I recently met with Valyn Perini, executive director of the OpenTravel Alliance and a Tnooz node, and the conversation turned to hotel distribution.
We covered some of the usual issues such as the limitations of working with older systems…and of course the ability of the OpenTravel XML messaging schema to help hotels more effectively merchandise their rooms and other services on property.
But we also started to talk about how people are searching for where they want to stay when they travel.
Many travelers bemoan the way many online travel sites and the hotel sites allow you to search for different properties and the way that the information is presented back to you.
The number of parameters you can key in on are very limited.
The same room type can be returned by the search in various configurations and it can be difficult to digest the differences between room types too.
And that’s just within a single property!
Trying to compare choices across different properties and different brands is nigh impossible today.
So how can hotel search get better? Â Has it even changed significantly in the past decade?
Well, Google has come up with a new way to search for hotels and I don’t think the hotels are probably very happy with it.
Yes, the search giant’s location-based search results are very good for showing choices based on proximity to a given location… but the only parameter that it provides is price.
It does level the playing field so that smaller hotels and inns can get placement alongside a Four Seasons or a InterContinental, but it’s hardly the way that people would compare different properties.
Nor do the hotels want to compete on price alone.
The W Ft. Lauderdale that my brother-in-law works at sells “cool” as much as anything (the Extreme WOW suite is outrageous) and it’s not the kind of place that my parents would probably want to stay at.
And it’s not otherwise comparable to a Comfort Inn other than that they both have beds.  A completely different experience, at a different price, for a different demographic.
So again let me ask: “How can hotel search get better?”
I recently ran across another great TED Talk by Gary Flake, a technical fellow at Microsoft, and the founder and director of Microsoft Live Labs demonstrating a visual representation of Pivot tables, long the treasure of data geeks, err, business intelligence professionals.
Take a look at the video below and you can begin to imagine how you might use this technology to browse through thousands of hotels, visualizing your search and drilling down through a variety of parameters until you find what you’re looking for.
It’s only a hair over six minutes and well worth the time.
As an aside, this is the second amazing TED Talk from Microsoft Live Labs, following the landmark demo of Photosynth in May 2007.
Why is it that they can create these amazing technologies in the labs, but what they ship to us is so bad? I’m confounded.
It’s true that it would require a standard taxonomy of how to represent different room attributes and property features.
Which leads me back to my lunch-partner Perini. I think the combination of the descriptive nature of XML, and the standardization through Open Travel could make this happen.
What do you think? Would this be a game changer?
Totally agree.
But the killer app is using that structure to provide highly relevant and highly personalized search results based on the travel persona for that particular itinerary.
Then, you get to the really challenging part –
Hotels are tough, but imagine the opportunities to provide a similar level of relevance for local travel-related activities. Would think that search/scheduling of destination activities would be about 10x (100x?) the challenge facing the high quality universal hotel search.
About 15 years ago (yes, I’m old…) the topic of discussion for hi-tech entrepreneurial types was “How can we use the Internet to make great search experiences?” Now bear in mind that the word “Internet” itself was just being discussed at the dinner table and most of the early ventures were eclipsed by early search engines such as Internet Explorer.
Although way to early for market acceptance or adoption, we believed the answer was to use a combination of artificial intelligence technologies (which indeed were in their infancy) combined with the available web languages… and hence the “AI geeks” started to prototype.
What is so refreshing to me is that the Internet (and it’s companion technologies) have matured to a point that exceptional search methods for the travel industry are no longer a meme. Add to this the current thought leaders in this space, including Robert Cole, Timothy O’Neil-Dunne and John Lambe and I think we’ve hit a trifecta of sorts.
So flashing forward to 2010… We have a “toy box” full of mature (or maturing) technologies:
• Social Graphing (RDFa via XHTML) – that enables any web page to become a rich object in a social graph.
• Behavioral Profiling – that is widely used in the search industry today and typically uses dynamic click stream analysis to create, exchange and utilize behavioral quadrants that contain granular facets.
• Content-Based Filtering – which includes domain-specific filtering techniques (such as filterbots) used to evaluate, associate and weight (classify) items within a specific domain, such as hotel.
• Collaborative Filtering – which is the process of filtering for information or patterns using techniques involving collaboration among multiple agents, viewpoints and data sources and is widely used to recommend products and inventory from “like minded” consumers.
• Neural Networks – that provide sophisticated approaches, such as self organizing maps, that are being used to cluster query logs in order to identify prominent groups of user query (such as search, pricing and availability) terms for further analysis.
So I say “let’s get ready and BUSY!” Let’s agree on what best enhances travel search technology and let’s get the infrastructure in place so that this data can be freely shared within open standards.
Bonnie
Ahh, the complexities of mapping one conceptual frame to another. George Lakoff (Women, Fire, and Dangerous Things among other interesting books) would be all over this.
At the transaction level – when I am asking about a property’s properties (oh dear we have overloading here too), we don’t have to have agreement about the terms. So a king sized room means er… um. Is that a big room, no statement about the room size but a big bed (whose definition of big?)?
Certainly for the exchange of messages we need to get agreement/standards. However there is absolutely nothing to require that the hotel stores the items in the same way that it exchanges messages.
There are 3 sizes of king-sized bed that I know of. A king in England is the same size as a US queen sized bed. I have fallen foul of that when buying linens in the UK and wondeing why they fit so poorly on my US bed. There is the standard US king, and then there is a California King. There is absolutely no standard nomenclature lying around, and probably nor should there be. The rich and famous have flunkies to make sure they get exactly what they want (imagine a phone call between a proper British hotelier, played by John Cleese perhaps) and the gofor for some soi-disant celebrity. Actually there are likely to be many phone calls. Precisely because there is no standard and somehow the translation must be done at the meaning or semantic level between the desire of the soi-disant celeb and the hotel property. The flunky;s job is to act as that intemediary – the meaning translator.
So we are often going to be surprised when we ask for things with one context in mind and that context doesn’t match the context of the recipient of the question.
Google is getting uncanny good at predicting (some of it heuristically, some of it algorithmically) what we mean when we ask things. The list of what we might be looking for when we start entering text into a Google search box is amazingly borad and surprisingly precise. This wired article http://www.wired.com/magazine/2010/02/ff_google_algorithm/
shows just the kind of thinking going on in search now.
So for a specific industry, perhaps we ought to be looking at how ordinary searchers look for stuff and harness the power of Google to sort the wheat from the chaff. Once that is done, then perhaps we can use message exchange patterns to get what we want transactionally out of the reservation systems.
Better still search mashus – bring Google+yelp+destination city code+weather+activities… all together. Get the results narrowed and then perform the transaction. We still confuse looking and booking – yet the activities are radically different.
Chris, thanks for the comment. You do make an excellent point that there’s challenges in getting agreement on syntax and that agreeing upon a messaging schema doesn’t equate to how a hotel might store the data or whether it presupposes a common syntax. While I think that it would be helpful were it so, it doesn’t need to be for the concept to be useful IMO.
I think it would be a big step forward if even a single hotel brand or chain would try something like this. Hopefully they could at least agree internally on what they mean. Trying to spread that across different brands or chains may be nigh impossible. And as @Robert suggests, extending it to activities is probably just fantasy at this point.
One thing I do want to add is that I’m not sure that I want to leave it to Google to get the search right. Many searches that I do have very different kinds of results returned, it’s far from perfect. I am far more in favor of your search mash-up idea. It’s exactly the reason that I think the Siri acquisition by Apple is so (potentially) brilliant. The concept of the “do engine” is far superior to a “search engine” in my view.
And lastly, while there may be confusion about what a “king bed” means (I had no idea there were more than 1 definition, although now as I think about it there is a California King), it would be useful, at least to me, to be able to easily compare square footage of a room or whether it included certain amenities which should be fairly simple to keep consistent.
All that being said, did you think the idea of presenting the search results in a different manner has merit?
Good discussion here.
Doing a 180 from the standardization theme we should consider the opportunity in mass differentiation using rich media. With high resolution pictures and video increasingly common place (and mobile – think iPhone, Evo & Others to come), a real and practical solution would be to provide Open Standard rich media tags and let consumers literally see with their own eyes what the hotel and amenities look like.
This is fair for hotels as well as they can put their best foot…err..media forward.