The Hidden Power of Conversational Search
Wearable tech is bringing conversational search into the spotlight, but how does it differ from normal search and why is it important to all organisations, not just the major search engines.
As Larry Kim of Wordstream says Google taught us to use “Caveman English” when searching. Typing words in order of importance into a search box has become natural to most internet users. The trouble is there is nothing natural about it, particularly when you are using voice as the input method.
Conversational search allows the user to search for something as if they were asking a friend such as “What’s the time in Alaska?” rather than the stilted, caveman way of “timezone Alaska”.
However, problems arise with conversational search when its basis is little more than an ASR (automated speech recognition) request with no intelligent reasoning behind it. Essentially it just turns the spoken word into text, guesses which words are of importance, looks up a “normal” search and gives a response, invariably incorrectly.
This is because conversational search adds a huge amount of ambiguity. That’s the reason Google taught us Caveman English in the first place, to help overcome this. In order to establish what the question is the words, their possible meanings, combined with a contextual understanding of previous interactions with the user, all needs to be reasoned out intelligently.
Furthermore, if there is still some ambiguity left, the search mechanism needs to be able to ask for clarification. It’s something we’ve been doing at Artificial Solutions for many years with our enterprise clients that use our NLI technology for their online customer service support. A customer might typically ask “How much will your insurance premium increase next January?” and the virtual assistant will then ask for more information such as “Are you talking about our Home Insurance policy or our Travel Insurance policy?”
So I can hear you asking, that’s all very interesting but I’m in retail or manufacturing, what do I care about how people talk to a search engine?
Well imagine you are the purveyor of the iBracelet, this Christmas’ biggest seller in women’s fitness accessories. Stylish, yet practical, it does everything from monitor heart-rate and calories burned to playing music from Spotify via a pair of wireless headphones. It’s even able to answer all those little spurious things one thinks of whilst at the gym or out for a morning jog.
How much is a golfing glove (Dad’s birthday is next week), is the new Grisham out yet, what’s the name of that new coffee shop on 10th, or did I add coffee to the shopping list? Using NLI, not only does the iBracelet understand that the “new Grisham” is a book, where 10th is and which app has the shopping list, it can also determine who answers those questions.
Whilst you can send the user to Google or Bing, you might choose to send some questions direct to a sports website or a book store or even use your own knowledge database depending on the question. You can even provide a voice response for the immediate answer to their query and send a link for them to look at or purchase the item later on that day.
With NLI, conversational search not only allows you to understand what your customer is asking about right now, but also enables you to enhance their experience by providing value add information, making your product a truly indispensible item.