weeks. I have other recordings on the subject. If we go back to the history of the PAN, so for multimodal transport, on what date did we decide — and moreover what made us decide — to create the NAP and what was the ecosystem at the time and what made us decide to create a national access point?

Maxime: spoiler, I wasn’t there at the time. I will speak with what I have heard from those who were present before. By the way, those who were present, maybe they will listen to me, especially Ishan. The NAP was created in 2017-2018. There were two reasons why it was created mainly, I think. The first reason is legal. There is a European regulation transposed into French law where this regulation requires the creation of this NAP. Basically, the centralization of transport data on the same site managed by a public administration. That was the first reason. And why is there this regulation?

It’s because there’s an observation that has been made, which is that historically, there are two worlds that don’t talk to each other much. There is the world of data producers. When we talk about data producers in the NAP, we are talking about those who operate the transport networks. For example, RATP, SNCF, they operate metros, trams, trains, and they generate bus schedules, metro schedules, etc. So they generate data schedules. And on the other hand, we have those who are called reusers: they will be interested in having access to this data, to create a particular service for users and passengers. And historically, these people had difficulty accessing this data.
Maxime Siret

The data was limited, not necessarily accessible, often paid for and often of poor quality. And so the observation was made that if we put an interlocutor between the two with a Marianne stamp, the State, which guarantees that the data is up-to-date, it is of quality and it is accessible in open data and in the long term, then we would be able to attract more reusers. And if we were to attract them more, behind them, more services could be created. And if more services are created, potentially more attraction for multimodality for the French and therefore more decarbonisation of transport, what is ultimately the ultimate goal, let’s say, of the PAN, is to decarbonise mobility. So for that, the French need to have access to passenger information and for that to happen, the data must be available. And so the PAN is a bit of a bridge: it’s as if there were a river, there was the left bank and the right bank, and in fact now there is the bridge and you can cross it. Well, that’s kind of the idea of the PAN. That’s why it was created historically.

Walid: ok, for the context again, see the episode with Tu-Tho. So indeed, we are starting from a situation in which producers and reusers do not talk to each other much, the data that is available is not necessarily complete and not necessarily of quality. It is therefore difficult for the players present or the players who would like to enter this market to offer services. And so, the answer to all this is the PAN. And at the time, was this data made available before the PAN? Where and by whom are they actually made available?

Maxime: in fact, they are not made available, they are hosted by each producer. And each reuser, each researcher, student, VSE, SME or large company who would like to have access to this data to provide a service and innovate — and also to enhance the producer’s data, because there is that too — he or she must go and contact these data holders individually and potentially enter into a contract with them to have access to this data. So, you potentially have to pay. It is done on a case-by-case basis. There is no generalization of this access, which means that it is a huge obstacle to access to data. And this obstacle is so enormous that there are very few services that are being created. And so, there is very little added value behind the data that remains the property of the producers. It’s not necessarily their job to innovate with this data, it’s to run buses. Finally, I caricature, if they hear me, that they do not protest. But that’s their job at the core. So the goal is precisely to create innovative services with this data. And for that, we have to free up data, we have to make it accessible.

Walid: When there is a decision to create this national access point, what are the steps that are put in place for the effective creation of this national access point? Because I assume that, as it is a European directive, each Member State is free to create its NAP in the form it wants. Certainly in other countries, there are other forms that have been chosen. What are the discussions at the French level? Was the current form that was chosen natural or were there discussions about the forms, where to host it, how to do it, etc.?

Maxime: yes, so the same thing, I wasn’t there and so I couldn’t answer you precisely. But indeed, there are different forms of NAPs, particularly in terms of financing.

There are European countries that have decided to have the NAP financed by private actors. In France, it’s purely public. It is now monofunded by the Ministry of Transport. Before that, it was by the Interministerial Digital Directorate (DINUM). And the question is, do we make this PAN a data host or a broadcaster, or both? This is a question that we still ask ourselves today. What was decided — and what is still the case after more than seven years of existence — is that the PAN hardly hosts any data, but disseminates it. So that is to say that the data is stored either at the producer or at data.gouv.fr.
Maxime Siret

I didn’t talk about this at the beginning, but transport.data.gouv.fr, which is the name of the PAN platform, is a vertical, if you will, of the more global platform that is data.gouv. On data.gouv, you’ll find a lot of verticals, including transport. The legal specificity in France means that data producers must — this is enshrined in the law — publish their data on data.gouv. And then transport.data.gouv retrieves them from data.gouv and displays them. There is very little data stored with us. The only ones that are stored at home are those that we generate ourselves. There are a few, maybe we can talk about them if you want. So the discussions are based on this: are we going to store everything or are we going to disseminate it instead? As the problem was not on storage but on dissemination, the decision was made to focus efforts on how we disseminate, and what additional services we are going to provide as a PAN to attract users and bring value to this data. So rather than focusing on how we store it, we will rather do how we disseminate it, how we guarantee the data, how we make it reliable, how we highlight it.

Source: Mobility Factory Webinar – Standardisation: tools for the quality of mobility data

Walid: So the fact of developing this PAN specialized in intermodality data, its form was natural, since in fact, it’s just a vertical of data.gouv.

Maxime: Yes, that’s right. Historically, it’s not necessarily in terms of multimodality. The PAN, at the beginning, is really a purely public transport component. Obviously, mobility has evolved between 2017 and 2025. There have been many more bicycles, scooters, carpooling, car-sharing. And so the product has evolved with mobility. Historically, it was buses, trams and metros: we don’t have the data, we have to make it accessible, we’re going to disseminate it via a Marianne stamped platform, we’re going to highlight it and we’re going to try to draw uses and create services around it. That was kind of the context.

State-owned startups

Walid: Do you know how it goes when the DINUM decides to launch this project on the PAN? It’s an internal project, they define a project manager who knows the subject and let’s go? Do you know — because you weren’t there at the time, you told me — but do you know a little bit about how it goes when there’s a theme like that that comes up? How do they actually do it?

Maxime: Yes, I think that at that time, 2017-2018, it was really the very beginning of the DINUM ecosystem, which must have been driven more or less politically by President Macron at the time. And so, they started what is called launching state start-ups. Maybe I’m not the best person, because in 2007, I was still a student… But the goal was to try to innovate in the public service, with the assurance of addressing issues of general interest with an agile method, in a small and interministerial team at the time. So that’s why they chose the word state start-up. So yes, they choose a project manager — product manager, which is called an intrapreneur now. The intrapreneur will have carte blanche on how to address this problem. He will have to defend a budget, say: “ok, I have to solve this and this as a problem, I need such and such resources and funding”, and behind it he will be able to put together a team which, historically and still very much today, is made up of people who are freelancers. So they will look for people on the market who are competent and adequate to solve this problem and try to move quickly, in product mode, a roadmap every 6 months, avoid the ministry’s projects that are a little big and established with periods that go from 2 to 3 years, etc. It was much more short-termist, with a lot, I think, of start-ups that didn’t prove their market and had to close after 6 or 12 months. That’s why it’s what we call a state incubator. Only once start-ups have really proven their usefulness and found their market, are they what we call sustainable. And once they are permanent, then we can decide to invest public money in this issue over the long term, because it has proven to be useful. And so, we make it permanent and we transfer it to its home ministry, with the objective of trying to infuse this agile product approach policy into the ministries, which historically have not necessarily been wired like that. That was the approach.

The project team and its sustainability

Walid: Two questions on that. When you are approached and you start working on the subject, are you a freelancer or do you become a contractor? And the second one, behind it, is: what are the metrics? How do we know that the start-up has achieved the objectives we had set for it and that it is worth investing in and transferring it to its supervisory ministry?

Maxime: yes, first question, I arrived — so I didn’t say it in the intro — but I arrived in September 2024.

Walid: Oh yes, it’s recent.

Maxime: yes, it’s recent, that’s why the historical context on the PAN, I don’t have it 100%. In any case, the rule today is that the intrapreneur of a state public service must be a public official. He is either a civil servant or a contractor. As I come from the private sector, I am a contractor. I have an assignment, I have a 3-year fixed-term contract at the ministry, which can be renewed, but I am not a freelancer. The PAN team is now made up exclusively of freelancers. That’s another subject, but I’d like to see skills internalized, because when a service is made permanent, it should be in the strict sense of the term — and broad, that is to say that the team is sustainable. This is not the case today. Originally, at least a few years ago, intrapreneurs may have been freelancers. I think that today, this is almost no longer the case, or else it is very unusual, it is no longer the doctrine.

And the second question, how do we decide to make it sustainable? That’s up to the intrapreneur, when he creates the start-up, to say: “this is the problem we solve, these are the objectives I set for myself, these are the metrics I set for myself, do you validate them?” And in fact, when I say “you”, it’s because the method of a state-owned start-up is that it works with sponsors. Often, the sponsors are the administration, several ministries or several departments within the ministries. And there is what is called an investment committee, which generally meets every 6 months. Every 6 months, we will go and report to the administration that finances us, saying: “This is what we have been doing for the past 6 months, this is what we intend to do, what we intend to solve, do you approve or not our roadmap for the next 6 months? Does or not what we have done over the past 6 months correspond to what had been set? Is it a public policy?”
Maxime Siret

And based on that, decisions are made for the next six months: do we continue like this, do we reduce, do we accelerate, do we close, etc. And as far as the NAP is concerned, it is a little different, because the NAP has the advantage — or the disadvantage — of being legally enshrined in French law. Today, there can be no questioning of the existence of the product, since the product is by definition integrated into the regulations. So the law would have to be changed to abolish the PAN, which is possible, but it’s not the same implication.

On the other hand, there are other start-ups where you can actually decide to close a certain state, with metrics that are what we call impact metrics. We have to be able to prove that the product we are developing responds to a real problem of the French and it has an impact on the daily lives of the French. Which, for the PAN, in this case, is very complicated. We are not a B2C product. The question of impact, and what we’re really for, how I manage to show that thanks to the work I do, I have X people in real life who will leave their car to go take the bus or take their bike, to go up the chain to that… It’s not that simple. It’s almost impossible to prove that I invested X euros, thanks to X euros I saved Y car trips and CO₂, etc. But for other products, it’s much easier.

Walid: You were at Google Maps, you must have seen the PAN coming. And what has it improved for you as well? That’s something I’m interested in, seen from the other side of the fence. But before we go any further on this subject, I would like to understand what all the features of the PAN are. So there, we talked about the fact that the PAN, it had its primary function, which is to make the data of the public transport data producer available in a centralized place. Does it have other additional functions or is that really its purpose?

Maxime: The primary goal is indeed to put data producers in touch with reusers. Not just public transport. Now, we have expanded, as I tell you, to self-service bikes, carpooling, car-sharing, electric charging stations, etc. So we put producers and reusers in touch. And obviously, this connection is accompanied by added services, because the goal, as I told you, is to be able to guarantee that the data is reliable and of high quality. So our goal will be to try to improve the quality of the data and make it more reliable. If the data is reliable and of high quality, then it will be read and used. It’s like a media library: if there are books, the books have to be well written and well presented so that you want to read them.

In our specific case, how do we ensure that the data is of high quality? We will animate the ecosystem, we will participate in webinars, interact with producers, train them on standards. There are data standards that exist so that everyone speaks the same language. We have to make sure that the data that is transmitted is homogeneous, because if tomorrow you put data that is heterogeneous, and that the data from Bordeaux are different in terms of the structure of Paris, Lyon, Angoulême, etc., behind it, the data aggregator, it will not be able to pick and choose like that if it has to adapt one by one to each producer.

So our goal will be to animate the ecosystem and ensure that the data complies with standards. We have what we call validators, who will validate the data that will be automatically transmitted to us. And it will display a report, and this report will be able to inform both the producer and the reuser, who subscribes to the notification, that the data present with us complies with the standard. And if they don’t respect it, we’ll have to follow up with producers to explain how to change their data so that they can be compliant. We’re the good guys, if you like, we’ve got a bit of a carrot. We are not going to fine them, etc., if they do not respect the law. On the other hand, if they don’t do it after many reminders either, behind us, we can send the police, which is the Transport Regulatory Authority (ART).”

We’re not going to fine. In any case, we have that feature. And then we have others. We have features for reusers, who will be able to subscribe to notifications, to services to be notified as soon as a new dataset arrives, as soon as it expires, as soon as someone comments on a dataset, that’s it. We have several services like that. We also used to have a converter — we don’t have it anymore for several reasons. So, we are trying to animate the ecosystem so that the data is of better quality and more available. And then, we have additional services, obviously, of communication, of ecosystem animation. And on this, it can be several forms. We also have a newsletter. That’s it, the goal is obviously to make the link and encourage the development of data. And for that, we think we’re the right entity to do it.

Walid: change management and community facilitation. And when you just talk about the validator, are we talking about a computer program that runs when you add your datasets, which validates that they comply with the quality of the expected data?

Maxime: it’s an algorithm that will validate. So, if it’s a transport game, it will validate that it’s a good public transport format. If it’s a bike dataset, it will validate that it’s a good bike format, etc.

In which ecosystem does the NAP evolve?

Walid: ok, we’ll come back to the formats later. And there, you talked about ART. In which ecosystem does the NAP evolve? That is to say, the PAN, it interacts — so we have understood that it interacts with producers, it interacts with reusers, it interacts with the ART — does it have other actors with whom it interacts? The State of course… Who are the people with whom, in general, the PAN intervenes?

Maxime: mainly with those who send us the data and reuse it. So the producers will be either private transport operators such as SNCF, RATP, BlaBlaCar, Titi, and co… or it can be — and this is often also the case — the regions, which are called the authorities organizing mobility, the AOMs. So we have both public and private entities that are, in quotation marks, in the lot of producers. And then we have the private ones who will reuse. Here, the system is very varied, and moreover it is destined to develop because we don’t know them well — but we’ll talk about that later. Historically, we have mainly and historically had large players in route planners such as Google Maps, Transit, Citymapper, Apple Maps. And then, we have a lot of small companies that need open data to make innovative services, on which we will also try to support their uses. So we have a lot of private players. And then, at the state level, indeed, it’s quite varied. Mainly, the Ministry of Transport, DINUM, data.gouv.fr and ART. But we also have other players, such as CEREMA, ADEME, with whom we also exchange for projects that are a little diverse and varied, and some associations or lobbies, whether at the American or European level. We have coordination projects at the European level of NAPs, so we also have exchanges with our counterparts in other countries, associations such as MobilityData, which is a kind of data lobby, with whom we also exchange. Because it’s important to work with them. In fact, it’s very rich, it’s very varied. I had a slide on it… Well, here, we’re in audio, but I had like 50 logos, so it’s quite varied.

Walid: Can you explain CEREMA and ADEME in a nutshell, just for people who don’t know? I ask you a following question.

Maxime: I don’t even know if I’m the right person to present them, but ADEME are satellite entities of the State. It is a state agency that allows us to analyse environmental studies to a small extent. CEREMA works with them mainly on the subject of transport data, since they analyse a lot of transport data in France. They also make what is called the PASIM database. It is a database that allows you to list all the authorities organising mobility in France. We need to know, in a given territory, who is responsible for the data, who must publish it, who is in charge of operating the transport. And CEREMA helps us answer this question.

And then, ADEME, we’re less close to them, but we still have subjects to discuss with them. On that note, it still remains on the scale of more prospective with them today. On the other hand, CEREMA is really a player with whom we work, especially on road data as well. There are a lot of topics with them in progress because CEREMA has a lot of history on road data with the DIR, the interdepartmental road directorates. CEREMA acts as an AOM, I would say, on road analysis issues. As they are AOM, they need data analytics and so they come to us to use them. And conversely, we need them to complete the data, so we also exchange with them widely.

Walid: Are you involved in any way in the normalization groups, such as we discussed with Tu-Tho?

Maxime: Yes, we are part of GT7, standardization of transport data. So yes, we’re part of that. And also, at the European level, there is a project called NAPCORE. NAPCORE stands for “National Access Point for Coordination in Europe”. So basically, it’s the project for the coordination of NAPs at the European level. And on this point, there are several sub-working groups that are thinking about how to set up standards at European level, and that they are respected and shared among all the Member States. There is sub-group 4.3, which is very technical, but we are also working on a European scale. Which is very interesting, but also super long in terms of temporality. It’s not the same as at the French level, obviously. We have much more latency on these subjects. So it’s a long time, on that.

Who owns the available data on the NAP?

Walid : On the data that is available and made available by the national access point, the first question I asked myself was who owns the data that is made available?

Maxime : Who owns the data? They are the authority organising mobility, so often it is mainly the regions or communities of municipalities. So it’s to them that the data belongs, but they delegate a service, which is called a PSD — so a public service delegation. The majority of them delegate. Otherwise, we can do what is called a management company. But when they delegate, they will delegate their service to a private company that will operate the transport and generate the data. And then, the data will be transmitted to the organizing authority, which will publish it with us. So it’s up to the authority to publish the data.

Walid: So when we talk about data, we’re talking about data that is rather static, rather declarative, or we’re going to talk about data that is just as dynamic?

Maxime: Yes, that’s a good question, because I didn’t mention it in the introduction, but the scope of the NAP is what we call supply data, at least today. These are not the actual data, but the supply data. So it’s what time the bus passes, how much it costs, how many lines there are. But it’s not going to be how many people got on the bus, how many lines were run on a given day, etc. It’s really the supply data.

Walid: can it be given in real time about buses on such and such a line?

Maxime: It must be, even. It must be both. In the sense of the regulations, it must be both. So we cover the field of static and real-time.

The NAP Data License

Walid: Do you impose a special license that allows reusers to reuse the data? Are you the one who imposes a license to be able to deposit this data on data.gouv.fr, or is there a pool of accepted licenses, and it is the organizing authority that defines the license it will use?

Maxime: Indeed, that’s a point on which we try to fight but in any case, we have a position on it. Indeed, we cannot impose the license. The regulations do not impose a specific license. So it’s up to the organizing authority to choose which license it applies to the data. And indeed, depending on the license chosen, the impact on the reuse of this data will be important. So we, as PAN, advocate open licensing. So, there are several open licenses: 1.0, 2.0… Basically, the open license is the one that is the least restrictive.

Walid: sorry, what is the open license for data? Is it a French license? What’s that?

Maxime: it’s the French license, open license v1.0 or 2.0. In fact, it is a license that does not impose conditions in exchange. That is to say, you can help yourself, and you can’t be asked for anything in exchange. So, in fact, it’s very unrestrictive. That is to say, suddenly, by using this license, we are sure that any entity will be able to take the data, we will not be able to say: “oh well, no, I don’t take it, because otherwise I have such and such a risk”, because the license, in fact, does not involve any risk. On the other hand, there are authorities who choose to put other licenses. Typically, the Open Database License (ODbL) — it’s a special license — or the Mobility license, which, on the other hand, will impose, in exchange for access to the data, a resharing of the data under conditions that can be more or less restrictive, and therefore more or less handicapping for certain personas of reusers, and therefore potentially be a brake on access to the data. So for example, big players like Google Maps are not necessarily going to integrate data that is under this or that license, because behind it, for them, it would represent a risk in terms of data, since they could be asked in exchange for data that would be in a much wider scope of application than the one for which the basic data was collected. So there are risks like that that are considered by people, and so we are trying to push for an open license, since we are convinced that if the license is open, the data will be widely shareable, and therefore we will have more and more impact and reuse behind it.

Walid: For listeners who want to know more about ODbL licenses, we talked about these topics. You can refer to episode 1 on Open Food Facts and the OpenStreetMap presentation episode. You can find out more by listening to these episodes. End of parenthesis. So, there is a licensing issue. This was one of the questions I had asked myself when watching the various seminars on the NAP. I wondered: did you have any means, and if it was persuasion, and also what arguments did you put forward? Because typically, I’m thinking of the argument that your data is accessible in Google Maps. Whether this is an acceptable argument for an organizing authority or not, I don’t know.

Maxime: I can answer directly, but yes. The smaller the organising authority, the fewer resources it has to send this data to third parties. And so, for an organizing authority — it doesn’t matter which one, but a small organizing authority — the fact of making the effort only once, that is to say to make the effort to put the data in our country, if it puts the data under an open license, it will have the guarantee that behind it, the data will be disseminated everywhere in the ecosystem. And so, rather than going to call SNCF Connect, Apple, Google, Transit, Citymapper, etc., it will do so once it is here, and behind us, we will play this role of broadcaster. And so, in terms of economy, and especially visibility, the guarantee of the open license is to have behind it that is distributed everywhere. If you have a more restrictive license, you don’t have that guarantee. And so, that’s an argument that we obviously often use to convince producers to open up their data as much as possible.

Data formats supported by the NAP

Walid: I’ll put a link, for those who are also interested, to a fairly recent project, with a link to FOSDEM conferences on a project called Transitous, which is a community service, which allows you to make an alternative, a rather promising open source routing system, which reuses data from the different European NAPs as well. I will put a link to a FOSDEM conference that explains all this. I will close the parenthesis once again.

To talk about formats, earlier you quickly introduced the fact that there are different formats. What format do we have — like, let’s say, large formats — to manage mobility data, and which ones are accepted on the PAN?

Maxime: So, we have to differentiate between European standards and industrial formats. Historically, there are industrial formats. For public transit, the industrial format — the one used by the industry at large — is the GTFS. So GTFS, correctly, the G is for Google, because Google created it. Now, it’s become open, and it’s open data, it’s really free to access. The G has been replaced by “General”, so General Transit Feed Specification. So it’s an industrial data format. It is not the data format that was chosen by the European Union, for several reasons, which is the NeTEx format. In the legal sense, we have the obligation — and it makes sense — to promote European standards.

Behind us, we are in contact with the ecosystem, we also need to listen to the needs of producers and industrialists. And so, our priority remains to have an impact, and in the broadest sense, to decarbonize mobility. So we, the NAP, until now, have always had a two-pronged approach: to do the best it can to reference European regulatory formats, and to accept industrial formats, by encouraging the industry to switch to regulatory formats, because there are interests in doing so beyond just ticking the legal box. So we do both: we reference industrial formats and European standards.
Maxime Siret

I put GTFS, NeTEx for public transport. It will be DATEX II for EVSEs.

Walid : What is EVSE — what is it?

Maxime : EVSE is charging infrastructure for electric vehicles — so it’s electric charging stations. GBFS format for bikes — so there, it’s instead of T for Transit, it’s B for Bike. So it’s the same type of format, let’s say, but to map the offer of self-service bikes. Then, we have formats that are a little more in-house depending on the type of data, where there is still a “need” for standardization. I’m thinking in particular of cycling facilities, for example, or parking, which are still very unclear between what we put in NeTEx and what we put in CSV. So when it’s not very clear, the NAP can have this role of participating in standardization groups to go and establish a plan. This is what the PAN did, for example, for charging station data, which created a Franco-French scheme, because at the outset, it was not standardised at the European level. And then, once it is standardised at the European level, the PAN will tend towards this standardisation and either change the format altogether, or put in place tools to convert formats from an industrial format to a regulatory format.

This is what the PAN has done, for example, historically on transport, where everyone sent in GTFS when NeTEx was needed, and the industry was not ready to provide NeTEx. So the PAN helped the industry by providing a GTFS → NeTEx converter, which has since been removed to encourage producers to natively produce in the regulatory NeTEx format.

The advantage of using the NeTEx format

Walid: I think NeTEx, we’ll come back to it in more detail in a future episode. But earlier you said that there were advantages to using the NeTEx compared to the GTFS. Can you explain in a few words what advantages there are that require data producers to change the format in which they send the data?

Maxime: Yes. First of all, it should be noted that the NeTEx format was not created strictly for the same use cases as GTFS. The GTFS is really a format for passenger information. NeTEx is a data format that allows transport operators to better manage their services, and not just to manage passenger information. Île-de-France Mobilités, the organising authority for mobility in the Île-de-France region, uses the NeTEx, and yet, behind it, it will supply and export both GTFS. So it’s not necessarily a use. The two should not be pitted against each other: the two are complementary. On the other hand, when it comes to modeling a transportation offer, if we really want to be purist and model 100% of the transportation offer, we will not be able to do it — at least not today — with the GTFS. In particular, everything that is accessible and price range. In terms of price range and accessibility, on the GTFS, we are quite limited today. Whereas on the NeTEx, we can tell if the ramp is 8% or 5% to get off the bus, how many meters it is in length, width, etc. Things that we can’t necessarily model. But let’s say that for passenger information, the GTFS probably meets 95% of the needs. And the remaining 5%, if we want to go and get them — but the data still has to be existing — we’ll have to go to the NeTEx, I don’t know. So that’s kind of the difference. But really, the two are not opposites, the two are complementary. And what we think about the NAP is that in the long term, we would like the data to be all produced in the rich NeTEx format, and then there is a GTFS export for traveler information needs. Because the GTFS format has advantages in terms of file size, it is much more flexible, much easier to handle, and it responds to many very simple uses than the NeTEx is perhaps a little less accessible to the average person. So we are aiming to have this NeTEx scheme → GTFS → passenger information applications.

Is all the data public? Are some of them paid?

Walid: ok. To finish on this subject, two questions. The first is: is all transport data public data, or is there still data that is private and that is not published, or that cannot be published? Is there any data that is also paid?

Maxime: good question. When I saw this plot, I told myself that I wasn’t necessarily going to have all the answers to this question. Of course, not all of them are public. Those listed in the regulations, obviously, are intended to be public and they are increasingly so. But for questions of professional secrecy and competition between operators, there is still data that is unfortunately not necessarily open. I am thinking in particular, for example, of everything that has been achieved. So the number of buses, how many buses get on at which stop, at this time… That’s something, it’s a fact that is not necessarily shared by the incumbent operator. For example, the operator that operates the buses in Bordeaux, tomorrow, there is a call for tenders, there is an incoming operator who wants to arrive to win the contract, but he will not necessarily have access to this data. And the vocation is to open up this data as much as possible. There are projects on this, in particular the OPRA project, but I don’t know much about it yet. The idea is to open up this data as much as possible. But indeed, even today, this data is not necessarily public. And so, if we want to have access to it, we’re going to have to pay.

Walid: Does she have access to the AOM?

Maxime: Yes, the AOM, it pays for that.

Walid: It could very well include in its calls for tenders that this data be public, or is it because it doesn’t want to do it for certain specific reasons?

Maxime: So, I’m not an expert in calls for tenders, I don’t want to talk nonsense. But I think that perhaps the operators do not necessarily give all the data to the AOM. There are some that they manage to keep, because it’s their professional secrecy, and it’s in their interest. Indeed, there are always cases of this. On the other hand, on the PAN, we only find free data. We don’t have paid data on the PAN. We are really on the PAN of open data. If then, for paid or more confidential data — I am also taking advantage of this for the audience — there is the Eona-X project. In fact, it could possibly be a topic for your next podcasts. [See the EONA-X conference of the Fabrique des Mobilités]

Walid : He’s on my list too

Maxime : Okay, that’s it. There’s Mélanie Vessier, who joined Eona-X — you could call her if you want, for example — who is DGITM. To make it a little restricted and confidential, we can create a “data space” where we can have a private exchange of data between players and industry, without it being open data. And that can be the case for sensitive, security, commercial data, etc. So in this respect, it’s a bit like the complement of the PAN, confidential version.

Concrete examples of reuses made possible by the NAP

Walid: ok. What I would like us to discuss are a few concrete examples. Because we talked about the fact that the ultimate goal was for the data to be available, for reusers to take it, and for it to improve the system so that there is a modal shift, and basically, at the end of the day, we reduce the CO₂ that we send into the atmosphere. Do you have any concrete examples to give us? So it can be in your experience, or examples of reusers, and the interest of the PAN for them, and therefore ultimately for the users, the end user of transport.

Maxime: The most striking examples are those that will have the most impact on the French. And so, often, we talk about the big operators of multimodal applications — Google, Apple, etc. Obviously, we have much more local use cases, with a more limited impact. If we talk about those that have the most impact — especially Google Maps, I was with them before, for 4 years — it’s very clear about that.

Before the PAN existed, an operator like Google Maps, in France, in 2018, had to reference perhaps a maximum of ten cities in France for internal reasons. That is to say, you had ten cities in France where, when you made an itinerary from the station to the town hall, for example, you went to the public transport tab, they told you: there is no way, you have to walk, take the taxi. What for? Because an operator like Google, which is global and global, did not have the necessary resources — and potentially, at that time, it did not see the added value — to dedicate energy to contact all the transport operators one by one.
Maxime Siret

Imagine, tomorrow, Google wants to reference all public transport around the world. For example, France. If we had to contact all the transport operators in France one by one, we had to staff a team — and a team of many people — and put a lot of resources in front of that. This was not Google’s priority in 2018. So, there was no data. So in concrete terms, what does this mean in terms of the impact of the French? It’s that a resident of Clermont-Ferrand, in 2018, he was at home, he wanted to go to the station and then go to Paris. Well, in fact, he could see that there were no buses. Whereas the bus existed. But if he wasn’t aware of that, potentially, he wouldn’t take it and he would take his car.

With the arrival of the PAN, players like Google and others have changed their approach. Because, as a result, rather than going to talk to each operator individually, they have an interlocutor — who is the PAN — who does the aggregation work for them and in exchange, they can integrate a lot of data very easily. It was my job before, so I can talk about that. So in the space of 4 years, we have gone from a dozen cities to more than 130-140 networks — now even more — available in cities in France. And a concrete use case is when I arrived in 2020, in Lyon, there was nothing. While Lyon… we’re talking about Lyon, after all. There was nothing. So they told you: you wanted to go from Fourvière to the station, they told you: walk, take your bike. Well now, there are public transport schedules. And I could see the number of requests. The number of queries per week on Google Maps in Lyon is in millions. So you can see the impact it can have on the French. You say to yourself: “well actually, out of the millions of requests, maybe I managed to convert 1% who finally took the bus rather than their car”. But 1% of millions each week, on the scale of France, is a lot. So it’s a very concrete impact of what the NAP has been able to bring on a large scale.

And then, on a small scale — and it’s less B2C, but it is in the end — that’s where it’s hard to show our impact, is that a lot of VSEs and SMEs have been able, thanks to our data, to carry out analyses and sell ancillary services to local authorities. For example, to tell them: “Look at how many bus lines there are in your area, at what time they pass, we could perhaps optimize it and put more buses here and fewer here, etc.” And so, afterwards, these services are provided to local authorities. The local authority, behind it, has been able to change its service, put more buses there, fewer buses here, and therefore, ultimately, improve the comfort and service provided to their citizens.

[On this subject, see the episode on transition.city, from Polytechnique Montréal]

These two examples — the most striking is obviously when you take reusers like Google Maps, you can see the impact directly. When I was with them and I was working for a month on the data of Clermont-Ferrand, when I launched Clermont-Ferrand on Google and I saw that in the space of a week, I had 600,000 queries — the figure is a random number, but it’s a bit like those orders of magnitude — 600,000 queries in one week in Clermont-Ferrand. I said to myself: “ok, I have 600,000 people who have made a simulation of a public transport trip in Clermont-Ferrand. Before, they had nothing. Maybe out of the 600,000, I informed maybe a tenth of them that now, they are aware that there is the E4 bus that passes at the stop below their house, and that they can take their bus.” This is a concrete example of what the NAP has been able to change in terms of access to passenger information.

Ways to compel producers to share quality data

Walid: The question I ask myself, which is a bit complementary to that: are there producers who take too long to submit their data, or who do not want to deposit their data? In that case, you said that you pass the file on to the ART, which then, behind it, can do things. I wondered what could be done in a concrete way to force a producer to share his data, and especially to share quality data. Because in the end, if he shares data but it’s not of quality, even if you want it to be of quality, in the end, the impact is less.

Maxime: Yes, that’s clear. That’s where the complexity of our work lies: it’s that we’re not producers. No matter how hard we try, if in front of us it doesn’t respond, it doesn’t respond. I still have cases, even today, where I have producers who have not responded for 6 months. In fact, we can’t move forward. In these cases, what happens is that behind it, indeed, when a producer doesn’t collaborate — so it can be voluntary or because he doesn’t have resources, it’s not just ill will — most of the time, it’s mostly: “I don’t have time, I don’t have resources, so let me go of the cluster with your data stories, I run my buses, that’s not it.” In these cases, we transfer the file to the ART, which will send reminder letters. She’s going to send a letter, she’s going to wait 6 months, she’s going to send a second letter… And so it’s a long time. Indeed, in the end, after a few reminders — I can’t tell you the precise rules because I don’t work at the ART — but after a few reminders, the threat is that there will be a fine. I can’t tell you if there has ever been a case where we have reached the point of a fine, but often reminder letters are enough and allow the situation to be unblocked. And I would even say that often, we don’t have to get to that point. Our work of supporting and animating the ecosystem allows us to unblock things. There are always recalcitrants, but most of the time, that’s enough.

There is also what we have put in place, which works well: it is a kind of public social pressure. In fact, when a producer puts in data — I’m talking about quality — I don’t know, Angoulême, I have nothing against them, it’s not the case, but let’s say that Angoulême sends me their data, their data is rotten, nothing to do. They will be published anyway. We will display a message, an error report, which will clearly indicate that the data is rotten. We will historize them, we will be able to see that over time, they are also rotten. And behind it, there will be a kind of public commentary, where the data ecosystem will be able to comment. So maybe Angoulême, when they see that Tartampion, a researcher or a student, says it’s not right, they’ll ignore it. But when they see a comment, two comments, three comments, five comments — and it’s Google that comments, that Apple comments — by dint of it, maybe they’ll say to themselves: “Ah, but actually, it might be worth it for me to move and correct.” So there’s a kind of public display that can work. Obviously, we’re not here to show all the bad players, but sometimes it works. Naturally, we have a producer who sends bad data. Directly, we have three people who comment, who say: “there, it’s not okay”, and then the following week, they correct.
Maxime Siret

The role of the NAP as a link between producer and reuser

Walid: ok. It’s something I’m thinking of mentioning, but that we haven’t explored before, and that’s the relationship with reusers. So what I understand is that you serve as a single point for the reuser, to comment, and you bridge the gap with the producer. So the reuser can leave comments for the producers, whereas before he didn’t do it at all, or he might have done it, but it was email exchanges, well it wasn’t necessarily public, you know.

Maxime: Exactly, yes. The goal is indeed to provide the platform for reusers. They only have us as a platform, and so for them, it’s easy to give feedback. They do it on the platform directly. Behind it, either the producer plays the game and responds to the reuser directly, or sometimes he ignores it. And in these cases, if he doesn’t know, we monitor it, and we’ll be able to go and follow up with the producer ourselves. So there are both, indeed. But yes, we make this link between the two. And that’s often how we manage to incrementally increase the quality of the data. It’s that at the beginning, the data won’t be very good, and then there will be a return. And then there’s Michel who will say that line 3 is missing, and then there’s Cécile who will say that the line is not red, but pink. And then there’s so-and-so who will say that… And gradually, over time, the returns will accumulate. And this feedback will be integrated by the producer, who will also share it with the entire ecosystem. And we’re going to create a little bit of this virtuous circle of the value of open data, by having this sharing and feedback from each person, rather than doing it in their own corner.

Walid: I think you mentioned it a little before, but so we can say that there are more reusers now that there is the PAN than before, then?

Maxime: We can say that there are more, that’s for sure. Unfortunately, we still can’t say how many. This is also a topical subject. Unfortunately, this is the disadvantage of open data, in the strict sense of the term as we have it today: it is that we do not have an exhaustive list of all the people who come to us. You, tomorrow, come to us, you download the data from Clermont-Ferrand, and you leave. In my system, I will have “+1 download” of the Clermont-Ferrand data. I wouldn’t know it’s Walid, from the podcast Projets Libres, and that you want to use it this or that way. Unfortunately, I won’t know that. I have a fairly broad vision of who comes to us, but even today, I don’t know all the customers of my data, because they come from me. This can be a concern when it comes to proving our usefulness internally: what we are for, what problems we solve, and how much money we need to solve these problems.

Walid: Do you encourage people to register you somewhere to say they use it? How does it work?

Maxime: So, we don’t do surveys, but we do encourage reusers to declare themselves as reusers. In fact, on our platform, we encourage people to declare that they are reusers, and then we encourage them to declare their reuse. So if you go to transport.data.gouv.fr, you have a “Reuse” tab, and you can see a bit of an example, an overview of all the data, of all the projects that have been carried out thanks to the data available on the NAP. But these are not all projects, these are the projects for which people have been willing to declare them. So yes, that’s maybe 10%, 5%, 15% — we don’t know — of the extent of what could… After that, maybe that’s also the magic of the thing. It’s that there’s always an element of mystery, and we won’t really know 100% what we’re for.

Relations with the academic world

Walid: To finish on this subject, there is a question that I hadn’t noted but that comes to me when we discuss, and that is: what are your relationships — you’ve talked about researchers several times — what are your relationships with researchers, with universities, in short, with people who work academically on data? Do you have reports, do you have projects with them, do you know what they do with the data they consume?

Maxime: Not enough. No, not enough. We don’t have many projects with them today. Historically, the target of the NAP was really the big route planners, because they were the ones that had been identified as the ones with which we were going to have the most impact quickly. Efforts have been turned towards them. Today, I think we have a vocation to expand. And we have three — so I’m changing the question a little bit — but we have a bit of three personas today. We have the big reusers, those who are experts in transport data. So it’s going to be either Google, Apple, or research firms that specialise in public transport that come to us. That’s a persona for which — which is a bit like the historical persona of the PAN — on which we have invested a lot.

After that, there is a persona a bit in between: these are people who work in transport but who are not data experts. In this, for example, you can have researchers or academicians, who are not necessarily experts in the GTFS format, but who have a transport project. For this transport project, they will need – episodically – this or that transport data. And so they’re going to have to take an interest in our product.

And newly, — and it’s happening more and more, it’s interesting — we have a third persona, where we have players who are totally outside the world of transport, but who, for projects, very specific needs, will want to integrate transport data. Typically, recently, we have had real estate agencies. A real estate agency, which, in fact, on its website, puts houses and apartments for sale and, to enhance them, it would like to automate the registration of public transport stops close to the properties they put up for sale. Because we know today that a property that is close to public transport is more valuable. So she would have liked to automate that. So, she comes to our house to pick up the transport stops, and display them on her map by putting “the house is close to the E4 stop”. We have things like that, which are evolving. On academics/researchers, historically — I wasn’t there at the beginning — but I don’t really have a specific example with this category of users.

The future of the NAP and its challenges

Walid: The last subject I would like to mention — the same, you talked about it a little bit at the beginning, in the intro — is the future of the PAN. Before we leave, I would like us to talk about the future of the PAN. What is your feeling? Where have you come at the moment? Have you arrived at a result that is satisfactory — we can always do better — but do you think that the result you have achieved, in terms of production, quality of data, is a satisfactory result? And is there still a long way to go? Then after that, it’s like: what are the challenges? What’s happening in front, the big stuff — whether it’s in terms of data, in terms of organization, funding — what can you tell us about that then?

Maxime: The answer — well, on what we have today in the NAP — the answer, it really depends on the mode of transport we’re talking about. The PAN, historically, has been dedicated to public transport. So today, public transport, we haven’t “finished the game”, in quotation marks, but we’re at the level 90 out of 100. That is to say, we have mapped almost all public transport in France. There is still work to be done on the quality and on the data carried out.

I didn’t say it at the beginning, but we have three verticals in the PAN to analyze what we do. It’s quantity, quality, and reusability. If we have data that is of quality and quantity, then it will be reused. For the PAN, today, for public transport, we have a lot of quantity, we have a lot of quality, so they are reused. But we could do better in terms of quality.

On the other hand, for the other modes of transport, which are much more emerging and much more recent, we are not at the same degree of maturity at the NAP. What for? Because the ecosystem is not either.

You take the ecosystem of scooters and bike-shares — until a year or two ago, the market was jungle, if you will. That is to say, you had Lime, Dott, Tier… Everyone arrived at the same time, every six months there was one who changed, one who sank, the other who was redeemed, etc. Inevitably, when the ecosystem is so immature, behind it, the data, it will not be mature either. So on these modes, a little more emerging, we are not yet at the final level of the game. On this, we are still looking for data. In VLS – self-service bicycles – we are still missing about 20% of the games, 20-30% of the flow of bicycles.
Maxime Siret

These are networks such as Vélib’ or free-floating networks such as Lime, Dott… So on this, we are not done because we still lack data. It’s the same thing for car-sharing, for carpooling. And then we have even newer, even newer modes — such as electric charging stations. Charging stations, every day in France, you may have hundreds of new stations that are created.

Walid: I can confirm, the one next to my home has just been put into service today.

Maxime: When you have this degree of maturity of evolution, it’s complicated for us to follow, if you want, this channel. So on this, in terms of these modes, we still have a lot to do, because we have to go and animate the ecosystem, we have to go and meet all the players who change overnight, we have to go and normalize the data and then we have to aggregate it. Because for charging stations, we now have more than 1800 different producers. So we have more than 1800 different data sources, and we are going to aggregate the 1800 sources to provide a single file that will consolidate all the terminals into a file, since what the reuser wants behind it is a single file.

So, to come back to your question, the challenges of evolution are very different depending on the mode we are talking about. Today, we still have a lot to do in terms of new modes. And probably the new modes will change and evolve again in the next few years, so we’ll always have a lot to do on that, that’s for sure.

Then, on the upcoming developments and the major challenges: for me, the biggest challenge is to better understand our uses.

In fact, it’s historically complicated to develop a product and legitimize it in terms of funding to the administration, if you’re not able to go to your decision-makers and tell them concretely what you’re for, who comes to you and why. And unfortunately, at least for me since I arrived, that’s kind of what I’ve been up against. They say to me: “ok, how much are we going to give you? Are we going to give you a million or a million two? Ok, and the 200,000 euros, what will they be used for? Are you going to go and see who will come to your house?” But that’s complicated when you’re not able to say who will come to your house — at least not in a very readable and exhaustive way. So there are issues behind this that we are in the process of putting in place.
Maxime Siret

And then, there’s another issue, which is very recent — and I think that by the time you release this podcast, we’ll be able to talk about it, so there’s no problem — which is that there is an issue of consolidating NAPs in France. There is the multimodal PAN, which has been talked about since the beginning. There is the road NAP. The challenge will be to merge these two NAPs and to make transport.data.gouv.fr the single national access point in France for all transport data, including roads. And so, there will be a challenge for the PAN to expand to road data, which are not necessarily the same types of data, not necessarily the same interlocutors. So on that, we’re going to have to start from scratch a bit. So there are also these stakes, which are strong for the future.

And along with that, the issue of financing of course. You know that we are in a very unstable, very complex political situation. Public funding is obviously pointed out and analysed in detail. And so, since I arrived, I’ve already been cut from the budget. So there is a strong challenge to convince decision-makers that public money would be — and will be — usefully invested in the NAP in the long term.

And so on this, for that, we come back to the problem of “who comes to us”, etc. So we have this strong challenge of financial sustainability of the product, because today, we are working a little by sight. Me, you see, I still don’t have a budget for 2026. I don’t know how much money I’m going to have, if I’m going to have to fire two people, or hire two, if I’m going to be able to prioritize my work on EVSE — well, on charging stations — or not. You see, there are a lot of issues that are unresolved, and that’s every six months. And as a result, it’s complicated to have a long-term vision when your short-term vision is a little blurry. So there are strong issues at stake in this too, in terms of sustainability.

Final Words

Walid: Okay, we’re coming to the end of the interview. I’ll just let you, before we leave, a final word. Do you want to get a message across, even if you just sent an interesting one just before, but do you have a particular message to get across before we leave?

Maxime: Particular no, I want to thank you for the invitation. It was very interesting to be able to present what we were doing at the PAN, which is a product that I find super interesting, and which meets a large number of needs, and which is not necessarily known. So I hope that it can call for other initiatives. If I have a final word, it’s that if the headphones have suggestions, they should come back after our exchanges and say to themselves: “but in fact why doesn’t the PAN do that? It’s stupid actually, it’s missing this, this and that, I don’t understand”, well I’m interested in having this feedback. Because we’re a bit in the handlebars on a daily basis. And sometimes, to have feedback from people who are not necessarily initiated, it feels good. And sometimes, we have ideas that fall thanks to that. So I’m interested in having this feedback from your audience.

Walid: Where do we contact you?

Maxime: The best way to contact us is contact@transport.data.gouv.fr. Generic address, but one that is consulted by everyone on the team, and I also have access to it. So by sending an email to that address, we will answer you very quickly and we will be available. We’re also available, we also have a platform on LinkedIn, where we communicate very regularly — so the public page transport.data.gouv.fr. And finally, we have a monthly newsletter, on which we communicate our work publicly. This is also an important value of the NAP: it is that everything we do is public. Our budgets are public, our works are public, our roadmap is public. So we share everything on a monthly basis. So if you want to follow our developments, you can subscribe to the newsletter which is accessible via the home page of transport.data.gouv.fr.

Walid: great. Well Maxime, thank you very much, it was a real pleasure. For the listeners, as usual, if you liked it, don’t hesitate to circulate this episode, to talk about it around you, to put comments on social networks. You can find all the information in the footer of the site or in the episode notes on your podcast platform — a podcast that is available on all the right platforms. So there you have it. And see you soon for other episodes on open data. See you all soon. Maxime, thank you very much, and we look forward to seeing you, in real life, at an event. We missed the transport meetup a few months ago, but here it is…

Maxime: It’s a small world, we’ll meet again without worry. Thank you Walid.

Walid: See you soon.

To go further

Peetube channel of Open Transport videos that inspire the episodes of Projets Libres: https://videos.projets-libres.org/c/transports/videos

Episode production

Remote check-in on October 27, 2025
Plot: Walid Nouh
Editing: Walid Nouh
Transcript: Walid Nouh

This article has been automatically translated from the original language into English.

License

This podcast is released under the CC BY-SA 4.0 license or later

[Transport] The PAN: centralising transport data to decarbonise our journeys