High-flying Web prices brought down to earth

There comes a time in the affairs of men when you just have to flee the country. I have a friend with a farm in Sweden, and a standing invitation to go and hide out there if the need arises (Tax inspectors, furious fathers and irate editors please note: other boltholes are available). A couple of weeks somewhere bucolic working on the eternal novel seem like just the thing at the moment, so it was off to the Web site of the appropriate cut-price airline to check prices. Forty quid return? Don't mind if I do -- but I hadn't confirmed the return date, so I cancelled the transaction and sent off an email to my farmer chum to finalise the timing.

The next morning, my inbox had two items of interest. Yes, said pally, those dates are fine. Come on over. And the airline had sent me a circular saying "Amazing! £15 off all European returns if you book by Tuesday!" My! What timing! Even in Sweden, £15 buys a small round of drinks, so it was with renewed joy that I returned to the airline Web site and repeated my selection. But I must have got something wrong -- my £40 fare had gone up to £65 overnight. Fortuitously I'd got a screenshot of the earlier price, so I checked -- nope, wasn't me. It was exactly the same journey, but the £15 discount seemed to have turned into a £25 premium. Even the airport taxes and other charges had all but doubled.

Low-cost airlines rely entirely on price manipulation, an approach invented by American carrier Southwest Airlines. The lower the demand and the earlier the booking for a seat, the lower the price -- and vice versa. The algorithms that the carriers use to determine this are the crown jewels of the operations, and as jealously guarded as anything in the Tower of London. Us punters cannot predict what price a seat will be until we actually buy it: the airlines themselves don't know until they've factored in load factors, the latest predictions and other mysterious magic that only they know. So I don't mind being caught with a higher price if I don't get my order in as quickly as I could. But this, where a massive price hike coincides with a major promotion, struck me as more than a little suspicious. Could conditions change so much overnight on the route?

I mentioned this in a couple of places online, and got three classes of response. One sort assumed that I didn't know how low-cost airline pricing worked, and explained it to me much as a whiskery grandfather would explain steam engines to a wide-eyed kid. Well, thanks. Another lot said that I should be grateful for cheap fares, and if the company wanted to rip me off I should accept that as part of the deal or just not fly. I'm not sure I want morality to be price dependent. The final class of replies said "yes, happened to me too, but what can you do?"

It seems as if the airline in question holds all the cards. Prices are necessarily very volatile and there is no way anyone can question the calculations: only they have all the data, and that's top secret. You can't even complain about the "£15 off" email, as there's no indication what the original price is -- it's a meaningless statement. But that's no excuse for misleading advertising, if that's what it is.

Fortunately, there is a way to tell. As any statistician will tell you, an individual datum is next to worthless but in the company of enough of its peers the truth will out. If I check a particular route every day, I can build up a picture of how the price changes over time -- and if there's a spike coincident with money-off promotions, then that starts to look significant. To be sure, I should check more than one route, and to be really sure I should also monitor similar routes on other comparable carriers. The ideal would be to extract every price possible every day, at which point you can probably have a fair stab at the company's pricing algorithm and set up your own airline, but that's statistical overkill.

There are problems. If I were running a Web ticketing operation, I'd be watching the requests to spot anyone harvesting data like this -- even if I wasn't being a bit cute with my marketing. You don't want to cause problems for genuine frequent fliers, but the stats cut both ways. If someone's gathering enough data to be mathematically significant, it'll be obvious to the data provider.

But that's OK. We live in a distributed age: I can find a hundred volunteers with broadband who can make a couple of requests a week. I can build a small agent that runs on any PC, picks up its weekly schedule from my server, does the business and reports back. With a little care, the spread of times, IP addresses and routes examined will disappear into the background noise of the normal requests. Over a couple of months, it is entirely plausible for a suitably motivated individual to collate enough information to say with a great deal of confidence exactly what is going on.

Investigations continue, as they say. Meanwhile, I very much hope that consumer organisations, trading standards bureaux and others will start to think in these sort of ways. It's too easy when confronted with a high tech, well-funded organisation to assume that we are helpless -- when the truth is that we have access to the same tools and opportunities. And to anyone thinking that mug punters are too thick and greedy to notice or care when they're being milked: think again, chum. To overpriced fares, we say: Book 'em, Danno.