How AI can power your customer experience
When Google created their home page experience back in the late nineties, a massive trend kicked off. Instead of putting up portal experiences, directories and other ways to navigate the emerging world wide web, the world was provided with a search box. Nothing more. Nothing less. Everybody understood how to use it - either you perform a ‘navigational search’ to navigate to the page you want, or you search for some piece of information, or a thing.
This trend picked up fast and everybody was putting search boxes in the middle of their home pages. A web page wasn’t worth having if it didn't have a search box and any online business was only worth its salt if you could search its catalogue of products, services, or content via a search box from their landing page.
Search innovation from that point on was focussed more around SEO/SEM and monetisation more than the search experience itself. Lots of money was made, but the customer was left out in the cold. The basic experience of product search and discovery went no further than the search box.
At Zenitech we started thinking about how search could be based more around the human than the product? What if search was driven by lifestyle and went deeper than just matching a bunch of words against an index?
We decided that the Automotive industry was a good place to start this discovery. Traditionally, the Automotive search experience has been based around ‘Manufacturer’, then ‘Model’, then a bunch of other things such as body type, transmission, fuel and so on - which is fine if you know exactly what manufacturer and model you want. And if you do, then it’s not really a search is it? You are just navigating to a product which you already know exists, perhaps to see if one is available in the listings.
But what if you are like me, or my mum and you know nothing about cars? What is the best way to find a vehicle which might be suitable for you? We found that the best way to suggest potential products to the customer would be to get the customer to talk about themselves. Everybody can do that. It goes like this..
A: “So you want to buy a car?”
B: “Yep”
A: “Okay - I won’t ask you too much about what kind of car. Why not tell me about yourself instead?”
B: “Sounds good”
A: “Cool - so let’s explore your key motivations. Are you a family person, or a swish executive, or perhaps a sporty petrolhead looking for something to rev at the traffic lights? If not, then perhaps you are a bit more environmentally conscious, or..”
When we start to have a conversation with our customers, then a whole new raft of opportunities can open up. And this is where search meets product discovery and we can start to make real conversational connections with human beings.
I haven’t really mentioned AI yet. But then, why should I? AI is just an algorithmic way of building a statistical model and using that model to make predictions or suggestions. AI isn’t a product in its own right. It’s a way of powering a product and in our case, we built a clever AI which knows how to categorise cars so that any human can come along, answer some basic questions about themselves and be provided with vehicle recommendations. Here’s what our basic UI looks like:
It couldn’t be simpler. The user orders their lifestyle preferences by dragging them around with their thumb. Hit ‘suggest me some cars’ and like magic, the system serves up some suggestions. No searching. No need to know anything about cars and hopefully something new or interesting will come up.
This demonstrates how Artificial Intelligence can allow us to take esoteric things, which are difficult to measure in any real terms, and classify them to provide some way of matching and comparing them. In this case, we took a human’s lifestyle preferences, prioritised those and then matched them against our pre-classified database of around 35,000 cars. AI provides all the firepower of figuring out what the indicators are, within a vehicle description, of suitability to different lifestyle preferences. To do that procedurally in code would be impossible.
What other industry verticals could this technique be applied to? We are looking at Real Estate, Insurance, Travel and others. Maybe you have a suggestion. If so, get in touch!
If you want to try our demo app, you can download it from the Apple App Store, or Google Play Store. Just search for ‘AutoCompare’,