Eric Nordling, heads up Expretio, a revenue management solutions specialist serving the rail sector in Europe and North America. He brings more than four decades of experience as both a user and designer of revenue management software. He most recently headed the pricing department at Amtrak, where he developed a pricing system which created the fares that were fed into Expretio’s Appia platform, and was subsequently recruited to join the company.
Sam Sherwood-Hale spoke to Expretio’s General Manager Eric Nordling, who has spent more than 40 years in revenue management across aviation and passenger rail, about the fundamental principles of ‘the RM discipline’, what rail can learn from aviation, and why the conditions are finally right for UK operators to make the most of it.
SSH: You’ve spent more than 40 years in revenue management. How has the discipline evolved, and what lessons from aviation are most relevant to rail today?
EN: The first thing I would say about how revenue management has evolved is that as computing power has improved, so has revenue management. In the early days, we had a limited amount of data and fewer than ten algorithms for forecasting. Now, with more modern computing technology, you can consume a lot more information, gathering years and years of data and looking at bookings much more rapidly. In terms of forecasting and optimisation algorithms, instead of eight or nine, we are now talking about hundreds.
If you only have eight or nine models forecasting versus a hundred, the chance of getting a better forecast is much higher with the hundred. And the models in these programmes are now far more sophisticated. They can even adapt on a continuous basis, and that kind of capability simply did not exist forty years ago.
The fundamentals of revenue management are the same in rail as they are in aviation – but the two industries got there by different paths, driven by different pressures.
In aviation, the catalyst was capacity. When Boeing introduced the 747 in 1970, aircraft size increased by roughly 400 per cent overnight — from around 100 seats on the 707 to 400 on the new widebody. Airlines suddenly had far more inventory than they knew how to sell. It was in fact a British carrier, BOAC, that first responded, introducing excursion fares in 1973. The logic was simple: offer a meaningfully lower price to passengers willing to commit early, book a return, and stay for an extended period. Lower prices stimulate demand, and that demand fills seats that would otherwise fly empty.
But there was a catch. Selling too many discounted seats left no room for the business traveller who books late and pays full fare. Airlines had to learn to predict late booking demand – and hold back enough inventory to serve it. That tension between selling cheap early and protecting space for premium late bookings is the foundational problem that revenue management was built to solve.
Very basically, all revenue management is understanding what the demand curve looks like for every itinerary. If you offer a high price, the quantity demanded is lower; offer a much lower price, and the quantity is much greater.
That idea of switching from one price to multiple prices is the fundamental underpinning of revenue management, and the revenue benefit is substantial. A single price of £20 across 60 seats generates £1,200. Offer multiple prices with fixed allocations and that rises to £1,370. But if you can truly understand the demand curve and adjust your allocations intelligently to match it, you can reach £2,040 under average demand conditions. The concept of offering multiple prices exists in airlines, hotels, rental cars, other industries, and of course in rail. And for rail, you have to understand these demand curves for every itinerary possible.
Governments are also under pressure to reduce subsidies, so the timing is aligning right now for UK rail to really take advantage of this technology, to increase revenues, reduce subsidies, increase ridership, and provide more satisfaction to the citizens of the UK. I believe that what will happen with fares reform is that there will be lower prices overall, and more people will be able to take advantage of rail travel. The timing is right.
SSH: Those algorithms are now so much more powerful, with hundreds available where previously there were ten. Are they still using the same underlying format you have just described?
EN: Yes. The underlying microeconomic concept is: if you understand the demand for a particular itinerary really well, and by that, I mean the demand for that itinerary on a specific departure, you can act on it. On a high-demand departure, the demand curve is shifted out to the right, so no cheap fares are offered because you can fill all the capacity with higher fares. On a low demand departure, you offer lots of cheap fares because the demand level is not there for the higher ones. Understanding all of these demand levels using large numbers of models helps you to set the right price.
SSH: Can you talk about how the complexity of routes and itineraries evolved in aviation, and what that means for rail?
EN: People want to travel all over the planet, so airlines created hubs where planes come in, exchange passengers, and go out again. Any given plane may have 20, 30, 40 or 50 different itineraries on it. A Boston to London Heathrow flight, for example: those passengers connect to 20, 30 or 40 other destinations. Airlines have to balance what people are willing to pay to travel to all these various places and decide which bookings to accept.
In rail, it gets more complicated. A train going from London to Edinburgh with 40 stops has over 800 different combinations of itineraries, 800 on a single train. One of the things I am impressed with at Expretio is that it actually forecasts each of those individual 800 itineraries and considers which ones would result in the best service to customers. If there are empty seats in one portion of the train, lower prices can be offered to stimulate demand on those empty sections. If the train is very full leaving London, it makes sense to prioritise the passengers travelling the longest distance over someone going just two stops.
All of this forecasting uses models to accurately predict where everybody wants to go, and then additional models to balance all these different itineraries, to determine which to accept and at what price. Rail gets to be a little more complicated because there are simply so many more itineraries per departure.
SSH: That complexity has been there for longer than this technology has existed. Is that why Britain has an opportunity now, because we have essentially been dealing with this problem poorly for a long time?
EN: Exactly. A simple example: if you have a train departing London heading towards a football match just a few stops away, there is enough demand there to fill the train with just those short trips, and then the train could be empty all the way to Edinburgh. In the old world, you did not have the ability to control all those itineraries. In the modern world, you do have the ability to predict them, control them and optimise for them.
SSH: As GBR comes together with new people and new structures, are you concerned that they might lean too heavily on AI in revenue management specifically?
EN: The AI that is out there today, these modern algorithms, are helpful and impressive. But at the end of the day, the human brain is still more capable. An experienced person who has worked in this field for years will see the patterns, know what is right and what is wrong, and understand the bigger, broader perspectives. They are still needed to reliably supervise these systems and make sure nothing is going awry. You definitely need the systems to keep track of everything, but you need the human judgment to keep everything in check.
AI can be very helpful, as long as it is supervised. If you simply say to AI: ‘do whatever you want’ that is a dangerous proposition. The human still needs to be in the loop. But an experienced human with years of knowledge can be more productive and check the veracity of AI much more readily.
SSH: Some have argued that consolidating everything into a single integrated platform is the most efficient approach to data management in rail. What is your view on that?
EN: An advantage of having everything in one place is that it is much easier to move data around. The disadvantage is: what are you doing with the data? If you are moving it around quickly but not feeding it into models that are forecasting 800 itineraries, if you are just approximating, what good is the rapid data input if you are not doing anything useful with it? The question you need to ask is: am I getting the best possible solution for this particular problem, or am I just getting speed? Revenue management is fundamentally about predicting the future, not reacting to it. The current enthusiasm for real-time data, while understandable, risks missing the point – by the time you’re reacting to what’s happening now, it’s often too late to optimize.
SSH: In your experience, what are the risks of treating revenue management as just one feature within a broader system?
EN: I go back to my business school days. My marketing professor said that in managing commercial affairs, the four most strategic and important considerations are the four Ps: Product, Price, Place and Promotion. For rail, the product is the itinerary. Where are you going to take me? Next is price, which is critical. What are you going to charge me? And pricing is what revenue management is all about: understanding where people want to go and what they are willing to pay. In any enterprise, the price you are offering consumers is often the most important, or second most important, criterion in their purchase decision. To treat revenue management as some kind of minor task to accomplish really belies the fundamental strategic importance of setting the right price. Place is distribution, and Promotion is advertising.
SSH: How should operators be thinking about pricing across different passenger types and travel patterns?
EN: You pay more for a quality item because it is going to last longer and produce better results. You pay more to be on the train at peak time because at the other end, there is something important for you to do. If you say it should be one low price for everyone, you end up with jam-packed trains and people cannot get on.
The trick is to find the happy median, a normal price that helps manage demand on peak departures, with plenty of discounts from that to help fill less popular services. One of the things we did when we updated the prices at Amtrak was thinking about small group travel. When individuals travel by themselves, the range of price appetite is quite broad. It could be business travelers or Wall Street types travelling between Washington and New York with big incomes and expense accounts, willing to pay high fares. But there are also families who have to figure out how to get between cities and cannot afford buying five tickets at higher prices. For these more price-sensitive passengers we offered automatic discounts for small groups.
So, we created a type of fare where, if you have more people travelling together, the price automatically goes down. We recognised that a family of five could potentially drive, so let us make the train a better value and make it more likely they will choose train over car. We built in automated discounting for larger parties. A family of five does not need to be on the seven o’clock business departure anyway. We can give them a low price on the ten o’clock departure instead. Let the business people go at seven and the families go at ten, let the business people pay a fair price and let the families get a really good deal.
My impression of UK rail is that there are a lot of empty seats. Sure, there are peak departures, but there are a lot of empty seats. And my impression is that with fare reform and good revenue management, there are going to be a lot more passengers, HAPPY ones at that, because the prices are going to go down. Ultimately, lower prices multiplied by more riders equals more revenue.
