Search behaviour is changing at a rapid pace. During an interactive session at SXSW, Microsoft, HotelTonight, Ozlo and Yummly gave us a peek at the future of searching. What if the self-learning assistant replaced our search engine?
Rangar Majunger, Bing’s Group Program Manager, has noted that the use of voice search is increasingly rapidly. Spoken queries are occurring more often via smart vocal assistants such as Alexa, Siri and Google Assistant. Alexa has already been coupled to Uber for booking trips. But Charles Jolley of Ozlo still sees significant restrictions in its use. In a demo, he shows that, in eight out of ten times, Alexa does not give the right answer. In Jolley’s opinion, the user experience must be improved, which is only a question of time. Ozlo is a personal assistant that shows increasingly better results based on data, natural language detection and earlier interactions. Gluten allergy? Ozlo displays only restaurants that serve gluten-free dishes.
Google is believed to be the future winner in the virtual assistant ecosystem, something that even Bing’s Majunger laughingly admits. By coupling Google Assistant to Google Home, Google will soon be able to advise people to switch suppliers based on their energy use. Assistant has already been rolled out on various Android devices and integrated into Google TV. Because of this, Google is capable of understanding us as users even better. It is only a matter of time before advertising is permitted in Assistant.
Amanda Richardson, VP of HotelTonight, states that her first priority is getting data organised. With competition like Booking.com – which spends 300 times as much on advertisements – HotelTonight is taking a ‘wait and see’ approach with respect to AI. Richardson does incorporate predictive intelligence in her marketing. The company adapts automated advertisements and hotels displayed on the basis of weather variables. In addition, it has learned when a user intends to book a business hotel without having entered an explicit search for this. Brian Wittlin, founder of the recipe app Yummly, agrees with Richardson. Yummly uses predictive intelligence to develop new recipes. This enables Yummly to develop thousands of new pages automatically and to send personalised recipes to users. Now half of Wittlin’s team is data scientists.
Back to search. What does Majunger see as other important trends in addition to voice? He is quite adamant about queries becoming increasingly longer. He sees users asking complete questions in Bing more and more often. He also indicates that questions entered as searches are a major part of their traffic. Rather than give answers, Yummly poses follow-up questions, in order to arrive at the best match via an active dialogue. Clicks and registrations are analysed in order to improve the relevance of follow-up questions. And conversion rates go up.
Listening to the panel members, I think about the impact on our work. Because AI needs data to become self-learning. Regarding the structuring of data and being able to personalise it, our role is inceasingly important. Speech recognition by AI assistants may still be in its early days, but technology is advancing rapidly. Thus we must reflect on how virtual assistants can disrupt the business of our customers. We must understand how we can make results more visible in Alexa and Google Assistant. Google: luckily, it will remain just a verb but, as consultants, let’s not forget to think outside of the search engine.