Google’s Autocomplete function burst onto the search scene in 2008, having been born and raised in Labs under the name Google Suggest. Searchers are now very familiar with its function of listing the search phrases it thinks you’re likely to be typing. It saves us time, corrects our spelling mistakes and helps us think of the rest of the phrase we want to use.
It works by using data on the most popular searches and then predicting the one being typed at any time as the user enters each character. The system is localised and personalised to the user and time-sensitive, meaning that “trending” search phrases quickly enter the suggestions when they are being searched in larger volumes than any others.
This process means we can therefore get an insight into the UK’s search habits by quantifying the suggested results Google presents. There are of course an endless amount of suggestions (many of them fascinating, funny or just worrying) so for this examination, I have chosen to look only at the suggested searches presented upon typing each single letter of the alphabet and single numbers. For each letter and number there are 10 suggestions in order of popularity. This method is not perfect by any means because it gives a skewed set of results – one word searches are the most probable entries based on the evidence of a single letter and therefore most suggestions are single words. As we know, there is a large percentage of searches that use more than one word (in the US, about 70% of searches are longer) but one word searches are the largest single group. Results however do get skewed towards one word suggestions and therefore names of companies, organisations, websites and people get an advantage. The data is still useful and interesting in this regard as we can see which names among this skewed group are favoured more.
The following is the analysis of the data gathered in October 2012 from Google.co.uk. We took the 10 suggested searches for 26 letters and 10 numerals and categorised the terms into the below groups. Note that the data does not show any weighting so while “bbc news” and “6 music” are both at the top of their respective lists and receive equal billing in our data, the former has a far greater volume. Therefore, our categorisation of data helps to tell us variety within a category but can’t accurately show dominance of one category over another. For example, only a handful of News sources are queried and it gets equal billing to Weather, yet traffic to news sites from search is greater.
What’s quite clear from the data is that Entertainment and Retail are the dominant uses of the internet. Online Services are significant, however. These are usually brands like “Dropbox”, “Google” and “Skype” but also general terms like “currency converter”. Nearly all other categories are marginal in comparison.
We’ve therefore broken down Entertainment and Retail into sub-categories to get a closer look at people’s interests.
Above is an interesting breakdown of the UK’s leisure interests, as they are expressed online. Gaming/gambling was so big we decided to subdivide it to show its component parts. It’s sad to see books only just getting in at 2% and even that was “50 Shades of Grey” in the uncompetitive “5” suggestions. I’d say this illustrates the long tail nature of search in books and reading, which is largely a solitary pursuit with hundreds of years of titles available to read at any time. TV, Film, Gaming and Music are much more skewed to what is showing at a given time and what is new.
In Retail, the landscape is less varied. Fashion dominates the sector and even when we’ve given Department Stores, Supermarkets and Online Giants (Amazon, eBay), who have fashion offerings, their own sub-categories. Again, this is arguably to do with variety where many brands can co-exist. In contrast, Entertainment sees fewer sites (Game, HMV, Toys R Us etc) appearing, reflecting the smaller number of major players and perhaps a market that has been losing out to digital formats and alternative methods of distribution.
Finally, as an alternative view, we’ve recreated the breakdown using only the first suggested search term. This gives us a view of just the top 36 terms by alphanumeric entry.
Even with this view Retail and Entertainment stand out. Technology gains a lot of ground but this is because many technology brands and terms tend to start with numbers: “3” (Three, the mobile operator), “7zip” and “O2” appears both for the letter O and numeral 0.
The above data is perhaps no surprise to anyone who’s seen other internet usage data and just underlines the fact that the dominant use of the internet in the UK is for shopping and fun. This is also borne out by the growth of online business in the past few years; much of which falls into these sectors. The Entertainment category is interesting. We know that people have been turning away from traditional media such as TV and radio but internet services to replace these media are still being formed and struggling to catch up with demand. It’s only recently that internet TV, on demand and streaming services are hitting the mainstream, yet the public appetite for it is already huge. First movers like Netflix are therefore seeing fantastic responses to their exclusive offerings. Retail has been better at reacting and developing an online presence and service. This is perhaps due to a greater security in investment where they don’t have to battle with piracy of their content and entry costs are lower, not requiring as much investment in the technological delivery of their product.