At the end of every conference comes the onslaught of obligatory summary posts of the top takeaways from the day. You can learn about all the best speakers, who had the most creative slides and catch some great summaries of the more intricate details of the presentations. But this time I thought I’d write something ...
Google has the search market share nailed. Pretty much globally. Search insiders, bloggers, analysts and experts tend to be in agreement that incremental improvements to search relevancy can only take the competition so far (or actually not very far at all). Consensus is that any serious threat to Google dominance will come from the search engine that nails the semantic intention.
Semantic intention? Semantic Web is about understanding connections and relationships to attach meaning and significance. I'm going to paraphrase the best example I have seen from Kaila Colben blogging for MediaPost Search Insider. Kaila's example is a search for 'Who is Darth Vader's son's sister?' Type that into a traditional search engine and you won't get the answer. You will get a collection of links that contain content that is highly relevant to the 'tokens' in my query. Tokens in the query string here are 'darth', 'vader', 'son' and 'sister'. So-called noise words (so, who, a, if, etc) are stripped out or de-prioritised according to which search engine you're using. The ability to understand and interpret the semantic intentions, generally conveyed by these 'noise words' has until now been the Holy Grail of search intelligence. Until that is... Wolfram Alpha.
Wolfram Alpha (the self-titled computational knowledge engine) is officially going to launch any minute, however they released a debut screencast tonight. Watch this first, pick your jaw back up from the floor then we'll continue... http://www.wolframalpha.com/screencast/introducingwolframalpha.html
So what is Google's cash cow and why is the Wolf a threat? Sponsored Search is by far and away the biggest cash machine in the Google arsenal of products. You know the links at the top of the page on Google, Yahoo! etc? Those links with the shaded background? Yep; they are Sponsored Search ads. Every time you click on one of those ads, the advertiser pays the search engine an amount of money. Such amount is set in a real-time bid ecosystem, taking account of the value of the product I am trying to sell and the amount of competition on the page and other such factors. Here's how it works.
Comscore reckon that the average UK user searches 24 times a month. 24 x 30MM (UK online) = 720MM total UK searches a month. At over 85% of the search market share that gives Google at least 612MM searches that could potentially make money. Of course not every search has an advert attached to it. Some searches might be too academic or just plain weird to have an advert that is relevant, so let's say for argument sake that Google shows adverts 60% of the time that's a potential of 367.2MM searches on which to show ads. Of course, the sponsored links will not always be clicked in favour of the natural search results, but imagine this happens 15% of the time. 15% of 367.2 is 55.08MM clicks at an educated guess of 15p average = £8.262 MM a month
So how can Wolfram's computational knowledge engine be a potential threat to this position?
At the moment, search engines allow advertisers to bid on all sorts of keywords that could be reasonably agreed to be relevant. So if my product is a low APR loan, Google will permit me to bid on terms such as low APR loan, cheap loan, low rate loan etc. My ad will only show up if a user searches for those words though. Now; with Wolfram Alpha, those searches that require interpretation or permit greater semantic intention, could be leveraged to serve appropriate and relevant ads to previously unmonetised queries. So that is the potential to hypothetically cover way more searches with a commercial result than Google. Additionally, the more relevant the ad is to the user search intention, the more likely the user is to click on it (click-thru rate). In the Google example above, if we imagine that instead of 15% of people clicking on the sponsored listing, there are 20% instead, then the revenue estimate would be over £11MM a month. Phew!
So anyway-whatever the plan for Wolfram Alpha in terms of commercialising the product, the volumes need to be there in the first place, which can only happen when people see it as a serious alternative to Google. I for one certainly can't wait to see it live in action, and assess if this really is the future of web search.