Can Retail AI Gain Situational Awareness?


“Hello, Dave. You look really nice in those jeans. Maybe you’d like to buy this shirt too?”

The retail market is being taken over by Artificial Intelligence (AI). At least, that’s what it feels like as AI emerges as the new pathway to give customers a personalized shopping experience.

As tech and retail leaders like Google, eBay and Amazon invest heavily in AI and machine learning technologies, the industry is grappling with the question of whether AI can lead to recommendations that are contextual and situationally aware to each individual customer.

The simple answer is no – but they can sure act like it. Machines have gotten really good at emulating awareness. It can sound a bit confusing, but let’s take a product purchase example and see how it relates to AI predictive behavior for shopping.

Whenever we purchase a product, there are multiple correlations between the person and the buying action. For example, the other day I bought a mango smoothie, there was a lot of context that went into the purchase:

  • Location: The store was close to my home and has a neighborhood feel;
  • Cause: I like the fact the owner pays his workers a living wage;
  • Weather: It was really hot out – nothing cools you off like a smoothie;
  • Transaction History: I buy mango smoothies a lot, at least once a week;
  • Likes: I just like mangos and smoothies – see my twitter posts!

I chose these various contexts because they’re incredibly frequent and telling when a consumer makes a purchase. So can a machine become aware of these situations, intuit my desires and make recommendations on my smoothie purchases accordingly?

That would take a level of sentience found only in Asimov and in other sci-fi novels. But, that doesn’t mean we can’t approximate it.

Each of these contexts creates data. My location, the weather, my transaction history, etc. are all part of my data set. AI and machine learning allow us to find the patterns hidden in this data, looking for repetition that will show what comes next.

Continuing with the smoothie example, these five data points are incredibly relevant in my purchase decision. This is where AI machine learning can be really powerful in retail.

Using AI machine learning, we can test each one of these data points, not just with the individual shopper, but in aggregate with other people who are like me – my demographic data, my location data, people who like smoothies, etc. It will try different combinations of weights among these data points until it adjusts and finds the right combination to drive a purchase.

The beauty of machine learning is that it can keep adjusting itself until I make the purchase. If hitting me with a deal when it’s hot doesn’t work, it can automatically reduce the ‘weather’ weighting. It can try an offer when I’m close to a store.

And given we live in the age of data, there are a multitude of other contexts and data sets and combinations that can be used to determine what drives my purchases.

So machines are just machines. They know what they’re programmed to know. There is no situational awareness when it comes to AI in retail. But, the good news for the marketers out there is that they can ACT as if they were aware. It’s artificial, but it’s also pretty intelligent.

By Cosmas Wong, co-founder and President, Grey Jean Technologies

Cosmas Wong is co-founder and President of Grey Jean Technologies, where he applies his experience of shepherding the most sensitive data to the marketing and retail space. An avid traveler and loyalty program enthusiast, he applies his personal passions to the consumer sector. An advisor for technology startups in the payments space, he is an expert at consumer purchase trends. Wong also founded Enso Financial, a leader in the analysis of trading strategy and financial data for the world’s largest hedge funds.