Understanding the change: Vertical Search

Since 2007 to present, online hotel reservations have been increasing by about 10% every year: Booking.com has just celebrated their half a million structures in the world. The turnover of the online travel business is over 150 billion dollars and even a start-up such as Airbnb has been able to enter in the ten-digit Silicon Valley companies, alongside with names like WhatsApp and DropBox. Yet, the volume of search for generic keywords related to travel on traditional search engines like Google, Bing or Yahoo is decreasing constantly.

The New Search
Let's analyze this GoogleTrend graph referred to the search volume from 2007 to present for the query "Hotel in Rome":

google trend

Now let's compare these data with two different keywords, Tripadvisor and Trivago:

google trend
google trend

As you can very clearly see, these two trends are in direct conflict.
This shows that the starting point for hotel reservations is no longer, as it used to be, the classic search engine, but it is what is technically called a vertical search engine.
But let's take a step back. Until a few years ago, going online to find a hotel could turn into a very frustrating experience: once you had chosen your destination you would have to navigate through dozens of proprietary websites, find the booking engine to check prices and availability or, in the more than likely case there was no such thing installed on the site, make direct contact with the hotel just to find out that there were no more rooms available or that their price was too high. So one would have to start over again, interrogating a new proprietary site and going all the way until the final reservation. Such a process could take hours, if not days, of web surfing, at the expense of the less SEO optimized sites or of those who did not have a booking engine. 
This explains why, until 2005-2006, the greatest number of searches was done with keys like "Hotel in Florence" or "B&B in Florence." In the absence of an aggregator, the only chance you had was to consult one by one the websites of the hotels you were interested to, and choose accordingly.
The decrease in the volume of these generic keywords does not necessarily correspond to a lesser interest in a specific destination (the tourists flow in Rome has not decreased in proportion with the decrease in the volume of Google searches), but it simply shows a different approach to the web search, which is hence defined vertical.

Vertical Search
A vertical search engine, unlike Google and the like, is focused on a specific segment of online content. Instead of indexing large portions of the web, as Google actually does, it is limited to certain pages, related to pre-defined or niche topic: typical examples of vertical engines are the online shopping sites like Ebay, Zalando, Asos and obviously Amazon.
If I want to buy a book it is more likely that my search starts on Bezos' portal rather than on Google: the results will be better, sorted and filterable, with the readers' reviews, the prices of the various editions and suggestions for similar authors. I can put the books in the shopping cart, save them in a wishlist or compare them with other titles, all in a matter of seconds: a huge difference if we compare this example with the previous one.
In short, the vertical user experience, when it comes to willingness to purchase, is much more satisfying and less frustrating than the generic one. This explains the huge resources employed by Big G. in projects like HotelFinder or GoogleFlights.
According to the New York Times, less than 15% of purchase oriented searches starts from a classic search engine. For a hotel, to be among the top positions with secondary keywords such as "Hotel + city name" is no longer essential as it was in the past, indeed we can safely say that it is almost useless. Most of the guests will discover your hotel for the first time on a vertical search engine, where vertical, in the travel industry, is synonymous with MetaSearch.
This explains the no. 1 chart of GoogleTrends and the increase in the search volume for Trivago and TripAdvisor (but the same is true for Kayak, Wego, etc.). SEO efforts to maintain a high ranking of words with high competitiveness ("Hotel in Florence", "Hotel Termini area", "New Year's Eve dinner") are increasingly less profitable in terms of ROI and the only keywords that are really worth to index are the brand of the hotel and, at most, a few more specific and niche long tails.
By studying the statistics of your tracking tool you will more and more often stumble upon reservations like this one:

Search by hotel name on classic search engine
(Google, Bing, Yahoo, etc.)

Short navigation (sometimes only the HomePage)

Click on Booking Engine

Booking completed

This reservation was most likely started on a vertical engine before the session during which the reservation has been confirmed, even if it is not possible to obtain confirmation from Google Analytics (or from any analysis system you use).

Unique Visitors and Cross-Device Navigation
Although cross-device analysis tools (which analyse the navigation performed by the same user, but on multiple devices) are becoming more and more reliable (Universal Analytics in particular), in principle we insist on evaluating the performance of our websites only in terms of unique visitors. A "unique visitor", in today’s multidevice market, certainly does not define a single human being, but rather a single device. Following this misunderstanding, a guest who starts his or her navigation from the office desktop in the morning, resumes it on his smartphone in the afternoon and finally books from his or her tablet in the evening, will be calculated not as a single visitor, but as three, so much for the reliability of the conversion rate which, consequently, will drop dramatically.
Two renowned Google Analytics experts, Sagnik Nandy and Justin Cutroni, have created a funny acronym to define this situation: W.T.M.D. |(Way Too Many Devices). By now it is very unusual to find someone who owns only one device. How many people do you know who have a desktop PC at home and don’t have a laptop? How many of these do not have at least one smartphone, one tablet, a reader, a game console or a wearable? Even I, as I am writing these lines, have five different devices connected, and all of them (just because they are connected) are potentially traceable.
According to forecasts, in 2015 there will be two connected devices for every person on the planet, for a total of more than fifteen billion devices. If you add the Google Android One project (an Android mobile with a high quality performance which should cost less than $ 100), which aims at disseminating smartphones also in Third World countries, it is not difficult to assume that this estimate may even be too prudent.
Let's go back for a moment to the above reservation example: until the day when it will be possible to unequivocally track every single inhabitant of the Earth, giving each person a unique ID (which, despite being a disturbing scenario for many, from the point of view of targeting potential customers would be a true revolution, as it is ver clear to Big G. who is following this path with the Android device/ Google account unique tracking), not associated to the device or browser used, we will have to use our deduction and common sense. In a reservation coming from a vertical search we can typically find at least two of the following characteristics:
Short Navigation:the time that goes from the landing on the site to the actual booking is no more than ten minutes (the average is about twice as much, almost equally distributed between the website and the booking engine) and this short dwell time suggests that the user already knows the hotel and knows exactly what he wants to buy (which means he has found it during a previous surfing session);
Landing on the site through specific search: this user has already visited one or more metasearch pages, so he knows the name of your hotel and types it in classic search engines;
Reduced dwell time in the website pages: This user does not linger too much in the various subsections of the site, but passes from the homepage to the booking engine in a few clicks. Why doesn't he visit other pages? Maybe he's not interested in your website. Not necessarily.

Assisted Conversion and Latency
Here we introduce the concept of latency, which is the time that elapses between the discovery of a hotel and the actual reservation. On average, it lasts one week during which the potential guests browse through offers from various distributors and / or compare your hotel with competitors or other hotels that meet their needs (ten, on average). Latency is very often multidevice which means that, with the exception of a few cases, it is impossible to track the entire navigation process (at least now, as we have seen). It is therefore likely that, during the latency period, the user has already visited your website a few times but, coming from another device (or another browser, or just after clearing his cache), he has not left any traces of this navigation, or better, he has been identified as a common visitor who, in the end, has not booked his room.

vertical search booking

Understanding the vertical search process and knowing how to correctly interpret the behavior of your website visitors will help you choose the most appropriate strategies and avoid expensive low-return campaigns resulting from incorrect analyses of the data in your possession.