Killing the keyboard​: AI’s going to do your shopping for you

Original article was published on Artificial Intelligence on Medium

We shop online in the same way as we did 20 years ago. Barely anything has changed. More people are doing it (of course), and sometimes they’re doing it in different places, but the mechanics of online product discovery are the same today as they were in 2000.

It goes like this: A customer sees a product they like, in a shop, on a billboard, display ad or whatever, and they start typing. If at the end of typing a description somewhere, they find the product, or one like it — and if checkout is easy and delivery cheap/quick — they buy it. Sometimes they hang around and buy some more things.

…or like this: A customer sees a product they like, and they start typing. If, after typing a description, they don’t find that product, or if it’s expensive, takes 6 weeks to be delivered, or other people say the product is crap, they look elsewhere. Sometimes they stop using that platform for look for products.

The entire digital commerce industry has been built off the back of making sure the first journey happens, avoiding the second at all costs. Why? Because if you can’t show the products people want to buy, they’ll stop buying from you.

It’s all about pushing the right products for the right text searches — product to intent matching. Commerce platforms have to be able to guess which of a bazillion in-stock products a customer wants to see, with only a couple of key taps to go on.

This product to intent matching requires two things:

  1. Having a good selection of products (from different brands) that you can show to customers.
  2. Having the technology to understand what someone is looking for and therefore what products you want to show.

Two decades since Amazon first made a profit, and some of the big digital players have become very good at product to intent matching. I can’t remember the last time I bothered looking at the second page of a shopping results page, anyway. Some others are going to get good, like Facebook.

The problem is that after 20 years of tinkering, they’re too good.

The algorithms are having to choose from ever greater product inventories, and show the right SKUs with ever fewer key taps to go on… mobile adoption and fat fingers have led to shorter search queries. At the same time, customers expect the ‘right’ results 100% of the time. Commerce platforms are having to guess a lot more, having hit the limit of what they can do with the hardware people use. The result is that we’re all seeing — and buying — the same thing. That’s bad business for so many reasons.

The keyboard birthed ecommerce, but ‘QWERTY’ is stifling product discovery.

That said, the hardware is changing and soon the keyboard may no longer be king. After two decades as the primary entry point to online shopping, the text search is going to be challenged by three new AI-powered generations of product discovery, in short order: