The other week, I was on Amazon looking for a present for my mother for Mother’s Day. I ended up ordering her a teapot, to make up for breaking her previous one. As soon as I completed the transaction, I decided to click back to the Amazon homepage to see what other deals I might be able to snag. As soon as I got there, however, I was greeted by a whole row of teapots — some of which looked identical to the one I had just bought.
Now, if you were just looking at it from an algorithmic perspective, it might make sense to offer up a variety of teapot styles to me, given my previous purchasing history. But once you take the machines out of it, it frankly seems a bit ridiculous to offer someone something you know they already have. And that, in a nutshell, is the problem that data-driven companies such as Amazon face. It’s all well and good to be able to predict when someone’s going to run out of toilet paper or milk and to send them a reminder notice, but how do you use someone’s browsing history to predict what kind of esoteric object they’re going to buy next?
Let’s say I’m going on vacation to Italy, and decide to buy a guidebook. If you’re buying the guidebook from Barnes & Noble, it might seem like there’s only so much they can do with this information; after all, Barnes & Noble only sells certain types of merchandise. But let’s say that Barnes & Noble’s systems decide to start serving you targeted ads on the basis of that guidebook purchase. Now, the easy way to do it would be to start sending you ads for a selection of Italian guidebooks — but, let’s face it, would you really be interested in buying another?
But if Barnes & Noble were to take a more sophisticated approach — that is, try and send you ads for products that aren’t guidebooks but might still be relevant to your trip — then they might actually be able to entice you into spending more money with them.
To better understand how AI can improve the customer journey, I sought out the advice of Marc Parrish, who previously served as Vice President of Retention and Loyalty Marketing for Barnes & Noble and is now the Chief Marketing Officer of ActionIQ, a leading enterprise Customer Data Platform. He explains that the future of marketing on both a large and small scale, from a B2B and B2C perspective, will involve greater and greater personalization. “Despite its reputation, Amazon’s personalization strategy is not very sophisticated,” explains Parrish. “Amazon looks at things that you put in your cart and abandoned, or suggests items you bought previously; in other words, it does a market basket analysis.” Market basket analysis operates on the theory that people who buy certain types of products are also likely to buy other, related products. While this strategy has proven effective for Amazon, says Parrish there’s a lot more they could be doing with the data — namely, combining personalization with the power of AI to create real-time, smart personalization that will improve the overall customer journey.
Generally speaking, data is only useful when it’s fresh, Parrish explains; it doesn’t always help companies to know that their customers were looking for a specific product three months ago. He advises companies to start implementing AI systems or algorithms that will allow them to act quickly, finding patterns and modeling behaviour based on the data. While this means giving the AI system some room to learn on its own, the payoff for brands is greater B2C engagement — as well as full use of a system that can act quickly based on the patterns it detects. Companies need to be proactive when it comes to AI implementation, and with behemoths like Google, IBM, and Amazon offering myriads of AI products, they’re spoiled for choice.
Ultimately, any brand that doesn’t currently have an online presence is at a severe disadvantage. For example, H&M has an advantage compared to a company like Primark because Primark doesn’t have an online experience. Additionally, consumers are smart: they can sense which companies are sophisticated and high tech, and which ones are not. They have expectations for the brands they shop at, especially now that most people have been exposed to the algorithmically advanced methods of Netflix and Amazon. Brands have to be willing to improve the customer journey, and be able to shift their mindset from a traditional retail model to a new one that privileges data and the insights that come with it.
“At the end of the day, the technology is only as good as the data you can get your hands on,” says Parrish. “Having sophisticated audience management tools is at the core of effective AI strategy.” He’s right. Frankly, the very act of collecting the data can put brands into the right mindset. Once brands start collecting data, the imagination will start working: they’ll be able to see what they can do with it, and see the concrete benefits of technological sophistication.
This post is part of our contributor series. The views expressed are the author's own and not necessarily shared by TNW.