Analytics for trains

Like Jeremy Corbyn(sic), I was consigned to the floor of a Cross Country train for 3.5 hours today. For a sizable fee of £175, I was locomoted between Reading and Manchester in some discomfort. And it really gets you thinking.

How can a fixed capacity train service be sold to overcapacity? The demand is of course peaky, very peaky. But then so are networks of parcel distribution (at Christmas), mobile phones (on new years eve) and swanky bars (on a Friday night).

Despite the advanced analytics on offer today, these capacity-critical services all seem to apply the same rule... "Deliver the worst service we can get away with".

Mostly this is driven by good business sense, but also by pure supply and demand. Train Operating Companies (TOCs) are contracted under the 'Cap and collar' model whereby if they lose money the government bails out a good proportion of their losses or, if they .ake money, takes a cut of their profits. Profits or losses vary widely across the rail network. In 2015 South Eastern Trains amassed £803 million in profit. Paying their shareholders a dividend of £25 million. And this was achieved with an average seat occupancy of 26%. Imagine how their profits would look if they could bump that up to 30% of even a third.

The problem here is that the contracts TOCs sign include a minimum service provision, which forces them to run off-peak services at a loss while trying to make up for that by stuffing peak time services to the gunwales. This means misery for daily commuters and the occasional traveller-on-a-schedule.

But is there a better way? Erm... have you been on a train in Germany? German trains run like clockwork and everyone has a seat. And, the cost of this experience ? Well.. that's where the problem lies.

Germany spends €17bn on rail subsidies per year compared with €4.5bn in the UK. That extra €12.5bn is equivalent to the UK government spending across roads, buses, trams and railways combined.

If we want a pleasant rail service we either have to pay for it, at the expense of the NHS, social security, defense, education or pensions, or we'll have to accept a service which more closely matches supply with demand.

Very basic analytics could allow us to optimise the off-peak services. For example, instead of providing time-boxed services (e.g. hourly), we should work on functional services (e.g. demand-driven). This might mean a train service between towns and cities twice or thrice in a day.  We could gradually reduce the public expectation of trains running when we want throughout the day, to having a reduced but reliable and comfortable service in the mornings, middays and evenings.

Sure, people would have to organise their activities differently to adjust to this new paradigm. But that, to me, is progress and good money sense.

So let's tear up the TOC contracts and deliver some societal change in the form of a better rail service for all.