Facebook, at a recent travel conference in the US, suggested its role as one of being in the “happiness business”.
Furthermore, citing some internal research, an official from the omnipresent social network went on to show how the American general public actually gets a pretty big high from the process of planning travel, and that it lasts longer than the act of travelling itself.
But does this happiness extend to when consumers are searching for flights?
Some cearly do not think so. A recent article in the Detroit Free Press pointed out that airfares were not fair, and that searching for appropriate flights and fares isn’t fun.
So perhaps it’s time to think through the process from the consumer’s perspective. In some respects, air search is still a V1.0 product. After 15+ years of being able to search for air on the internet, it really should be a heck of a lot better than it is.
Myth 1 – Air search is simple and easy
Searching for flights can be a daunting task for internet users. Many providers of internet travel search products often forget that a leisure consumer may only be travelling once or twice a year.
Invariably, it is difficult to find prices and routes, and there is a bewildering array of options. To the traveler, the differences between a low cost carrier and legacy airlines mean nothing.
The intersection of fares, availability, rules, schedules and now ancillary services and unbundled products (not to mention taxes and charges) is also very difficult to understand and compare.
Users can spend hours looking at different possible options and sites. Sorting them all and finding the appropriate flight is painful, and no one really helps the user in this process. Each of the websites offering assistance – and this applies to all – from direct airline sites, OTAs and meta search players offer a bewildering array of options.
There is no definitive single “right” answer that a user can trust. Mapping the possible right answers shows that inside an airline there can be two (sometimes even as many) “right” answers as there are places searched.
Air search is only one part of the need from Internet Users in travel. They want/need to have global information point to point. While they may know a lot about their city of departure there is very little they know about their city arrival, with timing to get to the end destination points.
And the game is complicated as the low cost carriers can land far away from city center or a connection can be from another airport (and some airports are really confusing). Try changing terminals for example in Dallas, or Madrid?
Changing planes in JFK is never a pleasant experience and can take literally hours. And where is all this information? Is it easy and simply explained?
Hardly a happy experience. Which is why inextricably tied to myth #2…
Myth 2 – It doesn’t matter where a user searches (the results are the same)
As a user, I look on metasearch sites, but also direct on airlines sites, and on OTAs. In mainland Europe, such as France and Germany for example, users also need to compare timing and rates between flights and the hi-speed rail eg. France’s TGV and Deutsche Bahn’s ICE.
So, as noted in Myth 1 above, different search sources return different results. There are now four basic types of results:
- Airline results from the internal airline reservation system (eg. HP Shares)
- Cached results created by a caching engine (eg. ITA’s QPX engine)
- GDS based results calculated either by a GDS based cache or by slamming the GDS system multiple times (literally on occasions hundreds) to build a result set that then are discarded
- Scraped results, in other words, results generated by someone else’s system copied and pasted into the serving site’s results.
The results, regardless of search site, are inconsistent and generate little trust for the user. Why are they different?
This leads us to the next myth…
Myth 3 – Air search works
Frankly, it doesn’t. For something that looks like a commodity product, an airline seat is complicated. Here is a smattering of the differences:
The brand.com website, often powered by ITA (such as American Airlines or Alaska Airlines) generates a result, but this is no longer dynamic. Meanwhile, the internal price from the call centre agent is dynamic and can be different.
Why? Because it has a different engine offering the price.
Outside airline-brand.com, we have OTAs that are powered by different cache-based engines, including ITA (powering Orbitz), and Expedia BFS, originally co-developed with Worldspan and now using Sabre-based data.
We have metasearch engines that rely on some general sources such as ITA or some that are scraping the screen results from each website (and let’s not forget there are some combinations of this).
This is ugly. The core problem here is that the airline supply side wants to assess the requestor before providing an answer – but the user wants speed.
The airlines have delegated to the GDSs and others the right to legally price their product. Thus there are many possible right answers and different players calculate the “right answers” differently.
Bottom line, there is no single definitive source for air search results, so nothing works particularly well. So could a user develop trust in this environment? And so who could you trust, this leads us to Myth #4
Myth 4 – Google Search is trustworthy and is a good model for air search
The magic wand theory. Google waves a stick and magically the answer appears, and it’s Google so it must be accurate. Most users appreciate Google Instant for its fast speed, but users don’t think too much about whether those products improve the quality of search results.
As we have seen, Google Flight search is blazingly fast but is it accurate? The first few weeks so far have shown the results to be incomplete. As of yet we are still seeing a similar metaphor to existing search sites – it adds a few bells and whistles but it’s not really doing anything fundamentally different, new or better.
As Google admitted at launch, Flight Search doesn’t include all US domestic airlines, or any international routes, so is not a great model for complete or useful search results in its current guise. Today, Google is also only handing off to the airlines. Will they only do that in the future? Well this leads us to Myth #5.
Myth 5 – The hand-off from search to booking is seamless and never fails
There are two elements to this issue. Form one is the external handoff from the search (metasearch such as Kayak and Skyscanner or a deal site) onto an OTA or an airline site in the workflow.
The second form of handoff is from an internal (ie. within site) search to the booking function. With different logic and different interpretations of the rules depending on the site or between sites, users get different answers.
Furthermore, the state of the answer matters. With a fundamentally dynamic situation regarding state of availability and rates, rules, etc, changes to availability and rate can occur quickly.
The number of instances when a result from a metasearch to the host site fails is quite frequent. There are no statistics that have been published on the subject but having seen several large system results I can assure you the number of failures is significant.
In my own case, I do a lot of searches and see the failures all too frequently. I am sure many others have seen this too. So by now perhaps you are ready for the last myth #6…
Myth 6: Flight search is merely a user interface problem?
Hipmunk, for example, has blazed a new trail by showing a different view of the flight search results. True, it’s very cool, but does it solve the core problems and does it provide more trust?
Passengers have many thousands of possible combinations to contend with, and expressing those could be interpreted as a user interface problem, but it’s far more complex than just the UI.
Data presentation is but one part of the situation, because air search is built on logic, data and compromise, based on rules that are not designed for implementation by a machine.
In fare rules (remember just, one component in the results) there are four major components – fare levels (prices, charges and taxes), rules (usually the 42 “mini” rule categories, some complex, some simple) routings eligible for the fare, and finally a thing called footnotes which are essentially rules one cannot express in simple “codifiable” statements.
So there you have it. Myths and ugly realities. The industry needs to do better to provide reliable and trust worthy results to the consumer. That, my friends, is a great challenge to have.
Related posts:
About Timothy O'Neil-Dunne
Great stuff Tim. Agree with everything for once!
I would only add that given all the complexity you just highlighted, some smart folks are doing a pretty good job.
Tim — I agree with much of what you said.
But — I think it’s an exaggeration. In the scheme of things, buying air tickets just isn’t that hard. How do I know? Millions of people do it with apparent success, since they fly seemingly when they want, where they want to go, and within their willingness to pay. And it’s a lot easier than it was 20 years ago before OTAs and airline web-sites. In some markets, yes, it is very challenging. But in many markets, it’s actually quite easy.
Is there variation that makes it frustrating at times? Yes. Are there technological hurdles? Sure.
But at the end of the day, people don’t trust airfare buying NOT because of the things you mention about the search process, but because of airline revenue management that changes the prices of a ticket seemingly minute to minute. They don’t trust that a fare will be $250 today and $180 tomorrow (as a recent ticket purchase I made was. Fortunately I waited and used Yapta to let me know the fare had dropped). I believe people are less paranoid that the fare they’re purchasing is available somewhere else cheaper right now than they are it WILL BE available cheaper on that same site the next day.
Unfortunately, that problem is even harder to solve. Airline revenue management is always 1-step ahead. Even Farecast/MSFT Bing’s algorithms aren’t good enough to save the consumer.
Just my 2 cents,
Evan
Nice to know that Facebook eventually catches up with what any self respecting agent has known for the last 30+ years or so … people get a high from planning travel and spend longer planning than travelling. Oh! Yawn! Then again, a lot of these Facebooky/ Googley types do spend a lot of time stating the blinding obvious (certainly about travel) whilst trying to make out they are making some fantastic new revelation…. whereas they are really making themselves look rather naive.
Basically, this article all makes sense. There are two types of travel: The very, very simple, short haul A to B flight and b) Everything else. It’s the “everything else” that’s a problem. There are a few issues. The main one is of course, that no matter how clever your website is, how advanced the technology becomes and how right the answer becomes – the fundamental problem remains that though the answer may be right, it is highly probable that the question asked is incorrect. Garbage in – garbage out.
Planning is undertaken because one lacks information – the planning process is assembling as much information as possible to ascertain which questions we need to resolve, then and only then can we point mouse at screen. It’s a bit like saying that my family and I wish to visit Uncle Bob. Thing is, we do not know where Uncle Bob lives, how long we may stay for or even if Uncle Bob wants to see us in the first place. So, we have to find out a lot of information before we can even start to seek a solution. Present travel search is akin to using search itself as a gatherer of information rather than a resolution of a problem. Now, the “Semantics” school of thought is moving towards changing that but present search, of any hue, is really variations of a rather well worn theme.
This also explains why the GDS is still popular. It answers very raw questions, with raw data. It doesn’t try to be clever. There is no “you wanted this but might like that” approach. It also explains why it is still the ultimate answer to resolving travel issues. (not talking about wha goes on behind the scenes at a GDS here, mind)
How does one find out what the “question” should be? There is, of course, a simple answer – visit a travel agent. Agents are quite good at asking all sorts of questions and from the replies received, building up a picture of how to offer the best value proposition to resolve any given travel event or proposal. Now, note that I say “best value” and not “cheapest” – there is a monumental difference between the two. Indeed, “cheapest” can be (and very often is) a very expensive solution. Of course, you can decide that the fee an agent charges can be avoided by booking online. Sure. Book online, muck it up and find that the fee becomes cents by comparison. Believe me, I and many of my brother and sister agents have spent much time untangling on line disasters. For a fee, of course.
You see, search and booking is not the end. It does not stop there. “Everything else” includes the (all too frequent) backup. For example, the other day I was telephoned at sparrows fart by a client who was not where they were supposed to be. There was nothing I could do – because the booking had been made through an OTA – who was not available and when the client did get through, suggested that they could do nothing either. (I found out later that the ticket was flexible it just involved something which OTA’s hate – faffing around sorting the issue out).
When I fix a travel event for a client, I have control. The client knows he can call me at any time, I know what can and cannot be done with the tickets (because I have built in to the ticket what I feel may be needed – both in terms of fare structure and how the ticket has been put together) partly through knowlege of the client, partly through knowledge of what they do and partly through bits of information picked up during the booking request. The point I am making, is that techy types assume – very, very wrongly – that once the ticket is issued and the money taken, the task ends there. It doesn’t.
So, these “myths” are very relevant. But, until the “semantics” boys come up with something and we can give it some rigorous testing, no matter how clever existing search becomes or how fast existing search becomes, those myths will remain.
Standing. Clapping.
I totally agree with all of the points above..it’s really frustrating as a IT systems developer having worked for multiple OTA’s, having improvements and innovation quashed by outdated GDS process flows, business requirements and call center operations teams who simply want to replicate their current manual process online. The problem with this is call center operatives are trained to make bookings, the customers shouldn’t require training as well!!
In my mind technology should be used to create booking flows that abstract the complexity of the airlines entrenched systems, not simply pass on the complication for the consumer to decipher. To make matters worse OTA’s then output a mass of t&c’s and “strategically” placed “up sells” to confuse the consumer a little more.
The balance between between simplicity and functionality needs to be restored!
Ah – if only it were so simple. Indeed my point is that there are several problems with today’s infrastructure.
1. There are seats available at fares which are legal which the systems do not – and will NEVER show. But a travel agent can find.
2. There are seats which look like they are available but never are. And NEVER will be.
The default situation is that it is assumed that the RM and YM systems are the cause of the problem. That was not my point. The myths are around that there are addressable issues that are not being addressed.
That’s worth about … 3 cents
Cheers
Congrat, great post Tim
Having worked on the front end problem of flights and trains, the complexity of the back end systems is difficult to hide especially with business requirements which can force you hand to feed way too much data to the customer.
Customers should set a budget for a trip based on reasonable average prices for their comfort (economy, usually) and not search for small savings here and there – this seems to be the biggest source of frustration I have seen. The big customer myth is that there must be a cheaper price somewhere because someone once paid 100-200$ less for a similar flight on a similar route. Airlines do sales and special offers sometimes, but getting those prices is luck.
Flexible dates search for flights more than a month in advance is the best way to shop if you’re on a tight budget. Shopping within 30 days of travel is not some mystical thing, there’s no need to compare massively. Meta search engines will give a good idea of the average market price. It would then be far better to shop within your favourite airlines and make sure you’re leaving from convenient terminals.
Pricing complexity is a dark pattern though. The more you look at it, the more your head starts to spin. http://wiki.darkpatterns.org/Price_Comparison_Prevention
#1 – a human travel agent (albeit a good one) knows the tricks of the trade to isolate the cheapest airfare _appropriate to customer need_. The cheapest isn’t always the best — for example a tour or cruise that might get cancelled would have been more appropriate to spend $50 more for the air fare with refundable ticket rules. Who tells you that?
#2 – With all the complexities listed above (that cannot be avoided since ultimately there is only one source for all air fare – ATPCo), perfecting search is practically a quantum problem (i.e. it’s never going to be perfect). An international result has literally thousands of combinations that can be returned by the search engine, and the search engine isn’t operating in the same system as the booking engine…which is the mainframe host and never going to be able to perform the complex calculations required for search.
#3 – Expanding on point #2, it’s all about the algorithm. Who’s smartest? I vote for the MIT grad students of ITA Software — for now.
….. so unless you really know how air fare works, a human knows more than the search algoriths and I’m not holding my breath for this to change anytime soon.
GDS Geekster… you are clearly hiding something in not revealing yourself.
I need to correct you in a place where you are TOTALLY wrong. Further I need to add some interpretation to some of your comments. Let me dive in.
Your statement:
(…since ultimately there is only one source for all air fare – ATPCo), Is inaccurate and incomplete.
There is only one true source for an airline’s fare. It is the airline. This is 100% universally true. ATPCo has the RIGHT to display many (most) of these fares as a result of agreements with airlines when an airline CHOOSES to file a fare through ATPCo. A GDS has a right to distribute (and calculate) these fares (including the rules) provided a full content agreement exists.
There are airlines who choose not to file fares through ATPCo. For example Ryanair. There are specific fares which are not available at all through ATPCo, for example the paper contracts that exist in Europe and in many markets. I, and many others who are experts in this area) can quote a litany of examples of fares that are not in ATPCo, available via a GDS or interpreted in any channel.
Do the GDSs and ITA (and others who are licensees of ATPCo and other systems) provide 100% factually all possible legal fares in calculation? No they don’t. Part of my issue here is that this is still a matter of interpretation.
Where Mr Geekster you and I do agree is that it is a quantum problem. However I believe that the technology to solve the quantum problem exists (to a higher level of confidence at a lower price than is currently being charged by the existing players). My big data post http://www.tnooz.com/2011/08/15/news/ten-reasons-why-big-data-will-change-the-travel-industry/ is an illustration of the capability. Also check out some of the other references to Big Data from TNooz and other sources. I recommend if you have not read it to peruse Fred Lalonde’s post: http://www.tnooz.com/2011/09/20/news/big-data-bringing-the-magic-back-to-travel-technology/.
Does the human know more than the search algorithms? That is debatable. I believe that a good agent can mine and find solutions in the ends of the bell curve of possible answers better than a machine can. However based on volumes no he cannot without good tools and the legal ability to execute such tools. I can point to several engines that can find legally valid fares today that are not GDS based.
The myths of air fares today contrive to provide consumers with less confidence in the resulting output. The emergence of mix and match fares – which may be legal but are hard to purchase (e.g. when one half of the fare is purchased and the other fails) is adding to this complexity.
What I believe long term is that someone will be able to solve it. And the ITA team (now members of the Google family) in Cambridge do not have an exclusive on this solution. Further I believe that viable solutions will emerge sooner than most people think.
Cheers
@gds geekster #1 No! Wrong. If you were booking flights with a cruise or tour – that is, vacation – you would either a) get them via the cruise operator or tour operator, so if the cruise or tour fell out the flights would be included in any refund or b) Get them independently, in which case you would have travel insurance (unless you were incredibly daft) which would cover any cancellation – either way, it’s academic. We are referring to business travel where an understanding of the clients likely requirements – and probably the agent will know more about these than the client – are paramount.
#2/3 I would venture that an algorithm can only work given a set of parameters. Thing is, in travel, you very often do not know what those parameters are. One has to make assumptions based on a persons’ travel history and knowledge of previous travel patterns and or events. Ergo, the premise that a system may be able to answer a given question with varying degrees of accuracy and speed … but has the *right* question been asked – holds true.
There are indeed fares that are not GDS based, rather they are – but you haveto be allowed to have access to them. This is normal commercial practice in any business. Its called “wholesale” and “retail” Why should any given agent not have access to wholesale fares, if they retail the product in large enough quantity? You want to suggest that customers should be able to, say, buy orange juice at Wallmart’s at their buying in price? Of course you can! How many containers of the stuff do you want (in one go) 10 or 12? By the same token, if I sell $4mill’s worth of airline X’s fares a year, yes I can have their CAT35 fares (that is, wholesale fares) why not?
Let me reiterate the points that the techy world forget: 1. There is more to travel than just flogging someone a seat/ holiday/ travel event – there is invariably lot of changing, ammending and just dealing with “stuff” involved – I am sure Google (say) are not going to want to have a full time global team fixing Uncle Tom’s issue with changing his flight from the 10:30 to the 11:30.
2. Cheapest is (very) often the wrong solution. For some reason, OTA’s and others assume cheapest is best; a fundamentally flawed approach. Couple this with some of the hopelessly unrealistic travel itineraries that some come up with – it’s best value that is the goal, not “cheapest”. Until such time as anyone can come up with a way of factoring in all the (business travel) uncertainties, specific to each individual travel event, the travel agent is the only way forward. Semantics notwithstanding.
Personally, I could not care how information gets to me, for me to interpret. I do need, however, that information to be clear, in a format which means I can play about with it, permutate it, think about it and decide if the look and feel of something is right … now, if that’s direct connect, GDS, web or frankly, by ruddy carrier pigeon the airlines and techys can sort that until the cows come home. That said, the GDS has done a pretty damn good job for the last 30 odd years and still does today. So far, even the supposed super-clever MIT graduates have come up with nothing better, so… blue screen, flavour of the month metasearch site (aka a crib sheet) and some help from Google Earth it is, then, for the time being.
Finally, techys – remember this. Firms pay their executives to land them and service their multi-million dollar deals and they pay PA’s to back up those deals. They do not pay their executives (or PA’s) to have to spend time working out the best way of getting from Dallas to Istanbul, with a meeting in Paris along the way… and for them to have to fire up a PC/ notebook/ app in the 10 mins they have in a taxi, with a dodgy network connection to faff about even more to change their flight from the 9am to the 10am. It is much more cost effective to have them call the likes of me (or fellow agent who deals with them) and say: “Hi.. Yes… it’s me… I’m running late, move my flight on an hour, will you?” *puts ‘phone down*