We all know how online shopping works. You spend hours browsing through different sites, different products, customer reviews, etc. before adding an item to your shopping cart. You then spend some more time contemplating the use or necessity of the purchase you’re about to make. The second part of contemplation can take a couple of seconds (if any) or even a couple of days. We tend to have a tipping point where we either purchase these items or we just remove them from our carts or wish list(s).
Amazon considered this a few years ago. This consideration on their part led to marketing the products you most look at as advertisements on third party applications. This is why you often see headphone advertisements pop up on your Instagram feed after you browse for headphones on ecommerce sites. Granted it gets annoying or scary at times but it does work. Nearly 1 out of every 20 recommended items are bought by users. This includes items you look for and items that are predicted using your user data.
Now imagine this. If all your search data, your social media feed and your application data was sent to a central predictive AI system which would decide what products you need the most, we could boost the accuracy of prediction. This idea laid the basis for anticipatory shipping. If you get the accuracy high enough, you don't need people to shop first before you ship the orders. You could send in a weekly or monthly “box” containing items that users would buy, at least as per predictions. Users can then select which of the items they need and want to keep, pay for them and return the rest.
There's a couple of benefits that a system like this gives to ecommerce platforms; amazon, in particular. It essentially wipes out competition simply because nobody is going to spend time on the net searching through different sites for what they want. Secondly, it forces users to actually see products with their own eyes. This causes a change in how they view the necessity or use of it.
If you look at a product in a non traditional setting, that is, in your home instead of a shop or a website, you tend to see how well it fits in with everything you have. You can better compare items and understand product specifics better given that you interact with it in a place of comfort. What's more, given the way the scheme would work, you now “return products” instead of “not buying” them. In other words, since it's harder for people to detach objects with which they interact from a place of comfort, you tend to increase the amount of products or variety of products you buy.
The reason it hasn’t been implemented yet is simply that the current accuracy is very low. However, a case can be made for implementing it even with a low accuracy. Essentially, if you get the system up and running, you get to collect more usable data and soon make up for short term losses if you get a high enough accuracy percentage in the future.
The essential takeaway is that adjusting the prediction machine's dial has a major impact on strategy. It transforms Amazon's business model from shopping-then-shipping to shipping-then-shopping, creates an incentive to vertically integrate into operating a product-returns service (including a fleet of trucks), and accelerates investment timing owing to first-mover advantage from higher returns.
We have had several tech commenters predict 2020 or 2021 as the year in which amazon finally adopts this model, given that the first articles on it came out as early as 2014. It certainly does have ethical ramifications that should and will be explored in the future, but it would transform how we look at online shopping or shopping in general.
Comments