This week, NVIDIA had a long session on Recommender Systems, and I believe they are the future of retail, and I don’t just mean for online retail. They will also revolutionize brick-and-mortar retail as the two concepts, brick and mortar and online, merge into something that eventually blends the best of both approaches.
Recommender Systems are AI-driven platforms. They’re currently used almost exclusively for online shopping and serve up offers that increasingly target a single customer’s unique needs and work to increase sales volumes. They also have the potential to improve customer satisfaction and loyalty if these systems allow the product experience to be over by just driving the close.
Let’s talk about Recommender Systems this week and how they are revolutionizing retail and things like social media.
Creating the perfect AI salesperson
A Recommender System works like a good retail salesperson but at scale. The best salespeople do their homework on their clients, know what the clients enjoy, know what they hate, know their hot buttons and know their priorities, and use that information to craft an offer that the client can’t refuse.
It is all about packaging up, and sometimes making up, information that will help them initially gain consideration and eventually close the sale. In theory, a Recommender System is like the best salesperson on steroids because it learns from anything about the prospect that is available online, as well as from prior engagements, which may have been with different companies that sold that intelligence on the prospect to them.
Using Amazon as an example, the cookies that have fed the service, past purchasing behavior, and things you have been interested in online fuel a set of recommendations that shoppers get when they arrive at the site. Be it books or movies, or appliances and toys, Amazon seems to know what you may be interested in and provides pictures of those items to drive you to the sale. But those efforts are relatively rudimentary when compared to what NVIDIA was talking about at the event.
For instance, NVIDIA spoke about dynamically building choices of products that were themselves designed to force a particular decision. For instance, a retailer needs to move Dyson Vacuums which are expensive, so they surface up a choice of vacuums where the Dysen appears to the closest match to what the buyer might eventually purchase. The user thinks they’re making the choice because they were manipulated into choosing what the vendor wanted them to buy.
The danger with Recommender Systems is being overly good at tricking buyers into buying something they don’t want and bleeding customer trust and satisfaction as a result. So, care must be taken not to manipulate customers into buying something they were tricked into buying, and instead being laser-focused on making sure the customer got the deal they were convinced they got.
This requires a great deal of intelligence about the buyer and a strong understanding of how systems like this can be manipulated to do harm and making sure that they assure a positive customer outcome instead.
This means that the implementation must have conflicting goals to both move merchandise and assure customer satisfaction and loyalty, or the latter will be sacrificed to the former and the business become unsustainable.
The system itself must be in a constant development loop to ensure it remains current and thus can carry out its objectives of driving sales and improving customer satisfaction and loyalty successfully. And, if done right, it should create a level of customer engagement that is not seen out of personalized brick-and-mortar shops with great sales reps.
And realize, that there is no reason why someone entering one of these physical stores couldn’t be identified and then guided to a unique set of items that best match that customer’s needs and drive additional sales.
Wrapping up: The future of retail
We don’t have enough great salespeople. Heck, we don’t have enough salespeople, period. And online, the quality, availability, and loyalty of salespeople has been suboptimal for some time. What a well-deployed Recommender System can do is digitize the sales experience, more consistently close sales opportunities, and provide a lower-cost, more effective alternative to salespeople who are currently in very short supply.
Deployed properly, Recommender Systems significantly increase sales and profits while improving customer satisfaction and assuring customer loyalty. This is how the best retail sites are likely to differentiate both for online and brick-and-mortar stores as the AI future becomes the present.